static bool
is_special_global_name(UChar *ptr, long len)
{
if (len <= 0) {
return false;
}
long pos = 0;
switch (ptr[pos]) {
case '~': case '*': case '$': case '?': case '!':
case '@': case '/': case '\\': case ';': case ',':
case '.': case '=': case ':': case '<': case '>':
case '\"': case '&': case '`': case '\'': case '+': case '0':
pos++;
break;
case '-':
pos++;
if (pos < len && is_identchar(ptr[pos])) {
pos++;
}
break;
default:
if (!isdigit(ptr[pos])) {
return false;
}
do {
pos++;
}
while (pos < len && isdigit(ptr[pos]));
break;
}
return pos == len;
}
static mrb_bool
is_special_global_name(const char* m)
{
switch (*m) {
case '~': case '*': case '$': case '?': case '!': case '@':
case '/': case '\\': case ';': case ',': case '.': case '=':
case ':': case '<': case '>': case '\"':
case '&': case '`': case '\'': case '+':
case '0':
++m;
break;
case '-':
++m;
if (is_identchar(*m)) m += 1;
break;
default:
if (!ISDIGIT(*m)) return FALSE;
do ++m; while (ISDIGIT(*m));
break;
}
return !*m;
}
static mrb_bool
symname_p(const char *name)
{
const char *m = name;
mrb_bool localid = FALSE;
if (!m) return FALSE;
switch (*m) {
case '\0':
return FALSE;
case '$':
if (is_special_global_name(++m)) return TRUE;
goto id;
case '@':
if (*++m == '@') ++m;
goto id;
case '<':
switch (*++m) {
case '<': ++m; break;
case '=': if (*++m == '>') ++m; break;
default: break;
}
break;
case '>':
switch (*++m) {
case '>': case '=': ++m; break;
default: break;
}
break;
case '=':
switch (*++m) {
case '~': ++m; break;
case '=': if (*++m == '=') ++m; break;
default: return FALSE;
}
break;
case '*':
if (*++m == '*') ++m;
break;
case '!':
if (*++m == '=') ++m;
break;
case '+': case '-':
if (*++m == '@') ++m;
break;
case '|':
if (*++m == '|') ++m;
break;
case '&':
if (*++m == '&') ++m;
break;
case '^': case '/': case '%': case '~': case '`':
++m;
break;
case '[':
if (*++m != ']') return FALSE;
if (*++m == '=') ++m;
break;
default:
localid = !ISUPPER(*m);
id:
if (*m != '_' && !ISALPHA(*m)) return FALSE;
while (is_identchar(*m)) m += 1;
if (localid) {
switch (*m) {
case '!': case '?': case '=': ++m;
default: break;
}
}
break;
}
return *m ? FALSE : TRUE;
}
static ID
rb_intern_uchars(const UChar *chars, const size_t chars_len, VALUE str)
{
const unsigned long name_hash = rb_str_hash_uchars(chars, chars_len);
LOCK();
ID id = (ID)CFDictionaryGetValue(sym_id, (const void *)name_hash);
UNLOCK();
if (id != 0) {
goto return_id;
}
if (str == Qnil) {
str = rb_unicode_str_new(chars, chars_len);
}
rb_sym_t *sym = NULL;
long pos = 0;
if (chars_len > 0) {
UChar c = chars[0];
switch (c) {
case '$':
id = ID_GLOBAL;
goto new_id;
case '@':
if (chars_len > 1 && chars[1] == '@') {
pos++;
id = ID_CLASS;
}
else {
id = ID_INSTANCE;
}
pos++;
break;
default:
if (chars_len > 1 && chars[chars_len - 1] == '=') {
// Attribute assignment.
id = rb_intern_str(rb_str_substr(str, 0, chars_len - 1));
if (!is_attrset_id(id)) {
id = rb_id_attrset(id);
goto id_register;
}
id = ID_ATTRSET;
}
else if (iswupper(c)) {
id = ID_CONST;
}
else {
id = ID_LOCAL;
}
break;
}
}
if (pos < chars_len && !isdigit(chars[pos])) {
for (; pos < chars_len; pos++) {
if (!is_identchar(chars[pos])) {
break;
}
}
}
if (pos < chars_len) {
id = ID_JUNK;
}
new_id:
id |= ++last_id << ID_SCOPE_SHIFT;
id_register:
//printf("register %s hash %ld id %ld\n", RSTRING_PTR(str), name_hash, id);
sym = sym_alloc(str, id);
LOCK();
CFDictionarySetValue(sym_id, (const void *)name_hash, (const void *)id);
CFDictionarySetValue(id_str, (const void *)id, (const void *)sym);
UNLOCK();
return_id:
return id;
}
static bool
sym_should_be_escaped(VALUE sym)
{
RB_STR_GET_UCHARS(RSYM(sym)->str, chars, chars_len);
if (chars_len == 0) {
return true;
}
bool escape = false;
for (long i = 0; i < chars_len; i++) {
if (!isprint(chars[i])) {
escape = true;
break;
}
}
if (escape) {
goto bail;
}
long pos = 0;
bool localid = false;
switch (chars[pos]) {
case '\0':
escape = true;
break;
case '$':
pos++;
if (pos < chars_len && is_special_global_name(&chars[pos],
chars_len - pos)) {
goto bail;
}
goto id;
case '@':
pos++;
if (pos < chars_len && chars[pos] == '@') {
pos++;
}
goto id;
case '<':
pos++;
if (pos < chars_len) {
if (chars[pos] == '<') {
pos++;
}
else if (chars[pos] == '=') {
pos++;
if (pos < chars_len && chars[pos] == '>') {
pos++;
}
}
}
break;
case '>':
pos++;
if (pos < chars_len) {
if (chars[pos] == '>' || chars[pos] == '=') {
pos++;
}
}
break;
case '=':
pos++;
if (pos == chars_len) {
escape = true;
goto bail;
}
else {
if (chars[pos] == '~') {
pos++;
}
else if (chars[pos] == '=') {
pos++;
if (pos < chars_len && chars[pos] == '=') {
pos++;
}
}
else {
escape = true;
goto bail;
}
}
break;
case '*':
pos++;
if (pos < chars_len && chars[pos] == '*') {
pos++;
}
break;
case '+':
case '-':
pos++;
if (pos < chars_len && chars[pos] == '@') {
pos++;
}
break;
case '|': case '^': case '&': case '/':
case '%': case '~': case '`':
pos++;
break;
case '[':
pos++;
if (pos < chars_len && chars[pos] != ']') {
escape = true;
goto bail;
}
pos++;
if (pos < chars_len && chars[pos] == '=') {
pos++;
}
break;
case '!':
pos++;
if (pos == chars_len) {
goto bail;
}
else {
if (chars[pos] == '=' || chars[pos] == '~') {
pos++;
}
else {
escape = true;
goto bail;
}
}
break;
default:
localid = !isupper(chars[pos]);
// fall through
id:
if (pos >= chars_len
|| (chars[pos] != '_' && !isalpha(chars[pos])
&& isascii(chars[pos]))) {
escape = true;
goto bail;
}
while (pos < chars_len && is_identchar(chars[pos])) {
pos++;
}
if (localid) {
if (pos < chars_len
&& (chars[pos] == '!' || chars[pos] == '?'
|| chars[pos] == '=')) {
pos++;
}
}
break;
}
if (pos < chars_len) {
escape = true;
}
bail:
return escape;
}