unicode-icu.c

   1 #include <unicode/ustring.h>
   2 #include <unicode/uchar.h>
   3
   4 static inline UChar *utf8_to_utf16(const char *str, int32_t *length)
   5 {
   6         UChar *ustr;
   7         UErrorCode status = U_ZERO_ERROR;
   8
   9         u_strFromUTF8(NULL, 0, length, str, -1, &status);
  10         status = U_ZERO_ERROR;
  11         (*length)++; /* for the NUL char */
  12         ustr = malloc(sizeof(UChar) * (*length));
  13         if (!ustr)
  14                 return NULL;
  15         u_strFromUTF8(ustr, *length, NULL, str, -1, &status);
  16         if (U_FAILURE(status)) {
  17                 free(ustr);
  18                 return NULL;
  19         }
  20         return ustr;
  21 }
  22
  23 static inline char *utf16_to_utf8(UChar *ustr, int32_t *length)
  24 {
  25         char *str;
  26         UErrorCode status = U_ZERO_ERROR;
  27
  28         u_strToUTF8(NULL, 0, length, ustr, -1, &status);
  29         status = U_ZERO_ERROR;
  30         (*length)++; /* for the NUL char */
  31         str = malloc(*length);
  32         if (!str)
  33                 return NULL;
  34         u_strToUTF8(str, *length, NULL, ustr, -1, &status);
  35         if (U_FAILURE(status)) {
  36                 free(str);
  37                 return NULL;
  38         }
  39         return str;
  40 }
  41
  42 static inline bool str_contains_upper(const char *s)
  43 {
  44         bool ret = false;
  45         int32_t length, i;
  46         UChar32 c;
  47         UChar *ustr = utf8_to_utf16(s, &length);
  48         if (!ustr)
  49                 return true;
  50         for (i = 0; i < length; /* U16_NEXT post-increments */) {
  51                 U16_NEXT(ustr, i, length, c);
  52                 if (u_isupper(c)) {
  53                         ret = true;
  54                         goto out;
  55                 }
  56         }
  57 out:
  58         free(ustr);
  59         return ret;
  60 }
  61
  62 static inline char *str_fold(const char *s)
  63 {
  64         int32_t length;
  65         char *str;
  66         UChar *ustr;
  67         UErrorCode status = U_ZERO_ERROR;
  68
  69         ustr = utf8_to_utf16(s, &length);
  70         if (!ustr)
  71                 return NULL;
  72         u_strFoldCase(ustr, length, ustr, length, U_FOLD_CASE_EXCLUDE_SPECIAL_I, &status);
  73         if (U_FAILURE(status))
  74                 return NULL;
  75         str = utf16_to_utf8(ustr, &length);
  76         free(ustr);
  77         return str;
  78 }