lib/utf8.c

   1 #include <stdlib.h>
   2 #include <stdio.h>
   3 #include <memory.h>
   4 #include "utf8.h"
   5
   6 static char utf8buf[16];
   7
   8 void writeUTF8(unsigned int charnum, char*dest)
   9 {
  10     dest[0] = 0;
  11     if(charnum < 0x80) {
  12         dest[0] = charnum;
  13         dest[1] = 0;
  14     } else if(charnum <0x800) {
  15         /* 0000 0080-0000 07FF   110xxxxx 10xxxxxx */
  16         dest[0] = 0xc0 | (charnum >> 6);
  17         dest[1] = 0x80 | (charnum & 0x3f);
  18         dest[2] = 0;
  19     } else if(charnum < 0x10000) {
  20         /* 0000 0800-0000 FFFF   1110xxxx 10xxxxxx 10xxxxxx */
  21         dest[0] = 0xe0 | (charnum >> 12);
  22         dest[1] = 0x80 |((charnum >> 6)&0x3f);
  23         dest[2] = 0x80 |((charnum     )&0x3f);
  24         dest[3] = 0;
  25     } else if(charnum < 0x200000) {
  26         /* 0001 0000-001F FFFF   11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
  27         dest[0] = 0xf0 | (charnum >> 18);
  28         dest[1] = 0x80 |((charnum >> 12)&0x3f);
  29         dest[2] = 0x80 |((charnum >> 6 )&0x3f);
  30         dest[3] = 0x80 |((charnum      )&0x3f);
  31         dest[4] = 0;
  32     } else if(charnum < 0x4000000) {
  33         /* 0020 0000-03FF FFFF   111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
  34         dest[0] = 0xf8 | (charnum >> 24);
  35         dest[1] = 0x80 |((charnum >> 18)&0x3f);
  36         dest[2] = 0x80 |((charnum >> 12)&0x3f);
  37         dest[3] = 0x80 |((charnum >> 6 )&0x3f);
  38         dest[4] = 0x80 |((charnum      )&0x3f);
  39         dest[5] = 0;
  40     } else if(charnum < 0x80000000) {
  41         /* 0400 0000-7FFF FFFF   1111110x 10xxxxxx ... 10xxxxxx */
  42         dest[0] = 0xfc | (charnum >> 30);
  43         dest[1] = 0x80 |((charnum >> 24)&0x3f);
  44         dest[2] = 0x80 |((charnum >> 18)&0x3f);
  45         dest[3] = 0x80 |((charnum >> 12)&0x3f);
  46         dest[4] = 0x80 |((charnum >> 6 )&0x3f);
  47         dest[5] = 0x80 |((charnum      )&0x3f);
  48         dest[6] = 0;
  49     } else {
  50         fprintf(stderr, "Illegal character: 0x%08x\n", charnum);
  51         dest[0] = 0;
  52     }
  53 }
  54
  55 char* getUTF8(unsigned int charnum)
  56 {
  57     memset(utf8buf, 0, sizeof(utf8buf));
  58     writeUTF8(charnum, utf8buf);
  59     return utf8buf;
  60 }
  61