How to convert Microsoft Locale ID (LCID) into language code or Locale object in Java

Question

I need to translate a Microsoft locale ID, such as 1033 (for US English), into either an ISO 639 language code or directly into a Java Locale instance. (Edit: or even simply into the "Language - Country/Region" in Microsoft's table.)

Is this possible, and what's the easiest way? Preferably using only JDK standard libraries, of course, but if that's not possible, with a 3rd party library.

Answer 1

The was the first hit on google for "Java LCID" is this javadoc:

gnu.java.awt.font.opentype.NameDecoder

private static java.util.Locale getWindowsLocale(int lcid)

Maps a Windows LCID into a Java Locale.

Parameters:
    lcid - the Windows language ID whose Java locale is to be retrieved. 
Returns:
    an suitable Locale, or null if the mapping cannot be performed.

I'm not sure where to go about downloading this library, but it's GNU, so it shouldn't be too hard to find.

Answer 2

As it started to look like there is no ready Java solution to do this mapping, we took the ~20 minutes to roll something of our own, at least for now.

We took the information from the horse's mouth, i.e. http://msdn.microsoft.com/en-us/goglobal/bb964664.aspx, and copy-pasted it (through Excel) into a .properties file like this:

1078 = Afrikaans - South Africa
1052 = Albanian - Albania
1118 = Amharic - Ethiopia
1025 = Arabic - Saudi Arabia
5121 = Arabic - Algeria 
...

(You can download the file here if you have similar needs.)

Then there's a very simple class that reads the information from the .properties file into a map, and has a method for doing the conversion.

Map<String, String> lcidToDescription;

public String getDescription(String lcid) { ... }

And yes, this doesn't actually map to language code or Locale object (which is what I originally asked), but to Microsoft's "Language - Country/Region" description. It turned out this was sufficient for our current need.

Disclaimer: this really is a minimalistic, "dummy" way of doing it yourself in Java, and obviously keeping (and maintaining) a copy of the LCID mapping information in your own codebase is not very elegant. (On the other hand, neither would I want to include a huge library jar or do anything overly complicated just for this simple mapping.) So despite this answer, feel free to post more elegant solutions or existing libraries if you know of anything like that.

Answer 3

You could use GetLocaleInfo to do this (assuming you were running on Windows (win2k+)).

This C++ code demonstrates how to use the function:

#include "windows.h"

int main()
{
  HANDLE stdout = GetStdHandle(STD_OUTPUT_HANDLE);
  if(INVALID_HANDLE_VALUE == stdout) return 1;

  LCID Locale = 0x0c01; //Arabic - Egypt
  int nchars = GetLocaleInfoW(Locale, LOCALE_SISO639LANGNAME, NULL, 0);
  wchar_t* LanguageCode = new wchar_t[nchars];
  GetLocaleInfoW(Locale, LOCALE_SISO639LANGNAME, LanguageCode, nchars);

  WriteConsoleW(stdout, LanguageCode, nchars, NULL, NULL);
  delete[] LanguageCode;
  return 0;
}

It would not take much work to turn this into a JNA call. (Tip: emit constants as ints to find their values.)

Sample JNA code:

• draw a Windows cursor
• print Unicode on a Windows console

Using JNI is a bit more involved, but is manageable for a relatively trivial task.

At the very least, I would look into using native calls to build your conversion database. I'm not sure if Windows has a way to enumerate the LCIDs, but there's bound to be something in .Net. As a build-level thing, this isn't a huge burden. I would want to avoid manual maintenance of the list.