char *String2Utf8(Object *string) {
int len = getStringLen(string);
unsigned short *unicode = getStringChars(string);
char *utf8 = (char*)sysMalloc(utf8CharLen(unicode, len) + 1);
return unicode2Utf8(unicode, len, utf8);
}
/*
* call-seq: getch
*
* Works like StringScanner#getch but is UTF8-aware
*/
static VALUE rb_cStringScanner_UTF8_getch(VALUE self) {
unsigned char *str;
long len = 0, pos = 0;
VALUE utf8Str, curStr;
int8_t lastCharLen=0;
#ifndef RUBINIUS
struct strscanner *scanner;
GET_SCANNER(self, scanner);
curStr = scanner->str;
pos = scanner->curr;
#else
curStr = rb_iv_get(self, "@string");
pos = FIX2LONG(rb_iv_get(self, "@pos"));
#endif
str = (unsigned char *)RSTRING_PTR(curStr);
len = RSTRING_LEN(curStr);
if (len > 0 && len > pos) {
lastCharLen = utf8CharLen(str, len);
if (lastCharLen < 0) {
rb_raise(rb_eArgError, "invalid utf-8 byte sequence");
}
utf8Str = rb_str_new((char *)str+pos, lastCharLen);
pos += lastCharLen;
#ifndef RUBINIUS
scanner->curr = pos;
#else
rb_iv_set(self, "@pos", LONG2FIX(pos));
#endif
AS_UTF8(utf8Str);
return utf8Str;
} else {
return Qnil;
}
}
int getStringUtf8Len(Object *string) {
return utf8CharLen(getStringChars(string), getStringLen(string));
}