Charset_Test.java

/*
 * Copyright (c) 2020, https://github.com/911992 All rights reserved.
 * License BSD 3-Clause (https://opensource.org/licenses/BSD-3-Clause)
 */

 /*
sample_code_java0
File: Charset_Test.java
Created on: May 28, 2020 7:19:32 PM
    @author https://github.com/911992
 
History:
    initial version: 0.1(20200528)
 */
package def;

/**
 *
 * @author https://github.com/911992
 */
public class Charset_Test {

    public static void main(String[] args) throws java.lang.Exception {
        String _hp_jp = "ハッピープログラミング";
        byte _hp_jp_utf8_arr[] = _hp_jp.getBytes("utf-8");
        System.out.printf("JP text len:%d , text bytes len(utf8):%d\n", _hp_jp.length(), _hp_jp_utf8_arr.length);

        String _hp_en = "Happy Programming";
        byte _hp_en_ascii[] = _hp_en.getBytes("utf-8");
        System.out.printf("En text len:%d , text bytes len(utf8):%d\n", _hp_en.length(), _hp_en_ascii.length);

        byte _hp_en_utf16[] = _hp_en.getBytes("utf-16be");
        System.out.printf("Text bytes len(utf-16be):%d\n", _hp_en_utf16.length);

        byte _hp_en_utf32[] = _hp_en.getBytes("utf-32le");
        System.out.printf("Text bytes len(utf-32le):%d\n", _hp_en_utf32.length);

        String _jp_tsu_str = "ッ"; //https://www.compart.com/en/unicode/U+30C3
        char _jp_tsu_c = _jp_tsu_str.charAt(0);
        byte _jp_tsu_arr[] = _jp_tsu_str.getBytes("utf-8");
        System.out.printf("Tsu (local) charpoint hex:0x%x\n", (long) _jp_tsu_c);//30c3
        StringBuilder _jp_tsu_arr_p = new StringBuilder();
        StringBuilder _jp_tsu_arr_p_s = new StringBuilder();
        for (int a = 0; a < _jp_tsu_arr.length; a++) {
            _jp_tsu_arr_p.append(String.format("%02x", _jp_tsu_arr[a]));
            if (a > 0) {
                _jp_tsu_arr_p_s.append(", ");
            }
            _jp_tsu_arr_p_s.append(String.format("[0x%02x] ", _jp_tsu_arr[a]));
        }
        System.out.printf("Tsu (arr) bytes:%s (%s)\n", _jp_tsu_arr_p.toString(), _jp_tsu_arr_p_s.toString()); //e38383
    }
    
//    OUTPUT:
//    
//    JP text len:11 , text bytes len(utf8):33
//    En text len:17 , text bytes len(utf8):17
//    Text bytes len(utf-16be):34
//    Text bytes len(utf-32le):68
//    Tsu (local) charpoint hex:0x30c3
//    Tsu (arr) bytes:e38383 ([0xe3] , [0x83] , [0x83] )
}