It is now part of the Calypso repository.
The published npm package will continue to be available as before, no changes necessary!
This lib enables translations, exposing three public methods:
It also provides a Higher-Order Component named localize(). Wrapping your component in localize() will give it the aforementioned functions as props. This is the suggested way of using them with React components.
Finally, this lib exposes a utility method for your React application:
translate() accepts up to three arguments (string, string, object), depending on the translation needs. The second and/or third parameter can be omitted:
/**
* @param {string} original - the string to translate, will be used as single version if plural passed
* @param {string} [plural] - the plural string to translate (if applicable)
* @param {object} [options] - properties describing translation requirements for given text
**/The following attributes can be set in the options object to alter the translation type. The attributes can be combined as needed for a particular case.
If you pass a single string into translate, it will trigger a simple translation without any context, pluralization, sprintf arguments, or comments. You would call it like this.
var i18n = require( 'i18n-calypso' );
var translation = i18n.translate( 'Some content to translate' );Translation strings are extracted from our codebase through a process of static analysis and imported into GlotPress where they are translated (more on that process here). So you must avoid passing a variable, ternary expression, function call, or other form of logic in place of a string value to the translate method. The one exception is that you can split a long string into mulitple substrings concatenated with the + operator.
/*----------------- Bad Examples -----------------*/
// don't pass a logical expression argument
var translation = i18n.translate( condition ? 'foo' : 'bar' );
// don't pass a variable argument
var translation = i18n.translate( foo );
// don't pass a function call argument
var translation = i18n.translate( foo( 'bar' ) );
/*----------------- Good Examples -----------------*/
// do pass a string argument
var example = i18n.translate( 'foo' );
// do concatenate long strings with the + operator
var translation = i18n.translate(
'I am the very model of a modern Major-General, ' +
'I\'ve information vegetable, animal, and mineral, ' +
'I know the kings of England, and I quote the fights historical ' +
'from Marathon to Waterloo, in order categorical.'
);The translate() method uses sprintf interpolation for string substitution (see docs for syntax details). The option.args value is used to inject variable content into the string.
// named arguments (preferred approach)
i18n.translate( 'My %(thing)s has %(number)d corners', {
args: {
thing: 'hat',
number: 3
}
} );
// 'My hat has 3 corners'
// argument array
i18n.translate( 'My %s has %d corners', {
args: [ 'hat', 3 ]
} );
// 'My hat has 3 corners'
// single substitution
i18n.translate( 'My %s has 3 corners', {
args: 'hat'
} );
// 'My hat has 3 corners'Because React tracks DOM nodes in the virtual DOM for rendering purposes, you cannot use string substitution with html markup as you might in a php scenario, because we don't render arbitrary html into the page, we are creating a virtual DOM in React.
Instead we use the interpolate-components module to inject components into the string using a component token as a placeholder in the string and a components object, similar to how string substitution works. The result of the translate() method can then be inserted as a child into another React component. Component tokens are strings (containing letters, numbers, or underscores only) wrapped inside double-curly braces and have an opening, closing, and self-closing syntax, similar to html.
NOTE: Always use a JSX element for passing components. Otherwise you will need to wrap your React classes with createFactory. Any wrapped content inside opening/closing component tokens will be inserted/replaced as the children of that component in the output. Component tokens must be unique:
// self-closing component syntax
var example = i18n.translate( 'My hat has {{hatInput/}} corners', {
components: {
hatInput: <input name="hatInput" type="text" />
}
} );
// component that wraps part of the string
var example2 = i18n.translate( 'I feel {{em}}very{{/em}} strongly about this.', {
components: {
em: <em />
}
} );
// components can nest
var example3 = i18n.translate( '{{a}}{{icon/}}click {{em}}here{{/em}}{{/a}} to see examples.', {
components: {
a: <a href="#" />,
em: <em />,
icon: <Icon size="huge" />
}
} );
You must specify both the singular and plural variants of a string when it contains plurals. If the string uses placeholders that will be replaced with actual values, then both the plural and singular strings should include those placeholders. It might seem redundant, but it is necessary for languages where a singular version may be used for counts other than 1.
// An example where the translated string does not have
// a number represented directly, but still depends on it
var numHats = howManyHats(), // returns integer
content = i18n.translate(
'My hat has three corners.',
'My hats have three corners.',
{
count: numHats
}
);
// An example where the translated string includes the actual number it depends on
var numDays = daysUntilExpiration(), // returns integer
content = i18n.translate(
'Your subscription will expire in %(numberOfDays)d day.',
'Your subscription will expire in %(numberOfDays)d days.',
{
count: numDays,
args: {
numberOfDays: numDays
}
}
);
// simplest case... just a translation, no special options
var content = i18n.translate( 'My hat has three corners.' );
// sprintf-style string substitution
var city = getCity(), // returns string
zip = getZip(), // returns string
content = i18n.translate( 'Your city is %(city)s, your zip is %(zip)s.', {
args: {
city: city,
zip: zip
}
} );
// Mixing strings and markup
// NOTE: This will return a React component, not a string
var component = i18n.translate( 'I bought my hat in {{country/}}.', {
components: {
country: <input name="someName" type="text" />
}
} );
// Mixing strings with markup that has nested content
var component = i18n.translate( 'My hat has {{link}}three{{/link}} corners', {
components: {
link: <a href="#three" />
}
} );
// add a comment to the translator
var content = i18n.translate( 'g:i:s a', {
comment: 'draft saved date format, see http://php.net/date'
} );
// providing context
var content = i18n.translate( 'post', {
context: 'verb'
} );
See the test cases for more example usage.
This module includes an instantiation of moment.js to allow for internationalization of dates and times. We generate a momentjs locale file as part of loading a locale and automatically update the moment instance to use the correct locale and translations. You can use moment() from within any component like this:
var thisMagicMoment = i18n.moment( "2014-07-18T14:59:09-07:00" ).format( 'LLLL' );And you can use it from outside of React like this.
var i18n = require( 'i18n-calypso' );
var thisMagicMoment = i18n.moment( "2014-07-18T14:59:09-07:00" ).format( 'LLLL' );The numberFormat method is also available to format numbers using the loaded locale settings (i.e., locale-specific thousands and decimal separators). You pass in the number (integer or float) and (optionally) the number of decimal places you want (or an options object), and a string is returned with the proper formatting for the currently-loaded locale. You can also override the locale settings for a particular number if necessary by expanding the second argument into an object with the attributes you want to override.
// These examples assume a 'de' (German) locale to demonstrate
// locale-formatted numbers
i18n.numberFormat( 2500.25 ); // '2.500'
i18n.numberFormat( 2500.1, 2 ); // '2.500,10'
i18n.numberFormat( 2500.33, { decimals: 3, thousandsSep: '*', decPoint: '@'} ); // '2*500@330'Using the method hasTranslation you can check whether a translation for a given string exists. As the translate() function will always return screen text that can be displayed (will supply the source text if no translation exists), it is unsuitable to determine whether text is translated. Other factors are optional key hashing as well as purposeful translation to the source text.
var i18n = require( 'i18n-calypso' );
i18n.hasTranslation( 'This has been translated' ); // true
i18n.hasTranslation( 'Not translation exists' ); // falseThe mixin has been removed from this distribution. Please use version 1 of i18n-calypso if you need to use the mixin.
localize is a higher-order component which, when invoked as a function with a component,
returns a new component class. The new component wraps the original component, passing all
original props plus props to assist in localization (translate, moment, and numberFormat).
The advantage of using a higher-order component instead of calling translate directly from
the i18n-calypso module is that the latter does not properly account for change events
which may be emitted by the state emitter object.
Typically, you'd wrap your exported function with localize:
// greeting.jsx
import React from 'react';
import { localize } from 'i18n-calypso';
function Greeting( { translate, className } ) {
return (
<h1 className={ className }>
{ translate( 'Hello!' ) }
</h1>
);
}
export default localize( Greeting );When the wrapped component is rendered, the render behavior of the original component is used, but with access to localization props.
// index.jsx
import React from 'react';
import { render } from 'react-dom';
import Greeting from './greeting';
render(
<Greeting className="greeting" />,
document.body
);I18n accepts a language-specific locale json file that contains the whitelisted translation strings for your JS project, uses that data to instantiate a Jed instance, and exposes a single translate method with sugared syntax for interacting with Jed.
In order to reduce file-size, i18n-calypso allows the usage of hashed keys for lookup. This is a non-standard extension of the Jed standard which is enabled by supplying a header key key-hash to specify a hash method (currently only sha1 is supported), as well as a hash length. For example sha1-4 uses the first 4 hexadecimal chars of the sha1 hash of the standard Jed lookup string. As a further optimization, variable hash lengths are available, potentially requiring multiple lookups per string: sha1-3-7 specifies that hash lengths of 3 to 7 are used in the file.
Instead of providing the full English text, like here:
{"":{"localeSlug":"de"},"Please enter a valid email address.":["","Bitte gib eine gültige E-Mail-Adresse ein."]}just the hash is used for lookup, resulting in a shorter file.
{"":{"localeSlug":"de","key-hash":"sha1-1"},"d":["","Bitte gib eine gültige E-Mail-Adresse ein."]}The generator of the jed file would usually try to choose the smallest hash length at which no hash collisions occur. In the above example a hash length of 1 (d short for d2306dd8970ff616631a3501791297f31475e416) is enough because there is only one string.
Note that when generating the jed file, all possible strings need to be taken into consideration for the collision calculation, as otherwise an untranslated source string would be provided with the wrong translation.