static inline const char* utf8_next(const char *str)
{
int n = utf8_char_len(str);
while (n > 0) {
str++;
if (*str == '\0')
return str;
n--;
}
return str;
}
t_size strstr_ex(const char * p_string,t_size p_string_len,const char * p_substring,t_size p_substring_len) throw()
{
p_string_len = strlen_max(p_string,p_string_len);
p_substring_len = strlen_max(p_substring,p_substring_len);
t_size index = 0;
while(index + p_substring_len <= p_string_len)
{
if (memcmp(p_string+index,p_substring,p_substring_len) == 0) return index;
t_size delta = utf8_char_len(p_string+index,p_string_len - index);
if (delta == 0) break;
index += delta;
}
return ~0;
}
// utf8 to LENDIAN unicode
static status_t
_utf8_to_lendian_unicode(
const char *src,
size_t *srcLen,
char *dst,
size_t *dstLen)
{
size_t srcLimit = *srcLen;
size_t dstLimit = *dstLen - 1;
size_t srcCount = 0;
size_t dstCount = 0;
status_t status = B_ERROR;
while ((srcCount < srcLimit) && (dstCount < dstLimit)) {
uint16 unicode;
uint16 *UNICODE = &unicode;
uchar *UTF8 = (uchar *)src + srcCount;
int err_flag;
if ((srcCount + utf8_char_len(src[srcCount])) > srcLimit)
break;
utf8_to_u_hostendian(UTF8, UNICODE, err_flag);
if (err_flag == 1)
return EINVAL;
unicode = B_HOST_TO_LENDIAN_INT16(unicode);
if ((dstCount + 1) > dstLimit) {
status = B_BUFFER_OVERFLOW;
break;
}
dst[dstCount++] = unicode & 0xFF;
dst[dstCount++] = unicode >> 8;
srcCount += UTF8 - ((uchar *)(src + srcCount));
status = B_OK;
}
*srcLen = srcCount;
*dstLen = dstCount;
return status;
}
int
main(void) {
init();
void * input = new(File, "./utf8-example");
void * content = File_read(input);
uint8_t * string = Object_inspect(content);
printf("[%s]\n", string);
uint8_t buffer[30];
buffer[0] = '\0';
size_t len = strlen(string);
size_t n;
for(int i = 0; i < len; i += n) {
n = utf8_char_len(string + i);
strncpy(buffer, string + i, n);
buffer[n] = '\0';
printf("[%s]\n", buffer);
}
return 0;
}
// utf8 to LENDIAN unicode
static status_t
_utf8_to_lendian_unicode(
const char *src,
int32 *srcLen,
char *dst,
uint32 *dstLen)
{
int32 srcLimit = *srcLen;
int32 dstLimit = *dstLen - 1;
int32 srcCount = 0;
int32 dstCount = 0;
while ((srcCount < srcLimit) && (dstCount < dstLimit)) {
uint16 unicode;
uint16 *UNICODE = &unicode;
uchar *UTF8 = (uchar *)src + srcCount;
int err_flag;
if ((srcCount + utf8_char_len(src[srcCount])) > srcLimit)
break;
utf8_to_u_hostendian(UTF8, UNICODE, err_flag);
if(err_flag == 1)
return EINVAL;
unicode = B_HOST_TO_LENDIAN_INT16(unicode);
dst[dstCount++] = unicode & 0xFF;
dst[dstCount++] = unicode >> 8;
srcCount += UTF8 - ((uchar *)(src + srcCount));
}
*srcLen = srcCount;
*dstLen = dstCount;
return ((dstCount > 0) ? B_NO_ERROR : B_ERROR);
}
utf8string::iterator::self& utf8string::iterator::operator++() {
base_it += utf8_char_len(*base_it);
return *this;
}
