utf8 转 unicode(c代码)

debug.h

#ifndef __DEBUG_H__
#define __DEBUG_H__

#define DEBUG 1

#if DEBUG
#include <stdio.h>
#include <stdarg.h>

#define DebugWhere() printf("FILE:%s,LINE:%d,FUNC:%s\n",\
        __FILE__,__LINE__,__FUNCTION__)
static inline void debug_p(const char *format,...)
{
    va_list args;

    va_start(args,format);
    vprintf(format,args);
    va_end(args);
}
#else
#define DebugWhere()

static inline void debug_p(const char *format,...)
{
}

#endif

#endif

main.c

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include "debug.h"
/*
 *  utf8编码转换为unicode编码
 *  输入:源字符串,目的字符串,数据长度
 *
 */

static int my_utf8_to_unicode(unsigned char* unicode, unsigned char* utf8, int len)
{
    int length=0;
    unsigned int code_int;
    int i;

    debug_p("len = %d\n",len);
    for(i = 0; i < len ; i++) {
        if(utf8[i] <= 0x7f) {
            unicode[length] = 0x00;
            length++;
            unicode[length] = utf8[i];
            length++;
        }
        else if(utf8[i] <= 0xdf) {
            code_int = (((unsigned char) utf8[i] & 0x1f)  << 6) + ((unsigned char) utf8[i+1] & 0x3f) ;
            unicode[length] = (unsigned char) (code_int >> 8);
            i +=1;
            unicode[length+1] = (unsigned char) code_int;
            unicode[length] = (unsigned char) (code_int >> 8);
            length += 2;
        } else {
            code_int = ((int) ((unsigned char) utf8[i] & 0x0f) << 12)+(((unsigned char) utf8[i+1] & 0x3f) << 6)+
                ((unsigned char) utf8[i+2] & 0x3f);
            i +=2;
            unicode[length+1] = (unsigned char) code_int;
            unicode[length] = (unsigned char) (code_int >> 8);
            length += 2;
        }
    }
    debug_p("length = %d\n",length);

    return (length);
}

int main()
{
    unsigned char unicode[128];
    unsigned char *utf8_11 = "工作愉快";
    unsigned char *utf8_1 = "工";
    unsigned char *utf8_2 = "作";
    unsigned char *utf8_3 = "愉";
    unsigned char *utf8_4 = "快";
    unsigned char buf[3] = {0};
    unsigned char buff[128] = {0};

    int i,len = 0;

    memset(unicode,0,sizeof(unicode));
    len = strlen(utf8_11);
    len = my_utf8_to_unicode(unicode,utf8_11,len);
    for(i = 0 ; i < len ; i++){
        printf("unicode[%d] = %02x\n",i,unicode[i]);
        sprintf(buf,"%02x",unicode[i]);
        strcat(buff,buf);
    }
    printf("hello: %s\n",buff);

    len = my_utf8_to_unicode(unicode,utf8_1,2);
    for(i = 0 ; i < len ; i++){
        printf("unicode[%d] = %02x\n",i,(unsigned char)unicode[i]);
    }

    len = my_utf8_to_unicode(unicode,utf8_4,2);
    for(i = 0 ; i < len ; i++){
        printf("unicode[%d] = %02x\n",i,(unsigned char)unicode[i]);
    }

}