Swift library that helps in localizing apps in a different, better, simpler, more powerful way than system localization does. It uses json files instead of strings files.

Swift 3

Swift 3 compatible version is available on swift3 branch. No pod available yet, though. Encountering issue with code sign and xcode 8 during validating podspec. Will try to solve it.


  • [x] Pluralization support - Without using stringdict files
  • [x] Length variations support - Supported since iOS 8.0 (instead of iOS 9.0 like system does) and avoids using stringsdict files
  • [x] Expressions - inequality and regular expressions
  • [x] Shared Expressions
  • [x] Built-in Expressions
  • [x] Works similarly to NSLocalizedString()
  • [x] Uses JSON files to minimize boilerplate code
  • [x] Comprehensive Unit Test Coverage
  • [x] Full documentation


Swifternalization helps in localizing apps in a smarter way. It has been created because of necessity to solve Polish language internalization problems but it is universal and works with every language very well.

It uses JSON files and expressions that avoid writing code to handle some cases that you have to write when not using this framework. It makes localizing process simpler.

Practical Usage Example

Description of practical usage example will use things that are covered later in the document so keep reading it to the end and then read about details/features presented here.


Let's assume the app supports English and Polish languages. Naturally app contains two Localizable.strings files. One is for Base localization which contains English translation and one is Polish language.

App displays label with information which says when object from the backend has been updated for the last time, e.g. "2 minutes ago", "3 hours ago", "1 minute ago", etc.


The label displays number and a hour/minute/second word in singular or plural forms with "ago" suffix. Different languages handles pluralization/cardinal numbering in slight different ways. Here we need to support English and Polish languages.

In English there are just two cases to cover per hour/minute/second word:

  • 1 - "one second ago"
  • 0, 2, 3... "%d seconds ago"
  • Same with minutes and hours.

In Polish it is more tricky because the cardinal numbers are more complicated:

  • 1 - "jedną sekundę temu"
  • 0, (5 - 21) - "%d sekund temu"
  • (2 - 4), (22-24), (32-34), (42, 44), ..., (162-164), ... - "%d sekundy temu"
  • Same logic for minutes and hours.

Following chapters will present solution without and with Swifternalization framework. Each solution describes Base (English) and Polish localizations.

Here is a table with Language Plural Rules which covers cardinal forms of numbers in many languages - Many language handle plurality in their own way.

Solution without Swifternalization

Localizable.strings (Base)
"one-second" = "1 second ago";
"many-seconds" = "%d seconds ago";

"one-minute" = "1 minute ago";    
"many-minutes" = "%d minutes ago";

"one-hour" = "1 hour ago";
"many-hours" = "%d hours ago";

Localizable.strings (Polish)
"one-second" = "1 sekundę temu"
"few-seconds" = "%d sekundy temu"
"many-seconds" = "%d sekund temu""    	    	

"one-minute" = "1 minutę temu"
"few-minutes" = "%d minuty temu	"
"many-minutes" = "%d minut temu" 	    	

"one-hours" = "1 godzinę temu"
"few-hours" = "%d godziny temu"
"many-hours" = "%d godzin temu";

There are 6 cases in English and 9 cases in Polish. Notice that without additional logic we're not able to detect which version of a string for hour/minute/second the app should display. The logic differs among different languages. We would have to add some lines of code that handle the logic for all the languages we're using in the app. What if there are more than 2 languages? Don't even think about it - this might be not so easy.

The logic is already implemented in Swifternalization framework and it fits to every language.

Solution with Swifternalization

This is how localizable files will look:

"time-seconds": {
    "one": "%d second ago"
    "other": "%d seconds ago"

"time-minutes": {
    "one": "%d minute ago"
    "other": "%d minutes ago"

"time-hours": {
    "one": "%d hours ago"
    "other": "%d hours ago"

"time-seconds": {
	"one": "1 sekundę temu",
	"few": "%d sekundy temu",
	"many": "%d sekund temu"

"time-minutes": {
	"one": "1 minutę temu",
	"few": "%d minuty temu",
	"many": "%d minut temu"

"time-hours": {
	"one": "1 godzinę temu",
	"few": "%d godziny temu",
	"many": "%d godzin temu"
  • "one", "few", "many", "other" - those are shared expressions already built into Swifternalization - covered below.
  • You can add own expressions to handle specific cases - covered below.

As mentioned the logic is implemented into framework so if you want to get one of a localized string you have to make a simple call.

Swifternalization.localizedString("time-seconds", intValue: 10)

or with I18n typealias (I-18-letters-n, Internalization):

I18n.localizedString("time-seconds", intValue: 10)

The key and intValue parameters are validated by loaded expressions and proper version of a string is returned - covered below.



Swifternalization drops necessity of using stringdicts files like following one to support pluralization in localized strings. Instead of that you can simply define expressions that cover such cases.

<plist version="1.0">
        <key>%d file(s) remaining</key>
                <string>%d file remaining</string>
                <string>%d files remaining</string>

No more stringsdict files!

Length Variations

iOS 9 provides new way to select proper localized string variation depending on a screen width. It uses stringsdict file with NSStringVariableWidthRuleType key.

Swifternalization drops necessity of using such file and it is not necessary to use this new key to use the feature.

With Swifternalization this length variations feature is available since iOS 8.0 because the framework has its own implementation of length variations.

To use length variations feature your translation file should has entries like this:

"forgot-password": {
	"@200": "Forgot Password? Help.",
	"@300": "Forgot Your Password? Use Help.",
	"@400": "Do not remember Your Password?" Use Help.""

The number after @ sign is max width of a screen or bounds that a string fits to. E.g. the second string will be returned if passed fitting width as a paramter is be greater than 200 and less or equal 300.

To get the second localized string the call looks like below:

I18n.localizedString("forgot-password", fittingWidth: 300) // 201 - 300

You can mix expressions with length variations. Following example shows it:

"car": {
    "ie:x=1": {
        @100: "One car",
        @200: "You've got one car"

    "more": "You've got few cars"


There are few expression types. Every expression type has their own parser and matcher but they work internally so you don't need to worry about them.

There are 3 types of expressions:

  • inequality - handles simple inequalities like: x<3, x>10, x=5, x<=3, and so on. Work with integer and float numbers.
  • inequality extended - extended version of inequality with syntax like this: 2<x<10, 4<=x<6. Work with integer and float numbers.
  • regex - uses regular expression. This is the most powerful ;)

Inequality Expressions

It is composed of several elements:

  • ie: - prefix of inequality expression
  • x - you have to always pass it, this means here is the place for a number that will be matched. Works with Ints and floating point numbers.
  • <, <=, =, >=, > - use one of inequality signs
  • 1, 3, 5, 6, ... - value to match is the last one in this expression


"cars": {
	"ie:x=1": "1 car",
	"ie:x=0": "no cars",
	"ie:x>1": "%d cars"

Inequality Extended Expressions

This is extended version of inequality expression. It is composed of 2 values, one value "marker" and two inequality signs.

  • iex: - prefix of inequality extended expression
  • x - place for number that will be matched. Works with Ints and floating point numbers.
  • Only < and <= are accepted.


"tomatos": {
	"iex:2<x<10": "%d tomatos is between 2 and 10"

Regex Expressions

This is the most powerful type of expression. It takes regular expression ;)

  • exp: - prefix of regex expression
  • string - it takes string with regular expression

Example: (police cars in Polish language)

"police-cars": {
	"exp:^1$": "1 samochód policyjny",
	"exp:(((?!1).[2-4]{1})$)|(^[2-4]$)": "%d samochody policyjne",
	"exp:(.*(?=1).[0-9]$)|(^[05-9]$)|(.*(?!1).[0156789])": "%d samochodów policyjnych"

Powerful stuff, isn't it? :>

PS. There is built-in solution for Polish language so you can use it with doing just this:

"police-cars": {
	"one": "1 samochód policyjny",
	"few": "%d samochody policyjne",
	"many": "%d samochodów policyjnych"

This is called "Shared Built-In Expression" and is covered below.

Shared Expressions

Shared expressions are expressions available among all the localization files. They are declared in expressions.json file divided by language and you can use them in localization files.

The functionality allows developer to observance of DRY principle and to avoid mistakes that exist because of reapeating the code in many places.

Normally you declare expression like this:

"ie:x>1": "Localized String"

If you want to use the same expression in multiple files there is no necessity to repeat the expression elsewhere. This is even problematic when you decide to improve/change expression to handle another cases you forget about - you would have to change expression in multiple places. Because of that there are Shared Expression. These feature allows you to create expression just in one place and use identifier of it in multiple places where you normally should put this expression.

What you need to do is to create expressions.json file with following structure:

	"base": {
		"one": "ie:x>1"

	"pl": {
		// ... other than "one" because "one" is available here too.

Now in pl.json, en.json and so on you have to use it as below:

"one": "Localized String"

Before you decide to create your own expression take a look if there is no built-in one with the same name or whether there is such expression but named differently. Maybe you don't need to do this at all and just use it.

Built-in expressions

Built-in expressions as name suggest are shared expressions built into framework and available to use with zero configuration. They are separated by country and not all country have its own built-in expressions. For now there are e.g. Base built-in expressions and Polish built-in expressions. Base expressions are available in every country and there are very generic to match all countries pluralization/cardinal numbering logic.

List of supported built-in shared expressions:

Base (English fits to this completely)
- one - detects if number is equal 1
- >one - detects if number is greater than 1
- two - detects if number is equal 2
- other - detects if number is not 1, so 0, 2, 3 and so on.

- few - matches (2-4), (22-24), (32-4), ..., (..>22, ..>23, ..>24)
- many - matches 0, (5-9), (10-21), (25-31), ..., (..0, ..1, ..5-9)

As you can see polish has no "one", ">one", etc. because it inherits from Base by default.

Getting Started

This chapter shows you how to start using Swifternalization and how to intergrate it with your code.


Documentation covers 100% of the code, Yay! There are two types of documentation. First covers only public API which is for those who only want to use the framework without looking inside. The second one covers all the API - public, internal and private.

You can find Public API and Full documentation with docset here in docs directory.

It is also hosted on my blog:



It works with iOS 8.0 and newer.

With CocoaPods:

pod 'Swifternalization', '~> 1.3.1'

If you are not using CocoaPods just import files from Swifternalization/Swifternalization directory to your project.

Swifternalization also supports Carthage.

Configuration - Optional

Before you get a first localized string you have to configure Swifternalization by passing to it the bundle where localized json files are placed.
If you do not call configure() it will use NSBundle.mainBundle() by default internally.

It will get languages from bundle which was used to configure it. So, if it does not translate you string, please make sure that you added Localizations in Project > Info > Localizations section - the same place like for regular localizations.

I18n.configure(bundle) // if files are in another bundle

Creating file with Shared Expressions

Shared Expressions must be placed in expressions.json. Syntax of a file looks like below:

    "base": {
        "ten": "ie:x=10",
        ">20": "ie:x>20",
        "custom-pl-few": "exp:(.*(?=1).[0-9]$)|(^[05-9]$)|(.*(?!1).[0156789])"

    "pl": {
        "few": "exp:(((?!1).[2-4]{1})$)|(^[2-4]$)",
        "two": "ie:x=2",
        "three": "ie:x=3"

In pseudo-language:

	"language-1": {
		"shared-expression-key-1": "expression-1",
		"shared-expression-key-2": "expression-2"

	"language-2": {
		"shared-expression-key-1": "expression-1"

Expressions from the files may be used inside localizable files. All the shared expressions for different languages are placed in the same file because there will be just few expressions for every language. Mostly the expression will be defined in base variant because if expression is in base it is also available in every other language too. So, "ten" is available in "pl", but "three" is not available in "base".

Creating file with localization per country

Localizable file contains translations for specific language. The files might look like below:

    "welcome-key": "welcome",

    "cars": {
        "one": "one car",
        "ie:x>=2": "%d cars",
        "ie:x<=-2": "minus %d cars"

Name of a file should be the same like country code. e.g. for English it is en.json, for Polish it is pl.json, for base localization it is base.json, etc.

There are few things that you can place in such files. More complex file will look like below:

	"welcome": "welcome",

	"cars": {
		"one": {
			"@100": "one car",
			"@200": "You have one car",
			"@400": "You have got one car"

		"other": "%d cars"

In pseudo-language:

	"key": "value",

	"key": {
		"expression-1": {
			"length-variation-1": "value-1",
			"length-variation-2": "value-2",
			"length-variation-3": "value-3"

		"expression-2": "value"

Getting localized string

Swifternalization allows you to work with its one class method which exposes all the methods you need to localize an app.

These methods have many optional paramters and you can omit them if you want. There are few common parameters:

  • key - A key of localized string.
  • fittingWidth - A width of a screen or place where you want to put a localized string. It is integer.
  • defaultValue - A value that will be returned if there is no localized string for a key passed to the method. If this is not specified then key is returned.
  • comment - A comment used just by developer to know a context of translation.

First method called localizedString(_:fittingWidth:defaultValue:comment:) allows you to get value for simple key without expression.


I18n.localizedString("welcome", fittingWidth: 200)
I18n.localizedString("welcome", defaultValue: "Welcome", comment: "Displayed on app start")

Next method localizedString(_:stringValue:fittingWidth:defaultValue:comment:) allows you to get a localized string for string value that match an expression. Actually the string value will contain number inside in most cases or some other string that you would like to match.

I18n.localizedString("cars", stringValue: "5")
// Other cases similar to above example

The last method localizedString(_:intValue:fittingWidth:defaultValue:comment:) allows you to get a localized string for int value. This method calls the above one and just turn the int value into string because all the framework operates on strings.

I18n.localizedString("cars", intValue: 5)


Swifternalization is open sourced so everyone may contribute if want to. If you want to develop it just fork the repo, do your work and create pull request. If you have some idea or question feel free to create issue and add proper tag for it.

There is no guide for contributors but if you added new functionality you must write unit tests for it.

Things to do in future releases

  • Add more built-in expressions for another countries.
  • Add support for float numbers in built in expressions that uses regular expressions.


Swifternalization is released under the MIT license.