refactor: 统一调用调度并安全化 IC 缓存

This commit is contained in:
2026-06-10 16:17:35 +08:00
parent af7c591414
commit 4aaa62812a
27 changed files with 1284 additions and 847 deletions
+143 -103
View File
@@ -31,11 +31,15 @@ fn boxed_primitive_value(this_val: JsValue) -> Option<JsValue> {
pub fn string_value_of(vm: &mut Vm, args: &[u8]) -> NativeResult {
let this_val = vm.reg(args[0]);
if this_val.is_string() {
return Ok(this_val);
return NativeResult::Ok(this_val);
}
match boxed_primitive_value(this_val) {
Some(v) => NativeResult::Ok(v),
None => NativeResult::Err(crate::builtins::error::create_type_error(
vm,
"String.prototype.valueOf called on non-String object",
)),
}
boxed_primitive_value(this_val).ok_or_else(|| {
crate::builtins::error::create_type_error(vm, "String.prototype.valueOf called on non-String object")
})
}
fn make_string_array(vm: &mut Vm, parts: &[String]) -> JsValue {
@@ -58,6 +62,27 @@ fn as_string(vm: &Vm, val: JsValue) -> String {
coercion::to_string(vm.kernel().string_forge().as_ref(), val)
}
fn char_len(s: &str) -> usize {
s.chars().count()
}
fn byte_index_at_char(s: &str, char_pos: usize) -> usize {
if char_pos == 0 {
return 0;
}
s.char_indices().nth(char_pos).map(|(idx, _)| idx).unwrap_or(s.len())
}
fn char_slice(s: &str, start: usize, end: usize) -> &str {
let start_byte = byte_index_at_char(s, start);
let end_byte = byte_index_at_char(s, end);
&s[start_byte..end_byte]
}
fn take_chars(s: &str, count: usize) -> String {
s.chars().take(count).collect()
}
fn is_regexp_obj(val: JsValue, vm: &Vm) -> bool {
if !val.is_object() {
return false;
@@ -81,52 +106,57 @@ fn is_regexp_obj(val: JsValue, vm: &Vm) -> bool {
pub fn string_index_of(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let n = char_len(&s);
if args.len() < 2 {
return Ok(JsValue::int(-1));
return NativeResult::Ok(JsValue::int(-1));
}
let search = as_string(vm, vm.reg(args[1]));
let pos = if args.len() > 2 {
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(s.len())
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(n)
} else {
0
};
if search.is_empty() {
return Ok(JsValue::int(pos as i32));
return NativeResult::Ok(JsValue::int(pos as i32));
}
let haystack = &s.as_bytes()[pos..];
let start_byte = byte_index_at_char(&s, pos);
let haystack = &s[start_byte..];
if search.len() == 1 {
if let Some(idx) = memchr(search.as_bytes()[0], haystack) {
return Ok(JsValue::int((pos + idx) as i32));
if let Some(idx) = memchr(search.as_bytes()[0], haystack.as_bytes()) {
let matched_byte = start_byte + idx;
return NativeResult::Ok(JsValue::int(char_len(&s[..matched_byte]) as i32));
}
} else if let Some(idx) = s[pos..].find(&search) {
return Ok(JsValue::int((pos + idx) as i32));
} else if let Some(idx) = haystack.find(&search) {
let matched_byte = start_byte + idx;
return NativeResult::Ok(JsValue::int(char_len(&s[..matched_byte]) as i32));
}
Ok(JsValue::int(-1))
NativeResult::Ok(JsValue::int(-1))
}
pub fn string_includes(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let n = char_len(&s);
if args.len() < 2 {
return Ok(JsValue::bool(false));
return NativeResult::Ok(JsValue::bool(false));
}
let search = as_string(vm, vm.reg(args[1]));
let pos = if args.len() > 2 {
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(s.len())
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(n)
} else {
0
};
if search.is_empty() {
return Ok(JsValue::bool(true));
return NativeResult::Ok(JsValue::bool(true));
}
let haystack = &s.as_bytes()[pos..];
let haystack = &s[byte_index_at_char(&s, pos)..];
if search.len() == 1 {
Ok(JsValue::bool(memchr(search.as_bytes()[0], haystack).is_some()))
NativeResult::Ok(JsValue::bool(memchr(search.as_bytes()[0], haystack.as_bytes()).is_some()))
} else {
Ok(JsValue::bool(s[pos..].contains(&search)))
NativeResult::Ok(JsValue::bool(haystack.contains(&search)))
}
}
@@ -134,31 +164,31 @@ pub fn string_char_at(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
if s.is_empty() {
return Ok(vm.intern(""));
return NativeResult::Ok(vm.intern(""));
}
return Ok(vm.intern(&s[0..1]));
return NativeResult::Ok(vm.intern(&take_chars(&s, 1)));
}
let idx = coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as i32;
if idx < 0 || idx as usize >= s.len() {
return Ok(vm.intern(""));
if idx < 0 || idx as usize >= char_len(&s) {
return NativeResult::Ok(vm.intern(""));
}
let ch = &s[(idx as usize)..(idx as usize + 1)];
Ok(vm.intern(ch))
let ch = s.chars().nth(idx as usize).map(|c| c.to_string()).unwrap_or_default();
NativeResult::Ok(vm.intern(&ch))
}
pub fn string_char_code_at(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
if s.is_empty() {
return Ok(JsValue::float(f64::NAN));
return NativeResult::Ok(JsValue::float(f64::NAN));
}
return Ok(JsValue::int(s.as_bytes()[0] as i32));
return NativeResult::Ok(JsValue::int(s.chars().next().unwrap() as i32));
}
let idx = coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as i32;
if idx < 0 || idx as usize >= s.len() {
return Ok(JsValue::float(f64::NAN));
if idx < 0 || idx as usize >= char_len(&s) {
return NativeResult::Ok(JsValue::float(f64::NAN));
}
Ok(JsValue::int(s.as_bytes()[idx as usize] as i32))
NativeResult::Ok(JsValue::int(s.chars().nth(idx as usize).unwrap() as i32))
}
pub fn string_concat(vm: &mut Vm, args: &[u8]) -> NativeResult {
@@ -167,12 +197,12 @@ pub fn string_concat(vm: &mut Vm, args: &[u8]) -> NativeResult {
for &arg_reg in args.iter().skip(1) {
result.push_str(&coercion::to_string(sf, vm.reg(arg_reg)));
}
Ok(vm.intern(&result))
NativeResult::Ok(vm.intern(&result))
}
pub fn string_slice(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let n = s.len() as i32;
let n = char_len(&s) as i32;
let start = if args.len() > 1 {
let v = coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as i32;
if v < 0 {
@@ -195,13 +225,13 @@ pub fn string_slice(vm: &mut Vm, args: &[u8]) -> NativeResult {
};
let start = start as usize;
let end = end as usize;
let result = if start < end { &s[start..end] } else { "" };
Ok(vm.intern(result))
let result = if start < end { char_slice(&s, start, end) } else { "" };
NativeResult::Ok(vm.intern(result))
}
pub fn string_substring(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let n = s.len() as i32;
let n = char_len(&s) as i32;
let mut start = if args.len() > 1 {
(coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as i32)
.max(0)
@@ -219,23 +249,23 @@ pub fn string_substring(vm: &mut Vm, args: &[u8]) -> NativeResult {
if start > end {
std::mem::swap(&mut start, &mut end);
}
let result = &s[start as usize..end as usize];
Ok(vm.intern(result))
let result = char_slice(&s, start as usize, end as usize);
NativeResult::Ok(vm.intern(result))
}
pub fn string_to_upper_case(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
Ok(vm.intern(&s.to_uppercase()))
NativeResult::Ok(vm.intern(&s.to_uppercase()))
}
pub fn string_to_lower_case(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
Ok(vm.intern(&s.to_lowercase()))
NativeResult::Ok(vm.intern(&s.to_lowercase()))
}
pub fn string_trim(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
Ok(vm.intern(s.trim()))
NativeResult::Ok(vm.intern(s.trim()))
}
pub fn string_repeat(vm: &mut Vm, args: &[u8]) -> NativeResult {
@@ -245,80 +275,84 @@ pub fn string_repeat(vm: &mut Vm, args: &[u8]) -> NativeResult {
} else {
1
};
Ok(vm.intern(&s.repeat(n)))
NativeResult::Ok(vm.intern(&s.repeat(n)))
}
pub fn string_pad_start(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let s_len = char_len(&s);
let target = if args.len() > 1 {
coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as usize
} else {
s.len()
s_len
};
let pad = if args.len() > 2 { as_string(vm, vm.reg(args[2])) } else { " ".to_string() };
if s.len() >= target || pad.is_empty() {
return Ok(vm.intern(&s));
if s_len >= target || pad.is_empty() {
return NativeResult::Ok(vm.intern(&s));
}
let needed = target - s.len();
let reps = needed.div_ceil(pad.len());
let mut out = pad.repeat(reps);
out.truncate(needed);
let needed = target - s_len;
let pad_len = char_len(&pad).max(1);
let reps = needed.div_ceil(pad_len);
let mut out = take_chars(&pad.repeat(reps), needed);
out.push_str(&s);
Ok(vm.intern(&out))
NativeResult::Ok(vm.intern(&out))
}
pub fn string_pad_end(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let s_len = char_len(&s);
let target = if args.len() > 1 {
coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as usize
} else {
s.len()
s_len
};
let pad = if args.len() > 2 { as_string(vm, vm.reg(args[2])) } else { " ".to_string() };
if s.len() >= target || pad.is_empty() {
return Ok(vm.intern(&s));
if s_len >= target || pad.is_empty() {
return NativeResult::Ok(vm.intern(&s));
}
let needed = target - s.len();
let reps = needed.div_ceil(pad.len());
let needed = target - s_len;
let pad_len = char_len(&pad).max(1);
let reps = needed.div_ceil(pad_len);
let mut out = s;
let pad_repeated = pad.repeat(reps);
out.push_str(&pad_repeated[..needed]);
Ok(vm.intern(&out))
out.push_str(&take_chars(&pad.repeat(reps), needed));
NativeResult::Ok(vm.intern(&out))
}
pub fn string_starts_with(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let n = char_len(&s);
if args.len() < 2 {
return Ok(JsValue::bool(false));
return NativeResult::Ok(JsValue::bool(false));
}
let search = as_string(vm, vm.reg(args[1]));
let pos = if args.len() > 2 {
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(s.len())
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(n)
} else {
0
};
Ok(JsValue::bool(s[pos..].starts_with(&search)))
NativeResult::Ok(JsValue::bool(s[byte_index_at_char(&s, pos)..].starts_with(&search)))
}
pub fn string_ends_with(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let n = char_len(&s);
if args.len() < 2 {
return Ok(JsValue::bool(false));
return NativeResult::Ok(JsValue::bool(false));
}
let search = as_string(vm, vm.reg(args[1]));
let end_pos = if args.len() > 2 {
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(s.len())
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize).min(n)
} else {
s.len()
n
};
Ok(JsValue::bool(s[..end_pos].ends_with(&search)))
NativeResult::Ok(JsValue::bool(s[..byte_index_at_char(&s, end_pos)].ends_with(&search)))
}
pub fn string_split(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
let parts = vec![s.clone()];
return Ok(make_string_array(vm, &parts));
return NativeResult::Ok(make_string_array(vm, &parts));
}
let sep_val = vm.reg(args[1]);
let limit = if args.len() > 2 {
@@ -334,13 +368,14 @@ pub fn string_split(vm: &mut Vm, args: &[u8]) -> NativeResult {
Some(p) => p,
None => {
let parts = vec![s.clone()];
return Ok(make_string_array(vm, &parts));
return NativeResult::Ok(make_string_array(vm, &parts));
}
};
let regex = unsafe { &*(fn_ptr as *const regex::Regex) };
// SAFETY: fn_ptr holds a Box<regex::Regex> pointer stored by regexp_constructor.
let regex = unsafe { &*(fn_ptr.as_ptr() as *const regex::Regex) };
let raw: Vec<&str> = regex.split(&s).collect();
let parts: Vec<String> = raw.iter().map(|p| p.to_string()).take(limit).collect();
return Ok(make_string_array(vm, &parts));
return NativeResult::Ok(make_string_array(vm, &parts));
}
let sep = as_string(vm, sep_val);
@@ -349,13 +384,13 @@ pub fn string_split(vm: &mut Vm, args: &[u8]) -> NativeResult {
} else {
s.split(&sep).map(|p| p.to_string()).take(limit).collect()
};
Ok(make_string_array(vm, &parts))
NativeResult::Ok(make_string_array(vm, &parts))
}
pub fn string_replace(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return Ok(vm.intern(&s));
return NativeResult::Ok(vm.intern(&s));
}
let pattern_val = vm.reg(args[1]);
let replacement = if args.len() > 2 { as_string(vm, vm.reg(args[2])) } else { String::new() };
@@ -365,22 +400,23 @@ pub fn string_replace(vm: &mut Vm, args: &[u8]) -> NativeResult {
let re = unsafe { &*re_ptr };
let fn_ptr = match re.native_fn() {
Some(p) => p,
None => return Ok(vm.intern(&s)),
None => return NativeResult::Ok(vm.intern(&s)),
};
let regex = unsafe { &*(fn_ptr as *const regex::Regex) };
// SAFETY: fn_ptr holds a Box<regex::Regex> pointer stored by regexp_constructor.
let regex = unsafe { &*(fn_ptr.as_ptr() as *const regex::Regex) };
let result = regex.replace_all(&s, replacement.as_str()).to_string();
return Ok(vm.intern(&result));
return NativeResult::Ok(vm.intern(&result));
}
let pattern = as_string(vm, pattern_val);
let result = s.replacen(&pattern, &replacement, 1);
Ok(vm.intern(&result))
NativeResult::Ok(vm.intern(&result))
}
pub fn string_match_fn(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return Ok(JsValue::undefined());
return NativeResult::Ok(JsValue::undefined());
}
let pattern_val = vm.reg(args[1]);
@@ -389,24 +425,25 @@ pub fn string_match_fn(vm: &mut Vm, args: &[u8]) -> NativeResult {
let re = unsafe { &*re_ptr };
let fn_ptr = match re.native_fn() {
Some(p) => p,
None => return Ok(JsValue::undefined()),
None => return NativeResult::Ok(JsValue::undefined()),
};
let regex = unsafe { &*(fn_ptr as *const regex::Regex) };
// SAFETY: fn_ptr holds a Box<regex::Regex> pointer stored by regexp_constructor.
let regex = unsafe { &*(fn_ptr.as_ptr() as *const regex::Regex) };
let matches: Vec<String> = regex.find_iter(&s).map(|m| m.as_str().to_string()).collect();
return Ok(make_string_array(vm, &matches));
return NativeResult::Ok(make_string_array(vm, &matches));
}
let pattern = as_string(vm, pattern_val);
if let Some(pos) = s.find(&pattern) {
return Ok(make_string_array(vm, &[s[pos..pos + pattern.len()].to_string()]));
return NativeResult::Ok(make_string_array(vm, &[s[pos..pos + pattern.len()].to_string()]));
}
Ok(JsValue::undefined())
NativeResult::Ok(JsValue::undefined())
}
pub fn string_search(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return Ok(JsValue::int(-1));
return NativeResult::Ok(JsValue::int(-1));
}
let pattern_val = vm.reg(args[1]);
@@ -415,30 +452,31 @@ pub fn string_search(vm: &mut Vm, args: &[u8]) -> NativeResult {
let re = unsafe { &*re_ptr };
let fn_ptr = match re.native_fn() {
Some(p) => p,
None => return Ok(JsValue::int(-1)),
None => return NativeResult::Ok(JsValue::int(-1)),
};
let regex = unsafe { &*(fn_ptr as *const regex::Regex) };
// SAFETY: fn_ptr holds a Box<regex::Regex> pointer stored by regexp_constructor.
let regex = unsafe { &*(fn_ptr.as_ptr() as *const regex::Regex) };
if let Some(m) = regex.find(&s) {
return Ok(JsValue::int(m.start() as i32));
return NativeResult::Ok(JsValue::int(m.start() as i32));
}
return Ok(JsValue::int(-1));
return NativeResult::Ok(JsValue::int(-1));
}
let pattern = as_string(vm, pattern_val);
if let Some(pos) = s.find(&pattern) {
return Ok(JsValue::int(pos as i32));
return NativeResult::Ok(JsValue::int(pos as i32));
}
Ok(JsValue::int(-1))
NativeResult::Ok(JsValue::int(-1))
}
pub fn string_trim_start(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
Ok(vm.intern(s.trim_start()))
NativeResult::Ok(vm.intern(s.trim_start()))
}
pub fn string_trim_end(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
Ok(vm.intern(s.trim_end()))
NativeResult::Ok(vm.intern(s.trim_end()))
}
pub fn string_code_point_at(vm: &mut Vm, args: &[u8]) -> NativeResult {
@@ -450,17 +488,17 @@ pub fn string_code_point_at(vm: &mut Vm, args: &[u8]) -> NativeResult {
};
let chars: Vec<char> = s.chars().collect();
if pos >= chars.len() {
return Ok(JsValue::undefined());
return NativeResult::Ok(JsValue::undefined());
}
let c = chars[pos] as u32;
if (0xD800..=0xDBFF).contains(&c) && pos + 1 < chars.len() {
let next = chars[pos + 1] as u32;
if (0xDC00..=0xDFFF).contains(&next) {
let cp = 0x10000 + ((c - 0xD800) << 10) + (next - 0xDC00);
return Ok(JsValue::int(cp as i32));
return NativeResult::Ok(JsValue::int(cp as i32));
}
}
Ok(JsValue::int(c as i32))
NativeResult::Ok(JsValue::int(c as i32))
}
pub fn string_normalize(vm: &mut Vm, args: &[u8]) -> NativeResult {
@@ -473,13 +511,13 @@ pub fn string_normalize(vm: &mut Vm, args: &[u8]) -> NativeResult {
"NFKD" => s.nfkd().collect(),
_ => s.nfc().collect(),
};
Ok(vm.intern(&result))
NativeResult::Ok(vm.intern(&result))
}
pub fn string_match_all(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return Err(JsValue::undefined());
return NativeResult::Err(JsValue::undefined());
}
let pattern_val = vm.reg(args[1]);
if is_regexp_obj(pattern_val, vm) {
@@ -487,19 +525,20 @@ pub fn string_match_all(vm: &mut Vm, args: &[u8]) -> NativeResult {
let re = unsafe { &*re_ptr };
let fn_ptr = match re.native_fn() {
Some(p) => p,
None => return Ok(JsValue::undefined()),
None => return NativeResult::Ok(JsValue::undefined()),
};
let regex = unsafe { &*(fn_ptr as *const regex::Regex) };
// SAFETY: fn_ptr holds a Box<regex::Regex> pointer stored by regexp_constructor.
let regex = unsafe { &*(fn_ptr.as_ptr() as *const regex::Regex) };
let matches: Vec<String> = regex.find_iter(&s).map(|m| m.as_str().to_string()).collect();
return Ok(make_string_array(vm, &matches));
return NativeResult::Ok(make_string_array(vm, &matches));
}
Ok(JsValue::undefined())
NativeResult::Ok(JsValue::undefined())
}
pub fn string_replace_all(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return Ok(vm.intern(&s));
return NativeResult::Ok(vm.intern(&s));
}
let pattern_val = vm.reg(args[1]);
let replacement = if args.len() > 2 { as_string(vm, vm.reg(args[2])) } else { String::new() };
@@ -508,13 +547,14 @@ pub fn string_replace_all(vm: &mut Vm, args: &[u8]) -> NativeResult {
let re = unsafe { &*re_ptr };
let fn_ptr = match re.native_fn() {
Some(p) => p,
None => return Ok(vm.intern(&s)),
None => return NativeResult::Ok(vm.intern(&s)),
};
let regex = unsafe { &*(fn_ptr as *const regex::Regex) };
// SAFETY: fn_ptr holds a Box<regex::Regex> pointer stored by regexp_constructor.
let regex = unsafe { &*(fn_ptr.as_ptr() as *const regex::Regex) };
let result = regex.replace_all(&s, replacement.as_str()).to_string();
return Ok(vm.intern(&result));
return NativeResult::Ok(vm.intern(&result));
}
let pattern = as_string(vm, pattern_val);
let result = s.replace(&pattern, &replacement);
Ok(vm.intern(&result))
NativeResult::Ok(vm.intern(&result))
}