Internationalization in PHP

August 16, 2013

This page was initially written in Lithuanian. The examples contain Lithuanian phrases.

One of the most actual problems that application developers have to solve is adapting application to different languages (internationalization, i18n) and multi-cultural preferences, i.e. number, date formats etc. (localication, l10n).

For example:

Internationalization problems are commonly solved using standart operating system tool - locales. I.e.:

setlocale(LC_TIME, 'lt_LT.UTF-8');
echo strftime('%c'); // 2013 m. rugpjūčio 16 d. 11:50:29
setlocale(LC_TIME, 'en_US.UTF-8');
echo strftime('%c'); // Fri 16 Aug 2013 11:50:29 AM EEST

However, locales do not solve all problems. Also, they must be installed in the operating system.

Under Linux OS you can see installed locale list using locale -a.

The project that solves more problems, related to unicode (UTF) internationalization, is ICU (Internationalization Components for Unicode):

ICU library has to be installed in the OS in order to use ICU (in case of Debian/Ubuntu, it is called libicu*)

PHP library is called intl. You can install it from PECL repository or if you’re using Ubuntu, as a package php5-intl.


Converting non-latin symbols to latin ones is a common problem, i.e.: when it is needed to form a clean URL.

Transliterator class can be used to perform such conversion:

$id = "Any-Latin; NFD; [:Nonspacing Mark:] Remove; NFC; [:Punctuation:] Remove; Lower();";
$transliterator = Transliterator::create($id);
$string = "ąčįū!?_-&% ĄČĘĖĮŠŲŪŽ";
echo $transliterator->transliterate($string);
// aciu aceeisuuz

If it is needed to generate URL slug, we can convert spaces to hyphens using RegExp:

echo preg_replace('/\s+/', '-', 'tekstas   tekstas2');
// tekstas-tekstas2

The argument passed to Transliterator::create() can be formed according the ICU transformation guide.

NFC and NFD are unicode normalization and denormalization functions.


Let’s assume we need to sort Lithuanian words: urvas, ūkas, Ukmergė, ugnis. The expected result is: urvas, ūkas, Ukmergė, ugnis.

Normally the PHP array is sorted like this:

$arr = ['urvas', 'ūkas', 'Ukmergė', 'ugnis'];
// Array ( [0] => Ukmergė [1] => ugnis [2] => urvas [3] => ūkas )

The result is incorrect due to wrong collation (the capital letter is sorted first, the u with macron is sorted the last).

It is possible to hint sort() to use system locale by passing SORT_LOCALE_STRING as second argument:

setlocale(LC_ALL, 'lt_LT.UTF-8');
$arr = ['urvas', 'ūkas', 'Ukmergė', 'ugnis'];
// Array ( [0] => ugnis [1] => ūkas [2] => Ukmergė [3] => urvas )

The result is correct, but it can have negative impact, because in this example the locale is set globally. Also, it must be installed in the operating system.

The other way is to use Collator class:

$arr = ['urvas', 'ūkas', 'Ukmergė', 'ugnis'];
$collator = new Collator('lt_LT');
// Array ( [0] => ugnis [1] => ūkas [2] => Ukmergė [3] => urvas )

Number formats

The number and currency formats are different in different languages. NumberFormatter can be used to display them correctly:

$ltNum = new NumberFormatter('lt_LT', NumberFormatter::CURRENCY);
echo $ltNum->formatCurrency(1234567890.25, 'LTL');
// 1,234,567,890.25 Lt
echo $ltNum->formatCurrency(1234567890.25, 'EUR');
// 1,234,567,890.25 €
echo $ltNum->formatCurrency(1234567890.25, 'USD');
// 1,234,567,890.25 US$

$enNum = new NumberFormatter('en_US', NumberFormatter::CURRENCY);
echo $enNum->formatCurrency(1234567890.25, 'LTL');
// LTL1,234,567,890.25
echo $enNum->formatCurrency(1234567890.25, 'EUR');
// €1,234,567,890.25
echo $enNum->formatCurrency(1234567890.25, 'USD');
// $1,234,567,890.25

As we see the currencies are displayed differently depending on locale.

NumberFormatter class can also convert numbers to text (spellout):

$ltNum = new NumberFormatter('lt_lt', NumberFormatter::SPELLOUT);
echo $ltNum->format(1234567890.25);
// vienas milijardas du šimtai trisdešimt keturi milijonų penki šimtai šešiasdešimt septyni tūkstančiai aštuoni šimtai devyniasdešimt kablelis du penki

$enNum = new NumberFormatter('en_US', NumberFormatter::SPELLOUT);
echo $enNum->format(1234567890.25);
// one billion two hundred thirty-four million five hundred sixty-seven thousand eight hundred ninety point two five

Text formatting

It is very important to ensure not only the correct text translation, but also, the format of dates, numbers and plural forms of the words.

MessageFormatter class is used to do it, i.e. to display dates correctly:

$enDate = new MessageFormatter('en_US', 'Today {0,date,short}');
echo $enDate->format(array(time()));
// Today 8/16/13
$enDate = new MessageFormatter('en_US', 'Today {0,date,long}');
echo $enDate->format(array(time()));
// Today August 16, 2013

$ltDate = new MessageFormatter('lt_LT', 'Šiandien {0,date,short}');
echo $ltDate->format(array(time()));
// Šiandien 2013-08-16
$ltDate = new MessageFormatter('lt_LT', 'Šiandien {0,date,long}');
echo $ltDate->format(array(time()));
// Šiandien 2013 m. rugpjūtis 16 d.

The locale and text format is passed to the constructor. The format() argument is the array of data that is used by formatter.

{0,date,short} means that the first format() array item is used as the date in short format.

MessageFormatter is also useful when working with plural forms:


$x = new MessageFormatter('lt_LT', 'Parduodu {0, plural, one{{0,number} obuolį} few{{0,number} obuolius} other{{0,number} obuolių}} ir {1, plural, one{{1,number} bandelę}few{{1,number} bandeles}other{{1,number} bandelių}} už {2,number,currency}');

echo $x->format(array(15, 2, 15));
// Parduodu 15 obuolių ir 2 bandeles už 15.00 Lt

More information in text formatting: Full list of plural forms:

Please give a feedback on working with intl functions with PHP since the functions are not fully documented in the manual. Thanks in advance :)