libutf.c

   1 /* libutf8 © 2012-2015 Connor Lane Smith <cls@lubutu.com> */
   2 #include "libutf.h"
   3 #include "util.h"
   4
   5 int
   6 runelen(Rune r)
   7 {
   8         if(r <= 0x7F)
   9                 return 1;
  10         else if(r <= 0x07FF)
  11                 return 2;
  12         else if(r <= 0xD7FF)
  13                 return 3;
  14         else if(r <= 0xDFFF)
  15                 return 0; /* surrogate character */
  16         else if(r <= 0xFFFD)
  17                 return 3;
  18         else if(r <= 0xFFFF)
  19                 return 0; /* illegal character */
  20         else if(r <= Runemax)
  21                 return 4;
  22         else
  23                 return 0; /* rune too large */
  24 }
  25
  26 int
  27 runetochar(char *s, const Rune *p)
  28 {
  29         Rune r = *p;
  30
  31         switch(runelen(r)) {
  32         case 1: /* 0aaaaaaa */
  33                 s[0] = r;
  34                 return 1;
  35         case 2: /* 00000aaa aabbbbbb */
  36                 s[0] = 0xC0 | ((r & 0x0007C0) >>  6); /* 110aaaaa */
  37                 s[1] = 0x80 |  (r & 0x00003F);        /* 10bbbbbb */
  38                 return 2;
  39         case 3: /* aaaabbbb bbcccccc */
  40                 s[0] = 0xE0 | ((r & 0x00F000) >> 12); /* 1110aaaa */
  41                 s[1] = 0x80 | ((r & 0x000FC0) >>  6); /* 10bbbbbb */
  42                 s[2] = 0x80 |  (r & 0x00003F);        /* 10cccccc */
  43                 return 3;
  44         case 4: /* 000aaabb bbbbcccc ccdddddd */
  45                 s[0] = 0xF0 | ((r & 0x1C0000) >> 18); /* 11110aaa */
  46                 s[1] = 0x80 | ((r & 0x03F000) >> 12); /* 10bbbbbb */
  47                 s[2] = 0x80 | ((r & 0x000FC0) >>  6); /* 10cccccc */
  48                 s[3] = 0x80 |  (r & 0x00003F);        /* 10dddddd */
  49                 return 4;
  50         default:
  51                 return 0; /* error */
  52         }
  53 }