Core Java Advanced LiveLessons

Internationalization

Slide navigation: Forward with space bar, → arrow key, or PgDn. Backwards with ← or PgUp.

.jpg

Understand the concept of a locale

Locales

A locale specifies the language, location, and preferences for formatting information of users of your software.
Examples:
- German users in Switzerland.
- Thai users who prefer Thai numerals.

Number formatting depends on the locale:

123,456.78 (United States)
123.456,78 (Switzerland)
๑๒๓,๔๕๖.๗๘ (Thailand with Thai numerals)

With dates, the differences are more pronounced:

March 22, 1961
22. März 1961
1961年3月22日

Specifying a Locale

A locale consists of up to five components:
1. A language, specified by two or three lowercase letters, such as de or zh.
2. Optionally, a script, specified by four letters with initial caps, such as Latn, Cyrl, Hant.
3. Optionally, a country or region, with two uppercase letters or three digits, such as US or CH.
4. Optionally, a variant (rare).
5. Optionally, an extension such as u-nu-thai or x-java.
You can get a Java Locale object from a locale string, from one of the predefined locales, or for the current locale of the user's computer:
```
Locale locale = Locale.forLanguageTag("en-US");
locale = Locale.CANADA_FRENCH;
locale = Locale.getDefault(Locale.Category.DISPLAY);
```

jshell

Locale en_US = Locale.forLanguageTag("en-US");
Locale fr_CA = Locale.CANADA_FRENCH;
en_US.getDisplayName()
en_US.getDisplayName(fr_CA)

Use locale-specific formatting for numbers and dates

Number Formats

The java.text.NumberFormat class has factory methods for formatting numbers, percentages, and currencies:

Locale loc = Locale.GERMANY;
NumberFormat formatter = NumberFormat.getCurrencyInstance(loc);
double amt = 123456.78;
String result = formatter.format(amt);
    // Yields 123.456,78€

Use the Currency class for currencies that are from a different locale:

formatter.setCurrency(Currency.getInstance("USD"));
System.out.println(formatter.format(dollars));

To read numbers, use the NumberFormat.parse method:

String input = "123.456,78€";
Number parsed = formatter.parse(input);
double x = parsed.doubleValue();

If the format isn't correct, a NumberFormatException is thrown.
When reading numbers from a Scanner, you can set its locale with the useLocale method.
The String.format and PrintWriter.printf methods take a Locale argument.

jshell

import java.text.*;
Locale de_DE = Locale.GERMANY;
NumberFormat formatter = NumberFormat.getCurrencyInstance(de_DE);
double amt = 123456.78;
formatter.format(amt)
NumberFormat.getCurrencyInstance(Locale.US).format(amt)
NumberFormat.getCurrencyInstance(Locale.forLanguageTag("th-TH-u-nu-thai")).format(amt)

Date and Time Formats

Use java.time.DateTimeFormatter, not the legacy java.text.DateFormat class.

Pick a format:

Style	Date	Time
`SHORT`	`7/16/69`	`9:32 AM`
`MEDIUM`	`Jul 16, 1969`	`9:32:00 AM`
`LONG`	`July 16, 1969`	`9:32:00 AM EDT`
`FULL`	`Wednesday, July 16, 1969`	`9:32:00 AM EDT`

Get a formatter:

DateTimeFormatter formatter =
   DateTimeFormatter.ofLocalizedDateTime(style).withLocale(locale);
    // Or ofLocalizedDate/ofLocalizedTime

Format a LocalDate, LocalDateTime, LocalTime, or ZonedDateTime:

ZonedDateTime appointment = ...;
String formatted = formatter.format(appointment);

Each of these date/time classes has a static parse method:
```
LocalTime time = LocalTime.parse("9:32 AM", formatter);
```

jshell

import java.time.*;
import java.time.format.*;
DateTimeFormatter formatter =
   DateTimeFormatter.ofLocalizedDate(FormatStyle.LONG);
LocalDate now = LocalDate.now();
formatter.withLocale(Locale.US).format(now);

Repeat with Locale.FRENCH and Locale.CHINESE

Month and Weekday Names

Sometimes, you need to display the names of months or weekdays independent of a date.

Pick a text style:

Style	Example
`FULL` / `FULL_STANDALONE`	January
`SHORT` / `SHORT_STANDALONE`	Jan
`NARROW` / `NARROW_STANDALONE`	J

The STANDALONE versions are for display outside a formatted date.
- For example, in Finnish, January is “tammikuuta” inside a date, but “tammikuu” standalone.

Then call the getDisplayName methods of the DayOfWeek and Month enumerations:

for (Month m : Month.values())
   System.out.println(m.getDisplayName(textStyle, locale));

The first day of the week depends on the locale:

DayOfWeek first = WeekFields.of(locale).getFirstDayOfWeek();

JShell

import java.time.*;
import java.time.format.*;
import java.util.stream.*;

List<String> weekdayNames(Locale locale) {
   return Stream.of(DayOfWeek.values()).
      map(m -> m.getDisplayName(TextStyle.FULL, locale)).
      collect(Collectors.toList());
}

weekdayNames(Locale.US)
weekdayNames(Locale.CHINESE)
weekdayNames(Locale.forLanguageTag("fi"))

import java.time.temporal.*

WeekFields.of(Locale.US).getFirstDayOfWeek()
WeekFields.of(Locale.GERMAN).getFirstDayOfWeek()

Work with strings in multiple languages

Collation

Don't use String.compareTo for comparing humanly readable strings.
Not even in English:
```
Athens
Zulu
able
zebra
Ångström
```
Use collation—sorting according to locale preferences.
- In Swedish, Å comes after Z.

The Collator class implements the Comparator<Object> interface:

Collator coll = Collator.getInstance(locale);
words.sort(coll);

You can optionally call setStrength to make the collator less selective.
- Collator.PRIMARY: e ≠ f, e = é, e = E
- Collator.SECONDARY: e ≠ f, e ≠ é, e = E
- Collator.TERTIARY (default): e ≠ f, e ≠ é, e ≠ E

String Normalization

In Unicode, some characters can be expressed in multiple ways.
- An é (U+00E9) is equivalent to a plain e (U+0065) followed by a ´ (combining acute accent U+0301).
The Unicode standard defines four normalization forms.
- C: Always use the composed characters (é).
- D: Always use the decomposed characters (e ´).
- KC, KD: Also decompose characters such as ™.

Use java.text.Normalizer to compute the normalized form of any string:

String city = "San Jos\u0301";
String normalized = Normalizer.normalize(city, Normalizer.Form.NFC);

With a collator, call setDecomposition to make the collator less selective.
- Collator.FULL_DECOMPOSITION: é = e´, ™ = TM
- Collator.NO_DECOMPOSITION (default): é = e´, ™ ≠ TM
- Collator.IDENTICAL: é ≠ e´, ™ ≠ TM

lesson07/collation

Letter Case

Forming uppercase and lowercase strings is locale-specific.
In Turkish, the lowercase of I is a dotless ı.
In German, the uppercase of ß is SS.

Supply Locale argument when calling toUpperCase/toLowerCase:

String street = "Einbahnstraße"
String uppercase = street.toUpperCase(Locale.GERMAN);
    // "EINBAHNSTRASSE"

Note that s.toLowerCase() and s.toUpperCase().toLowerCase() may be different.

Message Formatting

Messages with variable parts may require reordering in different languages:
- Pierre has 42 messages.
- Il y a 42 messages pour Pierre.

Use MessageFormat.format with placeholders:

String template = "{0} has {1} messages";
    // In French, we'll use "Il y a {1} messages pour {0}"
String text = MessageFormat.format(template, "Pierre", 42);

You can only use the static format method with the default locale. For other locales, construct a MessageFormat object:

MessageFormat mf = new MessageFormat(template, locale);
String text = mf.format(new Object[] { "Pierre", 42 }); // No varargs ☹

To format numbers, add number and currency, integer, percent, or a number format pattern of the DecimalFormat class (such as $,##0) to the placeholder:
```
template="Your current total is {0,number,currency}."
    // Will yield a string such as "Your current total is $1023.95."
```
For dates, add date or time, followed by the format short, medium, long, or full, or a format pattern of the SimpleDateFormat class (such as yyyy-MM-dd):
```
String message = MessageFormat("It is now {0,time,short}.", Date.from(Instant.now()));
```
Note that you need to convert java.time objects to the legacy java.util.Date type.

Choices

A choice format lets you generate messages such as:
```
No files copied
1 file copied
42 files copied
```
A choice format is a sequence of pairs of the form:
```
lowerLimit#formatString
```
The first limit is ignored.

The pairs are separated by | symbols:

String template = "{0,choice,0#No files|1#1 file|2#{0} files} copied";

Note that 0 appears twice: to make the choice, and to insert the count into the third choice.
You can use ≤ instead of #, < for a “less than” bound, and -∞ for the unused first bound:
```
-∞<No files|0<1 file|2≤{0} files
```
Caution: Strings enclosed in single quotes are included literally. Use '' to include single quotes in the template:
```
String template = "<a href=''{0}''>{1}</a>";
```

Organize locale-specific data in resource bundles

Resource Bundles

Use resource bundles to separate locale-specific parts from the rest of your application.
A resource bundle is either a set of property files or a class extending ResourceBundle.
A property file has extension .properties and contains key/value pairs:
```
computeButton=Rechnen
cancelButton=Abbrechen
defaultPaperSize=A4
```
Naming convention: bundleName_language_script_country
Hierarchical lookup: bundleName_de_CH before bundleName_de
Files are loaded by class loader.
- Place files for a bundle with name com.mycompany.messages into com/mycompany directory.

Load a bundle like this:

ResourceBundle bundle = ResourceBundle.getBundle(bundleName, locale);

Then you can look up keys:

String computeButtonLabel = bundle.getString("computeButton");

lesson07/retire

Java 9 News Flash - Resource Files

.png

Resource files are now UTF-8.
- No more file.preferences=Pr\u00e9f\u00e9rences