How do I convert an Arabic String date to a java 8 date object?

Question

I'm working on a web monitoring project in Arabic and I want to convert string date like this one:

الاثنين 24 أبريل 2017 - 15:00

to Java 8 date object. How can I do that?

Answer 1

Edit: with thanks to slim and Meno Hochschild for inspiration:

String dateTimeString = "الاثنين 24 أبريل 2017 - 15:00";

DateTimeFormatter formatter
        = DateTimeFormatter.ofPattern("EEEE d MMMM uuuu - HH:mm", new Locale("ar"));
LocalDateTime dateTime = LocalDateTime.parse(dateTimeString, formatter);
System.out.println(dateTime);

This prints:

2017-04-24T15:00

Answer 2

The answers of @Ole and @slim are working, but not for the reason they think.

First observation - the nu-extension is unnecessary for given example:

Oles suggestion would also work for the locale new Locale("ar", "SA") instead of Locale.forLanguageTag("ar-SA-u-nu-arab"). So what does the unicode-nu-extension here? Nothing. Next question:

What is the nu-extension supposed to do here?

The nu-code-word "arab" is specified by the unicode consortium to yield arabic-indic digits. But the input to be parsed does only have western digits 0-9 (which are historically overtaken from Arab people and specified as code word "latn" - a misnomer by the way). So if the nu-extension had really done its job here then parsing should have failed because arabic-indic digits are not 0-9 but:

٠ ١ ٢ ٣ ٤ ٥ ٦ ٧ ٨ ٩

Obviously, the nu-extension is not supported in general by new time-API in Java-8.

Does SimpleDateFormat support the nu-extension?

Using debugging of following code, I discovered that the nu-extension is only supported for Thai-numerals (see also official javadoc of class java.util.Locale but not for arabic-indic digits:

SimpleDateFormat sdf = 
    new SimpleDateFormat("EEEE d MMMM yyyy - HH:mm", Locale.forLanguageTag("ar-SA-nu-arab"));
Date d = sdf.parse(dateTimeString);
System.out.println(d);
String formatted = sdf.format(d);
System.out.println(formatted);
System.out.println(sdf.format(d).equals(dateTimeString));

sdf = new SimpleDateFormat("EEEE d MMMM uuuu - HH:mm", Locale.forLanguageTag("ar-SA-u-nu-thai"));
String thai = sdf.format(d);
System.out.println("u-nu-thai: " + thai);

I assume the class DateTimeFormatter of Java-8 also supports Thai numerals.

Conclusion:

Forget the nu-extension. Just construct the locale via the old-fashioned way without unicode extension and adapt Oles answer this way. It works because your input only has western digits 0-9.

For extensive i18n-support including the nu-extension for various numbering systems (if you have such input), you might consider external libraries (for example ICU4J or my lib Time4J).

Answer 3

I don't know enough Arabic to understand an Arabic formatted date. However this code:

Locale arabicLocale = new Locale.Builder().setLanguageTag("ar-SA-u-nu-arab").build();

LocalDate date = LocalDate.now();
DateTimeFormatter formatter = DateTimeFormatter.ofLocalizedDate(FormatStyle.FULL).withLocale(arabicLocale);

String formatted = date.format(formatter);
System.out.println(formatted);
System.out.println(formatter.parse(formatted));

Yields this output:

26 أبريل, 2017
{},ISO resolved to 2017-04-26

The code to create the Locale is from an answer to Setting Arabic numbering system locale doesn't show Arabic numbers

You can fine-tune this format by defining your own FormatStyle.

Answer 4

You have to specify the charset when parsing the string, assuming that the date you want to parse will always be in the format you provided this would work :

public static Date getDate(String strDate) throws Exception{
    strDate=new String(strDate.getBytes(),"UTF-8");

    Map<String, Integer> months = new HashMap<>();

    String JAN =  new String("يناير".getBytes(), "UTF-8");
    String FEB =  new String("فبراير".getBytes(), "UTF-8");
    String MAR =  new String("مارس".getBytes(), "UTF-8");
    String APR =  new String("أبريل".getBytes(), "UTF-8");
    String APR_bis =  new String("ابريل".getBytes(), "UTF-8");
    String MAY =  new String("ماي".getBytes(), "UTF-8");
    String JUN =  new String("بونيو".getBytes(), "UTF-8");
    String JUN_bis =  new String("يونيه".getBytes(), "UTF-8");
    String JUL =  new String("يوليوز".getBytes(), "UTF-8");
    String AUG =  new String("غشت".getBytes(), "UTF-8");
    String SEP =  new String("شتنبر".getBytes(), "UTF-8");
    String SEP_bis =  new String("سبتمبر".getBytes(), "UTF-8");
    String OCT =  new String("أكتوبر".getBytes(), "UTF-8");
    String OCT_bis =  new String("اكتوبر".getBytes(), "UTF-8");
    String NOV =  new String("نونبر".getBytes(), "UTF-8");
    String NOV_bis =  new String("نوفمبر".getBytes(), "UTF-8");
    String DEC =  new String("دجنبر".getBytes(), "UTF-8");
    String DEC_bis =  new String("ديسمبر".getBytes(), "UTF-8");



    months.put(JAN, 0);
    months.put(FEB, 1);
    months.put(MAR, 2);
    months.put(APR, 3);
    months.put(APR_bis, 3);
    months.put(MAY, 4);
    months.put(JUN, 5);
    months.put(JUN_bis, 5);
    months.put(JUL, 6);
    months.put(AUG, 7);
    months.put(SEP, 8);
    months.put(SEP_bis, 8);
    months.put(OCT, 9);
    months.put(OCT_bis, 9);
    months.put(NOV, 10);
    months.put(NOV_bis, 10);
    months.put(DEC, 11);
    months.put(DEC_bis, 11);


    StringTokenizer stringTokenizer = new StringTokenizer(strDate);

    Calendar calendar = Calendar.getInstance();


    while(stringTokenizer.hasMoreElements()) {

        stringTokenizer.nextElement();// to skip the first string which is the name of the day

        int day = Integer.parseInt(stringTokenizer.nextElement().toString().trim());

        String strMonth = stringTokenizer.nextElement().toString().trim();

        int month = months.get(strMonth);

        int year = Integer.parseInt(stringTokenizer.nextElement().toString().trim());

        calendar.set(year, month, day);


    }
    return calendar.getTime();

}

it gives this output :

  Fri Oct 20 15:26:47 WEST 2017

Answer 5

One solution could be to translate the date to English and parse it then:

private final static Map<String, Integer> monthMapping = new HashMap<>();
static {
    // list of all month.
    monthMapping.put("أبريل", "4");
}


public Date fromArabicToDate(String arabicInput) throws ParseException {
    String[] parts = arabicInput.split(" ");
    if (parts.length != 4) 
        throw new IllegalArgumentException();

    String dateInput = parts[0] + "-" + monthMapping.get(parts[1]) + "-" + parts[2];
    SimpleDateFormat parser = new SimpleDateFormat("YYYY-MM-DD");
    return parser.parse(dateInput);
}

I tried to copy the month but I don't believe I have done it correctly. The arguments of put get switched when parsing.

Or you have a look at Joda-Time. Maybe they have a solution. It was mentioned here.