static int to_ascii(OnigEncoding enc, UChar *s, UChar *end,
UChar buf[], int buf_size, int *is_over)
{
int len;
UChar *p;
OnigCodePoint code;
if (ONIGENC_MBC_MINLEN(enc) > 1) {
p = s;
len = 0;
while (p < end) {
code = ONIGENC_MBC_TO_CODE(enc, p, end);
if (code >= 0x80) {
if (code > 0xffff && len + 10 <= buf_size) {
sprint_byte_with_x((char*)(&(buf[len])), (unsigned int)(code >> 24));
sprint_byte((char*)(&(buf[len+4])), (unsigned int)(code >> 16));
sprint_byte((char*)(&(buf[len+6])), (unsigned int)(code >> 8));
sprint_byte((char*)(&(buf[len+8])), (unsigned int)code);
len += 10;
}
else if (len + 6 <= buf_size) {
sprint_byte_with_x((char*)(&(buf[len])), (unsigned int)(code >> 8));
sprint_byte((char*)(&(buf[len+4])), (unsigned int)code);
len += 6;
}
extern int
onigenc_str_bytelen_null(OnigEncoding enc, const UChar* s)
{
UChar* start = (UChar* )s;
UChar* p = (UChar* )s;
UChar* e;
while (1) {
if (*p == '\0') {
UChar* q;
int len = ONIGENC_MBC_MINLEN(enc);
if (len == 1) return (int )(p - start);
q = p + 1;
while (len > 1) {
if (*q != '\0') break;
q++;
len--;
}
if (len == 1) return (int )(p - start);
}
e = p + ONIGENC_MBC_MAXLEN(enc);
p += ONIGENC_MBC_ENC_LEN(enc, p, e);
}
}
native_int Encoding::string_character_length(const uint8_t* p, const uint8_t* e, OnigEncodingType* enc) {
native_int chars;
for(chars = 0; p < e; chars++) {
int n = Encoding::precise_mbclen(p, e, enc);
if(ONIGENC_MBCLEN_CHARFOUND_P(n)) {
p += ONIGENC_MBCLEN_CHARFOUND_LEN(n);
} else if(p + ONIGENC_MBC_MINLEN(enc) <= e) {
p += ONIGENC_MBC_MINLEN(enc);
} else {
p = e;
}
}
return chars;
}
int Encoding::mbclen(const uint8_t* p, const uint8_t* e, OnigEncodingType* enc) {
int n = ONIGENC_PRECISE_MBC_ENC_LEN(enc, (UChar*)p, (UChar*)e);
if (ONIGENC_MBCLEN_CHARFOUND_P(n) && ONIGENC_MBCLEN_CHARFOUND_LEN(n) <= e-p) {
return ONIGENC_MBCLEN_CHARFOUND_LEN(n);
} else {
int min = ONIGENC_MBC_MINLEN(enc);
return min <= e-p ? min : (int)(e-p);
}
}
extern int
onigenc_strlen_null(OnigEncoding enc, const UChar* s)
{
int n = 0;
UChar* p = (UChar* )s;
while (1) {
if (*p == '\0') {
UChar* q;
int len = ONIGENC_MBC_MINLEN(enc);
if (len == 1) return n;
q = p + 1;
while (len > 1) {
if (*q != '\0') break;
q++;
len--;
}
if (len == 1) return n;
}
p += ONIGENC_MBC_ENC_LEN(enc, p);
n++;
}
}