This is a plugin for gulp that allows you to replace your custom placeholders for different language versions of your translations. Call it i18n, localization, translation or whatever you want, in the end it's basically a token replacer with multiple source files to create multiple versions of your original.
There are a few other plugins out there that I've tried but couldn't quite get to do what I wanted. Maybe they work better for you: gulp-i18n gulp-localize gulp-static-i18n gulp-i18n-localize gulp-l10n Rather than dig in their code I wanted to use this opportunity to learn how to write plugins for gulp. So if none of those do the trick you can try this little plugin.
Features (cause we all love features):
- Custom token formats
- Custom language descriptors
- Custom filename format
- Custom translation source directory
- Read from .json, .js, .ini or .csv file (for examples check the test folder),
- Filters for individual files
$ npm install --save-dev gulp-international
var gulp = require('gulp');
var international = require('gulp-international');
gulp.task('default', function () {
return gulp.src('src/file.ext')
.pipe(international())
.pipe(gulp.dest('dist'));
});Or with your custom options if the out of box config doesn't work:
var gulp = require('gulp');
var international = require('gulp-international');
gulp.task('default', function () {
return gulp.src('src/file.ext')
.pipe(international({
whitelist: 'en_US'
}))
.pipe(gulp.dest('dist'));
});This module tries to be very flexible with it's configuration option. For many options you can define either a boolean, a string, a regular expression or an array with a mixture of string and regular expressions. If a boolean is passed in the option is simply set to true or false for all files. If the option is a string the filename is searched for this substring. If the option is regular expression it is tested against each file, and finally if the option is an array the test is executed against every element of the array and accepted if any of the elements returns true with the previous rules defined. You could even nest arrays if you wanted to.
An example:
options.ignoreTokens = [ 'de_DE', /en_*/ ]
// => matches any english language as well the German translation for Germany
Type: string (path)
Default: ./locales
This tells the script where to look for translation files. Currently supported are .json, .js, .ini and .csv files. Each format supports
nested keys (.ini only 1 level via sections). The name of the file is used for generating the output file. A source file with name
index.html with a translation file called en_US.json would result in an output file called index-en_US.html.
You don't have to use this format. Basically any name is valid for the file and will be used later on as the language placeholder.
Type: string
Default: ${path}/${name}-${lang}.${ext}
This option allows the user to configure the output format as desired. The default will generate all files in the same directory with the language suffix.
There are 4 placeholders available:
lang: The filename of the translation file without extension (e.g. 'en_US')path: The relative filepath inside the target directory (e.g. 'static')name: The original filename (e.g. 'index')ext: The original file extension (e.g. 'html')
Here are a few examples of what the results would be:
${path}/${name}-${lang}.${ext} = static/index-en_US.html
${lang}/${path}/${name}.${ext} = en_US/static/index.html
${path}/${name}.${ext}.${lang} = static/index.html.en_US
${lang}/${name}.${ext} = en_US/index.html // ignores the path and copies everything in the same dir
Type: Object
Default:
{
prefix: 'R.',
stopCondition: /[^\.\w_\-]/
}
The default configuration takes some queues from Android resources. The replacer is using the prefix to find tokens. The stopCondition is used to determine where a token ends (exclusively). The default settings look for either whitespace or a selection of special characters. Examples of matchable resources are:
- html:
<div>R.title<div> - javascript:
alert('R.message');,var test = a[R.item] - jade:
div.big#R.field(src="R.link") R.linkDescription
Alternatively you can also define a suffix that will determine where a token ends while including the result. An example configuration would be
{
prefix: '${',
suffix: '}'
}
This configuration would match any tokens formatted like this: <div>${title}</div>.
For help a flow graph of the files and content generated during the process:
source.file -> source-lang1.file -> translated content
-> source-lang2.file -> translated content
Type: boolean|string|string[]|RegExp|RegExp[]
Default: true
This option allows to limit the number of translations that can be used. The names that are whitelisted need to match the filename (without the extension). Any other files will still be loaded into the list of available dictionaries, but no files will be generated. The option will allow any locale if it's set to true. You can either define a single locale with a string/RegExp or multiple with an array of strings and RegExp's.
Type: boolean|string|string[]|RegExp|RegExp[]
Default: false
The opposite of the whitelist. Any language specified here will be ignored during processing. You can either define a single locale with a string or multiple with an array of strings.
Type: boolean|string|string[]|RegExp|RegExp[]
Default: true
Any non utf8 characters that do not map to html are replaced with html entities if this option is enabled. A translation such as "über" would be replaced with "über". If the setting is a regular expression then only files with a matching (original) filename will be escaped. If the option is instead a string then the filename is searched for this substring.
Type: boolean|string|string[]|RegExp|RegExp[]
Default: true
This enables warnings to be printed out if any tokens are missing. If the setting is a regular expression then only files with a matching (original) filename will throw warnings. If the option is instead a string then the filename is searched for this substring.
Type: boolean
Default: true
This enables caching of dictionaries, so that reruns of the plugin will not have to read in the translation tokens again. Depending on your configuration you might want to disable caching based on how your livereloads are configured.
Type: boolean|string|string[]|RegExp|RegExp[]
Default: false
Allows to disable throwing of any errors that might occur so that the pipe will continue executing. If the setting is a regular expression then only files with a matching (original) filename will ignore errors. If the option is instead a string then the filename is searched for this substring.
Type: boolean|string|string[]|RegExp|RegExp[]
Default: false
When set to true the plugin will perform all operations for a translation, but will pass on the original file along the pipe instead of the newly generated ones. If the setting is a regular expression then only files with a matching (original) filename will be ignored. If the option is instead a string then the filename is searched for this substring.
Flow graph:
source.file -> source.file -> original content
Type: boolean|string|string[]|RegExp|RegExp[]
Default: false
When set to true the plugin will ignore all tokens, but still create new files as if they were different for each language. This differs from a dryRun, which would instead pass on the original file. If the setting is a regular expression then only files with a matching (original) filename will be ignored. If the option is instead a string then the filename is searched for this substring.
Flow graph:
source.file -> source-lang1.file -> original content
-> source-lang2.file -> original content
Type: boolean|string|string[]|RegExp|RegExp[]
Default: false
When set to true the original file is passed along the pipe along with all translated files. If the setting is a regular expression then only files with a matching (original) filename will be included. If the option is instead a string then the filename is searched for this substring.
Flow graph:
source.file -> source.file -> original content
-> source-lang1.file -> translated content
-> source-lang2.file -> translated content
Type: boolean
Default: false
This option will log a lot more information to the console. Enable this if you can't quite get the plugin to do what you want it to and need some more details.
A number of formats are supported from where we can read translations:
{
"token1": "translation1",
"section1": {
"token2": "translation2",
"subsection1": {
"token3": "translation3"
}
}
}
This results in 3 tokens that look like this (With default delimiter settings):
R.token1 = translation1
R.section1.token2 = translation2
R.section1.subsection1.token3 = translation3
Similar to JSON you can just use node.js module and export your json:
module.exports = {
token1: 'translation1',
section1: {
token2: 'translation2', // This format also allows comments
subsection1: {
token3: 'translation3'
}
}
}
The tokens map the same way as a JSON file would. If you don't know which format to choose I would suggest this one.
In this format all except for the last readable field will make up the token. The last field will be the translation. To get the same output as with the previous examples the file would look something like this:
token1,,,translation1
token2,section1,,translation2
token3,section1,subsection1,translation3
This library doesn't really check for a correct CSV format and also doesn't assume a header line. It just scans for fields that it can use to build it's keys. Another valid format to read translations would be this non standard csv file:
token1,translation1
token2,section1,translation2
token3,section1,subsection1,translation3
If you have fields with , in them you can escape them ", which in turn can also be escaped using \".
The standard .ini format is supported as well, but only supports one level of nesting. You can simulate multiple levels though:
token1=translation1
[section1]
token2=translation2
[section1.subsection1]
token3=translation3
Sometimes you have dynamic scripts that you generate where you want to react differently based on the language that is being used. For
example let's say that you have a javascript file and need the language to be available from a variable. Simply set up the plugin to run
before the script is being executed on the script itself and include a line such as var lang = 'R.lang'. All you have to do to
make this working is have a key in your definition with the name "lang".
Again if you replace the string that is already there it is easy to integrate with existing formatting functions. Run the plugin before the script is being executed and you can make use of formatters such as sprintf, util.format, console.log and others. Since it's common to use gulp for browser projects here's a simple formatting function you can make use of (taken from StackOverflow):
/**
* Add formating function to string objects.
* @returns {string}
*/
String.prototype.format = function () {
var args = arguments;
return this.replace(/\{\{|}}|\{(\d+)}/g, function (m, n) {
if (m == "{{") { return "{"; }
if (m == "}}") { return "}"; }
return args[n];
});
};
With a function like this you can create smarter translations such as:
token1={0} is {1}, isn't it?
"R.token1".format('JS', 'awesome');
=> "JS is awesome, isn't it?"
Maybe I'll implement these one day, maybe not.
- Replace links to standard versions with internationalized versions (can probably also just be done with using tokens for imports)
- Extract non code strings from source and list them as still missing translations (difficult to doif this should work for any type of source file
- Warn about unused translation strings (probably easy to do, but not sure if it's worth the effort)
- Make translations available as environment variables in jade/js/coffeescript/etc. (although you can already replace strings anywhere)
- Support streams... although that seems like a pain to implement
- Support printing of token trees if they are nested (as a json object that can e.g. be parsed by another script... although at this point you might just as well include the original source file and not have it run through here.)
- Look for language files recursively
- Merge multiple language files (Using some sort of merge pattern? or by just merging files that have the same filename without ext & path)