/*
 * TestGbk.java
 *
 * Copyleft (C) 2006 RatKing. All wrongs reserved.
 */

import java.io.UnsupportedEncodingException;

/**
 * 测试GBK汉字.
 *
 * 用法举例：
 *     java TestGbk "我是谁？Who am I?"
 *
 * @author <a href="mailto:ratking＠ynet·com">RatKing</a>
 * @version 1.0 (build 20060721)
 *
 * @see <a href="http://www.haiyan.com/steelk/navigator/ref/gbk/gbindex.htm">《GBK 汉字内码扩展规范》</a>
 */
public class TestGbk {
    /** GBK编码高字节最小值 */
    public static final int GBK_HIGH_MIN = 0x81;
    /** GBK编码高字节最大值 */
    public static final int GBK_HIGH_MAX = 0xFE;
    /** GBK编码低字节最小值 */
    public static final int GBK_LOW_MIN = 0x40;
    /** GBK编码低字节最大值 */
    public static final int GBK_LOW_MAX = 0xFE;
    /** GBK编码低字节例外值 */
    public static final int GBK_LOW_EXCEPTION = 0x7F;
    
    public static void main(String[] args) {
        String str = null;
        if (args == null || args.length == 0) {
            str = "Hello, world!世界，你好！";
        } else if (args.length != 1
                || (args.length == 1
                && (args[0].equals("/?")
                || args[0].equals("-h")
                || args[0].equals("-help")
                || args[0].equals("--help")))) {
            System.err.println("用法：java TestGbk [字符串]");
            System.exit(1);
        } else {
            str = args[0];
        }

        int gbkCount = 0;
        int asciiCount = 0;
        for (int i = 0; i < str.length(); i++) {
            char c = str.charAt(i);
            boolean isGbk = isGbk(c);
            System.out.println(i + "：[" + c + "] " + isGbk);
            if (isGbk) {
                gbkCount++;
            } else {
                asciiCount++;
            }
        }
        System.out.println("字符串：\"" + str + "\"");
        System.out.println("GBK汉字（含全角标点符号）数：" + gbkCount);
        System.out.println("非GBK汉字数：" + asciiCount);
        System.out.println("字符串总字符数：" + str.length());
        System.out.println("字符串总字节数：" + (gbkCount * 2 + asciiCount));
    }

    /**
     * 是否为GBK内码的字符.
     * 对英文字母、阿拉伯数字、半角标点符号，返回false；
     * 对中文汉字、全角标点符号，返回true。
     * @param c 待检测的字符
     * @return 是否为GBK内码的字符
     */
    public static boolean isGbk(char c) {
        String str = String.valueOf(c);
        byte[] b = null;
        try {
            b = str.getBytes("GBK");
        } catch (UnsupportedEncodingException uee) {
            throw new IllegalStateException("内部错误：本计算机的Java环境不支持GBK编码: "
                    + uee.getMessage());
        }
        if (b.length == 2) {
            int ch0 = b[0]; // 高位字节
            int ch1 = b[1]; // 低位字节
            if (ch0 < 0) {
                ch0 += 256;
            }
            if (ch1 < 0) {
                ch1 += 256;
            }
            if (ch0 >= GBK_HIGH_MIN && ch0 <= GBK_HIGH_MAX
                    && ch1 >= GBK_LOW_MIN && ch1 <= GBK_LOW_MAX
                    && ch1 != GBK_LOW_EXCEPTION) {
                return true;
            }
        }

        return false;
    }
}