Skip to content

Transliteration & parsing

The "inverse and transform" half of ICU: convert text between scripts, detect spoofed look-alike strings, and parse locale-formatted numbers, money, and dates back into values.

Availability

Everything on this page needs a raw-ICU service — Transliterator, SpoofChecker, or a parser — that the Intl API does not expose. So these methods exist in PHP, Python, and Java only; the tabs below omit JS. See Platform notes.

Transliterate & romanize

transliterate(text, id) runs any ICU transform ("Greek-Latin", "Any-Hangul", "Any-Latin; Latin-ASCII", …). romanize(text) is the common case — to Latin, then stripped to ASCII — ideal for slugs and search keys.

Cosmo c = new Cosmo("en");
c.romanize("Москва");                                // "Moskva"
c.transliterate("Łódź café", "Any-Latin; Latin-ASCII"); // "Lodz cafe"

$c = new Cosmo('en');
$c->romanize('Москва');                              // "Moskva"
$c->transliterate('Łódź café', 'Any-Latin; Latin-ASCII'); // "Lodz cafe"

c = Cosmo("en")
c.romanize("Москва")                                 # "Moskva"
c.transliterate("Łódź café", "Any-Latin; Latin-ASCII") # "Lodz cafe"

An unknown transform id throws. The full list of available ids comes from supportedValues("transliterator") (the one supportedValues key beyond the standard six).

Spoof detection

confusable(a, b) asks whether two strings are visually confusable (e.g. a Cyrillic look-alike of a Latin brand). suspicious(text) flags a single string as spoof-prone — typically mixed-script.

Backed by UTS #39

Both checks implement Unicode Technical Standard #39, Unicode Security Mechanisms — the standard that defines the confusable-character data and mixed-script detection rules — through ICU's SpoofChecker.

Cosmo c = new Cosmo("en");
c.confusable("paypal", "раураl");    // true
c.confusable("hello", "world");      // false
c.suspicious("pаypal");              // true
c.suspicious("paypal");              // false

$c = new Cosmo('en');
$c->confusable('paypal', 'раураl');  // true   (Cyrillic look-alike)
$c->confusable('hello', 'world');    // false
$c->suspicious('pаypal');            // true   (mixed Latin/Cyrillic)
$c->suspicious('paypal');            // false

c = Cosmo("en")
c.confusable("paypal", "раураl")     # True
c.confusable("hello", "world")       # False
c.suspicious("pаypal")               # True
c.suspicious("paypal")               # False

Use these to defend account names, domains, and brand mentions against homograph attacks.

Parsing (inverse formatters)

Intl can only format, never parse — but ICU's parsers are available in the other three ports. Each is the inverse of a formatter on this site.

new Cosmo("de").parseNumber("1.234,56");         // 1234.56
new Cosmo("en").parseNumber("1,234.56");         // 1234.56
new Cosmo("en_US").parseMoney("$12.30");         // CurrencyAmount(12.30, USD)

Cosmo utc = new Cosmo("en_US", new Modifiers(null, null, "UTC"));
utc.parseDate("February 2, 2020", "long");       // java.util.Date
utc.parseMoment("2020-02-02", "yyyy-MM-dd");     // java.util.Date

new Cosmo('de')->parseNumber('1.234,56');       // 1234.56
new Cosmo('en')->parseNumber('1,234.56');        // 1234.56
new Cosmo('en_US')->parseMoney('$12.30');        // ['amount' => 12.3, 'currency' => 'USD']

$utc = new Cosmo('en_US', ['timeZone' => 'UTC']);
$utc->parseDate('February 2, 2020', 'long');     // DateTimeImmutable
$utc->parseMoment('2020-02-02', 'yyyy-MM-dd');   // DateTimeImmutable

Cosmo("de").parse_number("1.234,56")             # 1234.56
Cosmo("en").parse_number("1,234.56")             # 1234.56
Cosmo("en_US").parse_money("$12.30")             # {"amount": 12.3, "currency": "USD"}

utc = Cosmo("en_US", {"timeZone": "UTC"})
utc.parse_date("February 2, 2020", "long")       # datetime
utc.parse_moment("2020-02-02", "yyyy-MM-dd")     # datetime

parseMoney() returns the amount and the recognised currency (an array/dict in PHP/Python, an ICU CurrencyAmount in Java). Text that doesn't parse throws, so you can distinguish "0" from "unparseable".