unicode/utf8_chset.c

   1 /*
   2 ** Copyright 2000 Double Precision, Inc.
   3 ** See COPYING for distribution information.
   4 **
   5 ** $Id: utf8_chset.c,v 1.5 2004/05/23 14:28:25 mrsam Exp $
   6 */
   7
   8 #include "unicode.h"
   9 #include <stdio.h>
  10 #include <stdlib.h>
  11 #include <string.h>
  12
  13 /*
  14 ** UTF8.toupper/tolower/totitle is implemented by converting UTF8 to
  15 ** UCS-4, applying the unicode table lookup, then converting it back to
  16 ** UTF8
  17 */
  18
  19 static char *toupper_func(const struct unicode_info *u,
  20                           const char *cp, int *ip)
  21 {
  22         unicode_char *uc=unicode_utf8_tou(cp, ip), *p;
  23         char *s;
  24
  25         if (!uc) return (0);
  26
  27         for (p=uc; *p; p++)
  28                 *p=unicode_uc(*p);
  29
  30         s=unicode_utf8_fromu(uc, NULL);
  31         free(uc);
  32         return (s);
  33 }
  34
  35 static char *tolower_func(const struct unicode_info *u,
  36                           const char *cp, int *ip)
  37 {
  38         unicode_char *uc=unicode_utf8_tou(cp, ip), *p;
  39         char *s;
  40
  41         if (!uc) return (0);
  42
  43         for (p=uc; *p; p++)
  44                 *p=unicode_lc(*p);
  45
  46         s=unicode_utf8_fromu(uc, NULL);
  47         free(uc);
  48         return (s);
  49 }
  50
  51 static char *totitle_func(const struct unicode_info *u,
  52                           const char *cp, int *ip)
  53 {
  54         unicode_char *uc=unicode_utf8_tou(cp, ip), *p;
  55         char *s;
  56
  57         if (!uc) return (0);
  58
  59         for (p=uc; *p; p++)
  60                 *p=unicode_tc(*p);
  61
  62         s=unicode_utf8_fromu(uc, NULL);
  63         free(uc);
  64         return (s);
  65 }
  66
  67 static unicode_char *tou(const struct unicode_info *i, const char *p,
  68                          int *err)
  69 {
  70         return unicode_utf8_tou(p, err);
  71 }
  72
  73 static char *fromu(const struct unicode_info *i, const unicode_char *p,
  74                    int *err)
  75 {
  76         return unicode_utf8_fromu(p, err);
  77 }
  78
  79 const struct unicode_info unicode_UTF8 = {
  80         "UTF-8",
  81         UNICODE_UTF | UNICODE_MB | UNICODE_USASCII | UNICODE_HEADER_QUOPRI
  82         | UNICODE_BODY_QUOPRI,
  83         tou,
  84         fromu,
  85         toupper_func,
  86         tolower_func,
  87         totitle_func};
  88