chiark - git - mdw - disorder/blob - lib/utf8.h

   1 /*
   2  * This file is part of DisOrder
   3  * Copyright (C) 2004, 2005 Richard Kettlewell
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful, but
  11  * WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
  18  * USA
  19  */
  20 #ifndef UTF8_H
  21 #define UTF8_H
  22
  23 #define PARSE_UTF8(S,C,E) do {                  \
  24   if((unsigned char)*S < 0x80)                  \
  25     C = *S++;                                   \
  26   else if((unsigned char)*S <= 0xDF) {          \
  27     C = (*S++ & 0x1F) << 6;                     \
  28     if((*S & 0xC0) != 0x80) { E; }              \
  29     C |= (*S++ & 0x3F);                         \
  30     if(C < 0x80) { E; }                         \
  31   } else if((unsigned char)*S <= 0xEF) {        \
  32     C = (*S++ & 0x0F) << 12;                    \
  33     if((*S & 0xC0) != 0x80) { E; }              \
  34     C |= (*S++ & 0x3F) << 6;                    \
  35     if((*S & 0xC0) != 0x80) { E; }              \
  36     C |= (*S++ & 0x3F);                         \
  37     if(C < 0x800                                \
  38        || (C >= 0xD800 && C <= 0xDFFF)) {       \
  39       E;                                        \
  40     }                                           \
  41   } else if((unsigned char)*S <= 0xF7) {        \
  42     C = (*S++ & 0x07) << 18;                    \
  43     if((*S & 0xC0) != 0x80) { E; }              \
  44     C |= (*S++ & 0x3F) << 12;                   \
  45     if((*S & 0xC0) != 0x80) { E; }              \
  46     C |= (*S++ & 0x3F) << 6;                    \
  47     if((*S & 0xC0) != 0x80) { E; }              \
  48     C |= (*S++ & 0x3F);                         \
  49     if(C < 0x10000 || C > 0x10FFFF) { E; }      \
  50   } else {                                      \
  51     E;                                          \
  52   }                                             \
  53 } while(0)
  54
  55 int validutf8(const char *s);
  56 /* return nonzero if S is a valid UTF-8 sequence, else false */
  57
  58 #endif /* UTF8_h */
  59
  60 /*
  61 Local Variables:
  62 c-basic-offset:2
  63 comment-column:40
  64 End:
  65 */