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 int writeUTF8(unsigned int charnum, char*dest)
   9 {
  10     dest[0] = 0;
  11     if(charnum < 0x80) {
  12         dest[0] = charnum;
  13         dest[1] = 0;
  14         return 1;
  15     } else if(charnum <0x800) {
  16         /* 0000 0080-0000 07FF   110xxxxx 10xxxxxx */
  17         dest[0] = 0xc0 | (charnum >> 6);
  18         dest[1] = 0x80 | (charnum & 0x3f);
  19         dest[2] = 0;
  20         return 2;
  21     } else if(charnum < 0x10000) {
  22         /* 0000 0800-0000 FFFF   1110xxxx 10xxxxxx 10xxxxxx */
  23         dest[0] = 0xe0 | (charnum >> 12);
  24         dest[1] = 0x80 |((charnum >> 6)&0x3f);
  25         dest[2] = 0x80 |((charnum     )&0x3f);
  26         dest[3] = 0;
  27         return 3;
  28     } else if(charnum < 0x200000) {
  29         /* 0001 0000-001F FFFF   11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
  30         dest[0] = 0xf0 | (charnum >> 18);
  31         dest[1] = 0x80 |((charnum >> 12)&0x3f);
  32         dest[2] = 0x80 |((charnum >> 6 )&0x3f);
  33         dest[3] = 0x80 |((charnum      )&0x3f);
  34         dest[4] = 0;
  35         return 4;
  36     } else if(charnum < 0x4000000) {
  37         /* 0020 0000-03FF FFFF   111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
  38         dest[0] = 0xf8 | (charnum >> 24);
  39         dest[1] = 0x80 |((charnum >> 18)&0x3f);
  40         dest[2] = 0x80 |((charnum >> 12)&0x3f);
  41         dest[3] = 0x80 |((charnum >> 6 )&0x3f);
  42         dest[4] = 0x80 |((charnum      )&0x3f);
  43         dest[5] = 0;
  44         return 5;
  45     } else if(charnum < 0x80000000) {
  46         /* 0400 0000-7FFF FFFF   1111110x 10xxxxxx ... 10xxxxxx */
  47         dest[0] = 0xfc | (charnum >> 30);
  48         dest[1] = 0x80 |((charnum >> 24)&0x3f);
  49         dest[2] = 0x80 |((charnum >> 18)&0x3f);
  50         dest[3] = 0x80 |((charnum >> 12)&0x3f);
  51         dest[4] = 0x80 |((charnum >> 6 )&0x3f);
  52         dest[5] = 0x80 |((charnum      )&0x3f);
  53         dest[6] = 0;
  54         return 6;
  55     } else {
  56         fprintf(stderr, "Illegal character: 0x%08x\n", charnum);
  57         dest[0] = 0;
  58         return 0;
  59     }
  60 }
  61
  62 char* getUTF8(unsigned int charnum)
  63 {
  64     memset(utf8buf, 0, sizeof(utf8buf));
  65     writeUTF8(charnum, utf8buf);
  66     return utf8buf;
  67 }
  68