Start with a widget with some text in it:
Text("Hello, how are you?")
Translate it simply by adding .i18n
to the string:
Text("Hello, how are you?".i18n)
If the current locale is 'pt_BR'
, then the text in the screen will be
"Olá, como vai você?"
, the Portuguese translation to the above text. And so on for any other
locales you want to support.
You can also provide different translations depending on modifiers, for example plural
quantities:
print("There is 1 item".plural(0)); // Prints 'There are no items'
print("There is 1 item".plural(1)); // Prints 'There is 1 item'
print("There is 1 item".plural(2)); // Prints 'There are 2 items'
And you can invent your own modifiers according to any conditions. For example, some languages have
different translations for different genders. So you could create gender
versions for Gender
modifiers:
print("There is a person".gender(Gender.male)); // Prints 'There is a man'
print("There is a person".gender(Gender.female)); // Prints 'There is a woman'
print("There is a person".gender(Gender.they)); // Prints 'There is a person'
Also, interpolating strings is easy, with the fill
method:
// Prints 'Hello John, this is Mary' in English.
// Prints 'Olá John, aqui é Mary' in Portuguese.
print("Hello %s, this is %s".i18n.fill(["John", "Mary"]));
Try running the example.
I'm always interested in creating packages to reduce boilerplate. For
example, async_redux is about Redux without boilerplate,
and align_positioned is about creating layouts using
fewer widgets. This current package is also about reducing boilerplate for translations, so it
doesn't do anything you can't already do with plain old Localizations.of(context)
.
That said, this package is meant both for the one person app developer, and the big company team. It has you covered in all stages of your translation efforts:
-
When you create your widgets, it makes it easy for you to define which strings should be translated, by simply adding
.i18n
to them. These strings are called "translatable strings". -
When you want to start your translation efforts, it can automatically list for you all strings that need translation. If you miss any strings, or if you later add more strings or modify some of them, it will let you know what changed and how to fix it.
-
You can then provide your translations manually, in a very easy-to-use format.
-
Or you can easily integrate it with professional translation services, importing it from, or exporting it to any format you want.
Wrap your widget tree with the I18n
widget, below the MaterialApp
, together with
the localizationsDelegates
and the supportedLocales
:
import 'package:i18n_extension/i18n_widget.dart';
import 'package:flutter_localizations/flutter_localizations.dart';
...
@override
Widget build(BuildContext context) {
return MaterialApp(
localizationsDelegates: [
GlobalMaterialLocalizations.delegate,
GlobalWidgetsLocalizations.delegate,
GlobalCupertinoLocalizations.delegate,
],
supportedLocales: [
const Locale('en', "US"),
const Locale('pt', "BR"),
],
home: I18n(child: ...)
);
}
Note: To be able to import flutter_localizations.dart
you must add this to
your pubspec.yaml
:
dependencies:
flutter_localizations:
sdk: flutter
i18n_extension: ^1.4.2
The code home: I18n(child: ...)
shown above will translate your strings to the current system
locale. Or you can override it with your own locale, like this:
I18n(
initialLocale: Locale("pt", "BR"),
child: ...
Note: Don't ever put translatable strings in the same widget where you declared the I18n
widget, since they may not respond to future locale changes. For example, this is a mistake:
Widget build(BuildContext context) {
return I18n(
child: Scaffold(
appBar: AppBar(title: Text("Hello there".i18n)),
body: MyScreen(),
);
}
You may put translatable strings in any widgets down the tree.
The correct way to create a Locale
is to provide a language code (usually 2 or 3 lowercase
letters)
and a country code (usually 2 uppercase letters), as two separate Strings.
For example:
var locale = Locale("en", "US");
print(locale); // Prints `en_US`.
print(locale.languageCode); // Prints `en`.
You can, if you want, omit the country code:
var locale = Locale("en");
print(locale); // Prints `en`.
print(locale.languageCode); // Prints `en`.
But you cannot provide language code and country code as a single String. This is wrong:
// This will create a language called "en_US" and no country code.
var locale = Locale("en_US");
print(locale); // Prints `en_US`.
print(locale.languageCode); // Also prints `en_US`.
To help avoiding this mistake, the i18n_extension
may throw an error if your language code
contains underscores.
When you create a widget that has translatable strings, add this default import to the widget's file:
import 'package:i18n_extension/default.i18n.dart';
This will allow you to add .i18n
and .plural()
to your strings, but won't translate anything.
When you are ready to create your translations, you must create a dart file to hold them. This file
can have any name, but I suggest you give it the same name as your widget and change the termination
to .i18n.dart
.
For example, if your widget is in file my_widget.dart
, the translations could be in
file my_widget.i18n.dart
You must then remove the previous default import, and instead import your own translation file:
import 'my_widget.i18n.dart';
Your translation file itself will be something like this:
import 'package:i18n_extension/i18n_extension.dart';
extension Localization on String {
static var _t = Translations("en_us") +
{
"en_us": "Hello, how are you?",
"pt_br": "Olá, como vai você?",
"es": "¿Hola! Cómo estás?",
"fr": "Salut, comment ca va?",
"de": "Hallo, wie geht es dir?",
};
String get i18n => localize(this, _t);
}
The above example shows a single translatable string, translated to American English, Brazilian Portuguese, and general Spanish, French and German.
You can, however, translate as many strings as you want, by simply adding more translation maps:
import 'package:i18n_extension/i18n_extension.dart';
extension Localization on String {
static var _t = Translations("en_us") +
{
"en_us": "Hello, how are you?",
"pt_br": "Olá, como vai você?",
} +
{
"en_us": "Hi",
"pt_br": "Olá",
} +
{
"en_us": "Goodbye",
"pt_br": "Adeus",
};
String get i18n => localize(this, _t);
}
The locale you pass in the Translations()
constructor is called the default locale. For
example, in Translations("en_us")
the default locale is en_us
. All translatable strings in the
widget file should be in the language of that locale.
The strings themselves are used as keys when searching for translations to the other locales.
For example, in the Text
below, "Hello, how are you?"
is both the translation to English, and
the key to use when searching for its other translations:
Text("Hello, how are you?".i18n)
If any translation key is missing from the translation maps, the key itself will be used, so the text will still appear in the screen, untranslated.
If the translation key is found, it will choose the language according to the following rules:
-
It will use the translation to the exact current locale, for example
en_us
. -
If this is absent, it will use the translation to the general language of the current locale, for example
en
. -
If this is absent, it will use the translation to any other locale with the same language, for example
en_uk
. -
If this is absent, it will use the key itself as the translation.
Other translation packages ask you to define key identifiers for each translation, and use those.
For example, the above text key could be helloHowAreYou
or simply greetings
. And then you could
access it like this: MyLocalizations.of(context).greetings
.
However, having to define identifiers is not only a boring task, but it also forces you to navigate to the translation if you need to remember the exact text of the widget.
With i18n_extension
you can simply type the text you want and that's it. If some string is already
translated, and you later change it in the widget file, this will break the link between the key and
the translation map. However, i18n_extension
is smart enough to let you know when that happens, so
it's easy to fix. You can even add this check to tests, as to make sure all translations are linked
and complete.
When you run your app or tests, each key not found will be recorded to the static set
Translations.missingKeys
. And if the key is found but there is no translation to the current
locale, it will be recorded to Translations.missingTranslations
.
You can manually inspect those sets to see if they're empty, or create tests to do that automatically, for example:
expect(Translations.missingKeys, isEmpty);
expect(Translations.missingTranslations, isEmpty);
Note: You can disable the recording of missing keys and translations by doing:
Translations.recordMissingKeys = false;
Translations.recordMissingTranslations = false;
Another thing you may do, if you want, is to set up callbacks that the i18n_extension
package will
call whenever it detects a missing translation. You can then program these callbacks to throw errors
if any translations are missing, or log the problem, or send emails to the person responsible for
the translations.
To do that, simply inject your callbacks into Translations.missingKeyCallback
and
Translations.missingTranslationCallback
.
For example:
Translations.missingTranslationCallback =
(key, locale) =>
throw TranslationsException("There are no translations in '$locale' for '$key'.");
As explained, by using the Translations()
constructor you define each key and then provide all its
translations at the same time. This is the easiest way when you are doing translations manually, for
example, when you speak English and Spanish and are creating yourself the translations to your app.
However, in other situations, such as when you are hiring professional translation services or
crowdsourcing translations, it may be easier if you can provide the translations by locale/language,
instead of by key. You can do that by using the Translations.byLocale()
constructor.
static var _t = Translations.byLocale("en_us") +
{
"en_us": {
"Hi.": "Hi.",
"Goodbye.": "Goodbye.",
},
"es_es": {
"Hi.": "Hola.",
"Goodbye.": "Adiós.",
}
};
You can also add maps using the +
operator:
static var _t = Translations.byLocale("en_us") +
{
"en_us": {
"Hi.": "Hi.",
"Goodbye.": "Goodbye.",
},
} +
{
"es_es": {
"Hi.": "Hola.",
"Goodbye.": "Adiós.",
}
};
Note above, since "en_us" is the default locale you don't need to provide translations for those.
To combine translations you can use the *
operator. For example:
var t1 = Translations("en_us") +
{
"en_us": "Hi.",
"pt_br": "Olá.",
};
var t2 = Translations("en_us") +
{
"en_us": "Goodbye.",
"pt_br": "Adeus.",
};
var translations = t1 * t2;
print(localize("Hi.", translations, locale: "pt_br");
print(localize("Goodbye.", translations, locale: "pt_br");
Sometimes you have different translations that depend on a number quantity. Instead of .i18n
you
can use .plural()
and pass it a number. For example:
int numOfItems = 3;
return Text("You clicked the button %d times".plural(numOfItems));
This will be translated, and if the translated string contains %d
it will be replaced by the
number.
Then, your translations file should contain something like this:
static var _t = Translations("en_us") +
{
"en_us": "You clicked the button %d times"
.zero("You haven't clicked the button")
.one("You clicked it once")
.two("You clicked a couple times")
.many("You clicked %d times")
.times(12, "You clicked a dozen times"),
"pt_br": "Você clicou o botão %d vezes"
.zero("Você não clicou no botão")
.one("Você clicou uma única vez")
.two("Você clicou um par de vezes")
.many("Você clicou %d vezes")
.times(12, "Você clicou uma dúzia de vezes"),
};
String plural(int value) => localizePlural(value, this, _t);
The plural modifiers you can use are zero
, one
, two
, three
, four
, five
, six
, ten
,
times
(for any number of elements, except 0, 1 and 2), many
(for any number of elements,
except 1, including 0), zeroOne
(for 0 or 1 elements), and oneOrMore
(for 1 and more elements).
Also, according to a Czech
speaker, there must be a special modifier for 2, 3 and 4. To that end you can use
the twoThreeFour
modifier.
Note: It will use the most specific plural modifier.
For example, .two
is more specific than .many
.
If no applicable modifier can be found, it will default to the unversioned string.
For example, this: "a".zero("b").four("c:")
will default to "a"
for 1, 2, 3, or more than 5 elements.
You can actually create any modifiers you want. For example, some languages have different
translations for different genders. So you could create .gender()
that accepts Gender
modifiers:
enum Gender {they, female, male}
int gnd = Gender.female;
return Text("There is a person".gender(gnd));
Then, your translations file should use .modifier()
and localizeVersion()
like this:
static var _t = Translations("en_us") +
{
"en_us": "There is a person"
.modifier(Gender.male, "There is a man")
.modifier(Gender.female, "There is a woman")
.modifier(Gender.they, "There is a person"),
"pt_br": "Há uma pessoa"
.modifier(Gender.male, "Há um homem")
.modifier(Gender.female, "Há uma mulher")
.modifier(Gender.they, "Há uma pessoa"),
};
String gender(Gender gnd) => localizeVersion(gnd, this, _t);
Your translations file may declare a fill
method to do interpolations:
static var _t = Translations("en_us") +
{
"en_us": "Hello %s, this is %s",
"pt_br": "Olá %s, aqui é %s",
};
String get i18n => localize(this, _t);
String fill(List<Object> params) => localizeFill(this, params);
Then you may use it like this:
print("Hello %s, this is %s".i18n.fill(["John", "Mary"]));
The above code will print Hello John, this is Mary
if the locale is English,
or Olá John, aqui é Mary
if it's Portuguese.
It uses the sprintf package internally. I don't know how closely it follows the C sprintf specification, but here it is.
If you have a translation object you can use the localize
function directly to perform
translations:
var translations = Translations("en_us") +
{
"en_us": "Hi",
"pt_br": "Olá",
};
// Prints "Hi".
print(localize("Hi", translations, locale: "en_us");
// Prints "Olá".
print(localize("Hi", translations, locale: "pt_br");
// Prints "Hi".
print(localize("Hi", translations, locale: "not valid");
To change the current locale, do this:
I18n.of(context).locale = Locale("pt", "BR");
To return the current locale to the system default, do this:
I18n.of(context).locale = null;
Note: The above will change the current locale only for the i18n_extension
, and not for Flutter
as a whole.
To read the current locale, do this:
// Both ways work:
Locale locale = I18n.of(context).locale;
Locale locale = I18n.locale;
// Or get the locale as a lowercase string. Example: "en_us".
String localeStr = I18n.localeStr;
// Or get the language of the locale, lowercase. Example: "en".
static language = I18n.language;
There is a global static callback you can use to observe locale changes:
I18n.observeLocale =
({Locale oldLocale, Locale newLocale}) => print("Changed from $oldLocale to $newLocale.");
This package is optimized so that you can easily create and manage all of your translations yourself, by hand.
However, for large projects with big teams you probably need to follow a more involved process: Export all your translatable strings to files in some external format your professional translator, or your crowdsourcing tool uses (see formats below). Continue developing your app while waiting for the translations. Import the translation files into the project and test the app in each language you added. Repeat the process as needed, translating just the changes between each app revision. As necessary, perform additional localization steps yourself.
The following formats may be used with translations:
-
JSON: Can be used, however it lacks specific features for translation, like plurals and gender.
-
ARB: This is based on JSON, and is the default format for Flutter localizations. https://github.com/google/app-resource-bundle/wiki/ApplicationResourceBundleSpecification
-
ICU: https://format-message.github.io/icu-message-format-for-translators/
-
XLIFF: This is based in XML. https://en.wikipedia.org/wiki/XLIFF
-
CSV: You can open this with Excel, save it in .XLSX and edit it there. However, beware not to export it back to CSV with the wrong settings (using something else than UTF-8 as encoding). https://en.wikipedia.org/wiki/Comma-separated_values
-
YAML: Can be used, however it lacks specific features for translation, like plurals and gender.
Currently, only .PO
and .JSON
importers are supported out-of-the-box.
Note: Those importers were contributed by Johann Bauer. If you want to help creating importers for any of the other formats above, please PR here: https://github.com/marcglasberg/i18n_extension.
Add your translation files as assets to your app in a directory structure like this:
app
\_ assets
\_ locales
\_ de.po
\_ fr.po
...
Then you can import them using GettextImporter
or JSONImporter
:
import 'package:i18n_extension/io/import.dart';
import 'package:i18n_extension/i18n_extension.dart';
void loadTranslations() async {
TranslationsByLocale translations =
Translations.byLocale("en_GB")
+ await GettextImporter().fromAssetDirectory("assets/locales");
}
Note: If you need to import any other custom format, remember importing is easy to do because
the Translation constructors use maps as input. If you can generate a map from your file format, you
can then use the Translation()
or Translation.byLocale()
constructors to create the translation
objects.
An utility script to automatically export all translatable strings from your project was contributed
by Johann Bauer. Simply
run flutter pub run i18n_extension:getstrings
in your project root directory and you will get a
list of strings to translate in strings.json
. This file can then be sent to your translators or be
imported in translation services like Crowdin, Transifex or Lokalise. You can use it as part of your
CI pipeline in order to always have your translation templates up to date.
Note the tool simply searches the source code for strings to which getters like .i18n
are applied.
Since it is not very smart, you should not make it too hard:
print("Hello World!".i18n); // This would work.
// But the tool would not be able to spot this since it doesn't execute the code.
var x = "Hello World";
print(x.i18n);
As previously discussed, i18n_extension will automatically list all keys into a map if you use some
unknown locale, run the app, and manually or automatically go through all the screens. For example,
create a Greek locale if your app doesn't have Greek translations,
and it will list all keys into Translations.missingTranslationCallback.
Then you can read from this map and create your exported file. There is
also this package
that goes through all screens automatically.
Q: Do I need to maintain the translation files as Dart files?
A: Not really. You do have a Dart file that creates a Translation
object, yes, and this
object is optimized for easily creating translations by hand. But it creates them from maps. So if
you can create maps from some file you can use that file. For example, a simple code generator that
reads .json
und outputs Dart maps would do the job:
var _t = Translations("en_us") + readFromJson("myfile.json")
.
Q: How do you handle changing the locale? Does the I18n class pick up changes to the locale automatically or would you have to restart the app?
A: It should pick changes to the locale automatically. Also, you can change the locale manually
at any time by doing I18n.of(context).locale = Locale("pt", "BR");
.
Q: What's the point of importing 'default.i18n.dart'?
A: This is the default file to import from your widgets. It lets the developer add .i18n
to
any strings they want to mark as being a "translatable string". Later, someone will have to remove
this default file and add another one with the translations. You basically just change the import
later. The point of importing 'default.i18n.dart' before you create the translations for that widget
is that it will record them as missing translations, so that you don't forget to add those
translations later.
Q: Can I do translations outside of widgets?
A: Yes, since you don't need access to context
. It actually reads the current locale
from I18n.locale
, which is static, and all the rest is done with pure Dart code. So you can
translate anything you want, from any code you want. You can also define a locale on the fly if you
want to do translations to a locale different from the current one.
Q: By using other translation methods if I type localizations.howAreYou
, I know that there's a
localization key named howAreYou
because otherwise my code wouldn't compile. There is no way to
statically verify that "How are you?".i18n
will do what I want it to do.
A: i18n_extension does away with identifier keys like howAreYou
. Not having those was one
thing I was trying to achieve. I hate having to come up with these keys. I found that the developer
should just type the text they want and be done with it. In other words, in i18n_extension you are
not typing a key, you are typing the text itself (in your default language). So there is no need to
statically verify anything. Your code will always compile when you type a String, and that exact
string will be used for your default language. It will never break.
Q: But how can I statically verify that a string has translations? Just showing the translatable string as defined in the source code will not hide that some translations are missing?
A: You can statically verify that a string should have translations because it has .i18n
attached to it. What you can't do is statically verify that those translations were actually
provided for all supported languages. But this is also the case when you use older methods. With the
older methods you also just know it should have translations, because it has a translation key, but
the translation itself may be missing, or worse yet, outdated. With i18n_extension at least you know
that the translation to the default language exists and is not outdated.
Q: What happens if a developer tries to call i18n on a string without translations, wouldn't that be harder to catch?
A: With i18n_extension you can generate a report with all missing translations, and you can even add those checks to tests. In other words, you can just freely modify any translatable string, and before your launch you get the reports and fix all the translations.
Q: There are a lot of valid usages for String that don't deal with user-facing messages. I like to
use auto-complete to see what methods are available (by typing someString.
), and seeing loads of
unrelated extension methods in there could be annoying.
A: The translation extension is contained in your widget file. You won't have this extension in
scope for your business classes, for example. So .i18n
will only appear in your auto-complete
inside of your widget classes, where it makes sense.
Q: Do I actually need one .i18n.dart
(a translations file) per widget?
A: No you don't. It's suggested that you create a translation file per widget if you are doing
translations by hand, but that's not a requirement. The reason I think separate files is a good idea
is that sometimes internationalization is not only translations. You may need to format dates in
specific ways, or make complex functions to create specific strings that depend on variables etc. So
in these cases you will probably need somewhere to put this code. In any case, to make translations
work all you need a Translation object which you can create in many ways, by adding maps to it using
the +
operator, or by adding other translation objects together using the *
operator. You can
create this Translation objects anywhere you want, in a single file per widget, in a single file for
many widgets, or in a single file for the whole app.
Also, if you are not doing translations by hand but importing strings from translation files, then
you don't even need a separate file. You can just add
extension Localization on String { String get i18n => localize(this, Translations("en_us") + load("file.json")); }
to your own widget file.
Q: Won't having multiple files with extension Localization
lead to people importing the wrong
file and have translations missing?
A: The package records all your missing translations, and you can also easily log or throw an
exception if they are missing. So you will know if you import the wrong file. You can also add this
reports to your unit tests. It will let you know even if you import the right file and translations
are missing in some language, and it will let you know even if you import from .arb
files and
translations are missing in some language.
Q: Are there importers for X?
A: Currently, only .PO
and .JSON
importers are supported out-of-the-box. Keep in mind this
lib development is still new, and I hope the community will help writing more importers/exporters.
We hope to have those for .arb
.icu
.xliff
.csv
and .yaml
, but we're not there yet.
However, since the Translations
object use maps as input/output, you can use whatever file you
want if you convert them to a map yourself.
Q: How does it report missing translations?
A: _At the moment you should just print Translations.missingKeys
and Translations.missingTranslations
. We'll later create a Translations.printReport()
function
that correlates these two pieces of information and outputs a more readable report.
Q: The package says it's "Non-boilerplate", but doesn't .i18n.dart
contain boilerplate?
A: The only necessary boilerplate for .i18n.dart
files
is static var _t = Translations("...") +
and String get i18n => localize(this, _t);
. The rest are the translations themselves. So, yeah,
it's not completely without boilerplate, but saying "Less-boilerplate" is not that catchy.
The Flutter packages I've authored:
- async_redux
- fast_immutable_collections
- provider_for_redux
- i18n_extension
- align_positioned
- network_to_file_image
- image_pixels
- matrix4_transform
- back_button_interceptor
- indexed_list_view
- animated_size_and_fade
- assorted_layout_widgets
- weak_map
My Medium Articles:
- Async Redux: Flutter’s non-boilerplate version of Redux ( versions: Português)
- i18n_extension ( versions: Português)
- Flutter: The Advanced Layout Rule Even Beginners Must Know ( versions: русский)
My article in the official Flutter documentation:
---
Marcelo Glasberg:
https://github.com/marcglasberg
https://twitter.com/glasbergmarcelo
https://stackoverflow.com/users/3411681/marcg
https://medium.com/@marcglasberg