c++实现UTF-16转UCS4

1.UTF-16编码规则

在UTF-16中，表示数据有单16位和双16位（32位）两种，1101 10xx xxxx xxxx，区间就是D800~DBFF为前16位，1101 11xx xxxx xxxx，区间就是DC00~DFFF为后16位，其余为单16位。

2.源码实现

#include <iostream>
#include <cstring>

using namespace std;

class UTF16 {
public:
    static int toUCS4(const unsigned short *utf16, unsigned short *ucs4);
};

int UTF16::toUCS4(const unsigned short *utf16, unsigned short *ucs4)
{
    if(utf16[0] >= 0xd800 && utf16[0] <= 0xdfff)
    {
        if(utf16[0] < 0xdc00)
        {
            if(utf16[1] >= 0xdc00 && utf16[1] <= 0xdfff)
            {
                ucs4[1] = (utf16[0] & 0x3ff);
                ucs4[0] = (utf16[1] & 0x3ff);
                ucs4[0] = ((ucs4[1] << 10) | ucs4[0]);
                ucs4[1] = ((ucs4[1] >> 6) | 1);

                //printf("%04x\n", ucs4[0]);
                //printf("%04x\n", ucs4[1]);

                return 2;
            }

            return -1;
        }

        return -1;
    }
    else
    {
        ucs4[0] = utf16[0];
        ucs4[1] = 0x00;
    }

    return 1;
}

int main()
{
    const unsigned short utf16[4] = {0xd802, 0xdc01, 0x00, 0x00};
    unsigned short ucs4[2];
    unsigned int *p = (unsigned int *)ucs4;

    UTF16::toUCS4(utf16, ucs4);

    printf("%08x\n", *p);

    return 0;
}

3.编译源码

$ g++ -o test test.cpp -std=c++11

4.运行及其结果

$ ./test
00010801