VALUE string_spec_rb_str_ptr_assign_call(VALUE self, VALUE str) {
char *ptr = rb_str_ptr(str);
ptr[1] = 'x';
rb_str_concat(str, rb_str_new2("d"));
return str;
}
VALUE string_spec_rb_str_ptr_assign_funcall(VALUE self, VALUE str) {
char *ptr = rb_str_ptr(str);
ptr[1] = 'x';
rb_funcall(str, rb_intern("<<"), 1, rb_str_new2("e"));
return str;
}
VALUE string_spec_rb_str_ptr_iterate(VALUE self, VALUE str) {
int i;
char* ptr;
ptr = rb_str_ptr(str);
for(i = 0; i < RSTRING_LEN(str); i++) {
rb_yield(INT2FIX(ptr[i]));
}
return Qnil;
}
VALUE string_spec_rb_str_ptr_assign(VALUE self, VALUE str, VALUE chr) {
int i;
char c;
char* ptr;
ptr = rb_str_ptr(str);
c = FIX2INT(chr);
for(i = 0; i < RSTRING_LEN(str); i++) {
ptr[i] = c;
}
return Qnil;
}
int
ox_sax_collapse_special(SaxDrive dr, char *str, int line, int col) {
char *s = str;
char *b = str;
while ('\0' != *s) {
if ('&' == *s) {
int c = 0;
char *end;
s++;
if ('#' == *s) {
uint64_t u = 0;
char x;
s++;
if ('x' == *s || 'X' == *s) {
x = *s;
s++;
end = read_hex_uint64(s, &u);
} else {
x = '\0';
end = read_10_uint64(s, &u);
}
if (0 == end) {
ox_sax_drive_error(dr, NO_TERM "special character does not end with a semicolon");
*b++ = '&';
*b++ = '#';
if ('\0' != x) {
*b++ = x;
}
continue;
}
if (u <= 0x000000000000007FULL) {
*b++ = (char)u;
#if HAS_ENCODING_SUPPORT
} else if (ox_utf8_encoding == dr->encoding) {
b = ox_ucs_to_utf8_chars(b, u);
} else if (0 == dr->encoding) {
dr->encoding = ox_utf8_encoding;
b = ox_ucs_to_utf8_chars(b, u);
#elif HAS_PRIVATE_ENCODING
} else if (ox_utf8_encoding == dr->encoding ||
0 == strcasecmp(rb_str_ptr(rb_String(ox_utf8_encoding)), rb_str_ptr(rb_String(dr->encoding)))) {
b = ox_ucs_to_utf8_chars(b, u);
} else if (Qnil == dr->encoding) {
dr->encoding = ox_utf8_encoding;
b = ox_ucs_to_utf8_chars(b, u);
#else
} else if (0 == dr->encoding) {
dr->encoding = UTF8_STR;
b = ox_ucs_to_utf8_chars(b, u);
} else if (0 == strcasecmp(UTF8_STR, dr->encoding)) {
b = ox_ucs_to_utf8_chars(b, u);
#endif
} else {
b = ox_ucs_to_utf8_chars(b, u);
/*
ox_sax_drive_error(dr, NO_TERM "Invalid encoding, need UTF-8 encoding to parse &#nnnn; character sequences.");
*b++ = '&';
*b++ = '#';
if ('\0' != x) {
*b++ = x;
}
continue;
*/
}
s = end + 1;
continue;
} else if (0 == strncasecmp(s, "lt;", 3)) {
c = '<';
s += 3;
col += 3;
} else if (0 == strncasecmp(s, "gt;", 3)) {
c = '>';
s += 3;
col += 3;
} else if (0 == strncasecmp(s, "amp;", 4)) {
c = '&';
s += 4;
col += 4;
} else if (0 == strncasecmp(s, "quot;", 5)) {
c = '"';
s += 5;
col += 5;
} else if (0 == strncasecmp(s, "apos;", 5)) {
c = '\'';
s += 5;
} else {
ox_sax_drive_error_at(dr, NO_TERM "special character does not end with a semicolon", line, col);
c = '&';
}
*b++ = (char)c;
col++;
} else {
if ('\n' == *s) {
line++;
col = 0;
}
col++;
*b++ = *s++;
}
}
*b = '\0';
return 0;
}
