feat: 支持 String 对象及成员方法

This commit is contained in:
爱特每羊队
2026-06-07 10:36:02 +08:00
parent 2063f4731b
commit ee6fcad370
8 changed files with 804 additions and 2 deletions
+164
View File
@@ -44,6 +44,28 @@ pub struct ErrorMethods {
pub eval_error: *const (),
}
pub struct StringMethods {
pub index_of: *const (),
pub includes: *const (),
pub char_at: *const (),
pub char_code_at: *const (),
pub concat: *const (),
pub slice: *const (),
pub substring: *const (),
pub to_upper_case: *const (),
pub to_lower_case: *const (),
pub trim: *const (),
pub repeat: *const (),
pub pad_start: *const (),
pub pad_end: *const (),
pub starts_with: *const (),
pub ends_with: *const (),
pub split: *const (),
pub replace: *const (),
pub match_fn: *const (),
pub search: *const (),
}
pub struct BuiltinWorld {
pub object_proto: P<JsObject>,
pub array_proto: P<JsObject>,
@@ -393,6 +415,148 @@ impl BuiltinWorld {
);
}
pub fn bind_string_methods(
&self,
methods: &StringMethods,
string_forge: &StringForge,
shape_forge: &ShapeForge,
) {
let proto_ptr = P::as_ptr(&self.string_proto) as *mut JsObject;
let proto = unsafe { &mut *proto_ptr };
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"indexOf",
methods.index_of,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"includes",
methods.includes,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"charAt",
methods.char_at,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"charCodeAt",
methods.char_code_at,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"concat",
methods.concat,
1,
);
let _ = Self::bind_method(proto, shape_forge, string_forge, "slice", methods.slice, 2);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"substring",
methods.substring,
2,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"toUpperCase",
methods.to_upper_case,
0,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"toLowerCase",
methods.to_lower_case,
0,
);
let _ = Self::bind_method(proto, shape_forge, string_forge, "trim", methods.trim, 0);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"repeat",
methods.repeat,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"padStart",
methods.pad_start,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"padEnd",
methods.pad_end,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"startsWith",
methods.starts_with,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"endsWith",
methods.ends_with,
1,
);
let _ = Self::bind_method(proto, shape_forge, string_forge, "split", methods.split, 1);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"replace",
methods.replace,
2,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"match",
methods.match_fn,
1,
);
let _ = Self::bind_method(
proto,
shape_forge,
string_forge,
"search",
methods.search,
1,
);
}
pub fn bind_method(
proto: &mut JsObject,
shape_forge: &ShapeForge,
+2
View File
@@ -13,6 +13,8 @@ oxide_types = { path = "../oxide_types" }
bumpalo.workspace = true
fast-float.workspace = true
ryu.workspace = true
memchr.workspace = true
regex.workspace = true
[dev-dependencies]
proptest.workspace = true
+1
View File
@@ -2,3 +2,4 @@ pub mod array;
pub mod error;
pub mod global;
pub mod object;
pub mod string;
+381
View File
@@ -0,0 +1,381 @@
use oxide_kernel::shape_forge::EMPTY_SHAPE_ID;
use oxide_types::object::JsObject;
use oxide_types::value::JsValue;
use crate::coercion;
use crate::native::NativeResult;
use crate::vm::Vm;
use memchr::memchr;
use regex::Regex;
fn this_string(vm: &Vm, args: &[u8]) -> String {
coercion::to_string(vm.kernel().string_forge().as_ref(), vm.reg(args[0]))
}
fn make_string_array(vm: &mut Vm, parts: &[String]) -> JsValue {
let proto = vm.kernel().builtin_world().array_proto.as_ptr() as *mut JsObject;
let n = parts.len().min(31);
let arr = vm.epoch().alloc(JsObject::new_array(
EMPTY_SHAPE_ID,
JsValue::from_js_object(proto),
n,
vm.epoch().bump(),
));
unsafe {
for (i, s) in parts.iter().enumerate().take(n) {
let sv = vm.intern(s);
(*arr).set_prop(i as u8, sv);
}
(*arr).set_prop_count(n as u8);
}
JsValue::from_js_object(arr)
}
fn as_string(vm: &Vm, val: JsValue) -> String {
coercion::to_string(vm.kernel().string_forge().as_ref(), val)
}
pub fn string_index_of(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return 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())
} else {
0
};
if search.is_empty() {
return Ok(JsValue::int(pos as i32));
}
let haystack = &s.as_bytes()[pos..];
if search.len() == 1 {
if let Some(idx) = memchr(search.as_bytes()[0], haystack) {
return Ok(JsValue::int((pos + idx) as i32));
}
} else if let Some(idx) = s[pos..].find(&search) {
return Ok(JsValue::int((pos + idx) as i32));
}
Ok(JsValue::int(-1))
}
pub fn string_includes(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return 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())
} else {
0
};
if search.is_empty() {
return Ok(JsValue::bool(true));
}
let haystack = &s.as_bytes()[pos..];
if search.len() == 1 {
Ok(JsValue::bool(
memchr(search.as_bytes()[0], haystack).is_some(),
))
} else {
Ok(JsValue::bool(s[pos..].contains(&search)))
}
}
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 Ok(vm.intern(&s[0..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(""));
}
let ch = &s[(idx as usize)..(idx as usize + 1)];
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 Ok(JsValue::int(s.as_bytes()[0] 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));
}
Ok(JsValue::int(s.as_bytes()[idx as usize] as i32))
}
pub fn string_concat(vm: &mut Vm, args: &[u8]) -> NativeResult {
let mut result = this_string(vm, args);
let sf = vm.kernel().string_forge().as_ref();
for &arg_reg in args.iter().skip(1) {
result.push_str(&coercion::to_string(sf, vm.reg(arg_reg)));
}
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 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 {
(n + v).max(0)
} else {
v.min(n)
}
} else {
0
};
let end = if args.len() > 2 {
let v = coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as i32;
if v < 0 {
(n + v).max(0)
} else {
v.min(n)
}
} else {
n
};
let start = start as usize;
let end = end as usize;
let result = if start < end { &s[start..end] } else { "" };
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 mut start = if args.len() > 1 {
(coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as i32)
.max(0)
.min(n)
} else {
0
};
let mut end = if args.len() > 2 {
(coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as i32)
.max(0)
.min(n)
} else {
n
};
if start > end {
std::mem::swap(&mut start, &mut end);
}
let result = &s[start as usize..end as usize];
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()))
}
pub fn string_to_lower_case(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
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()))
}
pub fn string_repeat(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let n = if args.len() > 1 {
(coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as usize)
.min(10000)
} else {
1
};
Ok(vm.intern(&s.repeat(n)))
}
pub fn string_pad_start(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let target = if args.len() > 1 {
coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as usize
} else {
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));
}
let needed = target - s.len();
let reps = needed.div_ceil(pad.len());
let mut out = pad.repeat(reps);
out.truncate(needed);
out.push_str(&s);
Ok(vm.intern(&out))
}
pub fn string_pad_end(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
let target = if args.len() > 1 {
coercion::to_number(vm.reg(args[1]), vm.kernel().string_forge().as_ref()) as usize
} else {
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));
}
let needed = target - s.len();
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))
}
pub fn string_starts_with(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return 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())
} else {
0
};
Ok(JsValue::bool(s[pos..].starts_with(&search)))
}
pub fn string_ends_with(vm: &mut Vm, args: &[u8]) -> NativeResult {
let s = this_string(vm, args);
if args.len() < 2 {
return 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())
} else {
s.len()
};
Ok(JsValue::bool(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));
}
let sep = as_string(vm, vm.reg(args[1]));
let limit = if args.len() > 2 {
coercion::to_number(vm.reg(args[2]), vm.kernel().string_forge().as_ref()) as usize
} else {
usize::MAX
};
let parts: Vec<String>;
if sep.starts_with('/') && sep.len() > 2 && sep.ends_with('/') {
let pattern = &sep[1..sep.len() - 1];
if let Ok(re) = Regex::new(pattern) {
let raw: Vec<&str> = re.split(&s).collect();
parts = raw.iter().map(|p| p.to_string()).take(limit).collect();
} else {
parts = vec![s.clone()];
}
} else if sep.is_empty() {
parts = s.chars().map(|c| c.to_string()).take(limit).collect();
} else {
parts = s.split(&sep).map(|p| p.to_string()).take(limit).collect();
}
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));
}
let pattern = as_string(vm, vm.reg(args[1]));
let replacement = if args.len() > 2 {
as_string(vm, vm.reg(args[2]))
} else {
String::new()
};
let result = if pattern.starts_with('/') && pattern.len() > 2 && pattern.ends_with('/') {
let pat = &pattern[1..pattern.len() - 1];
if let Ok(re) = Regex::new(pat) {
re.replace_all(&s, replacement.as_str()).to_string()
} else {
s.clone()
}
} else {
s.replacen(&pattern, &replacement, 1)
};
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());
}
let pattern = as_string(vm, vm.reg(args[1]));
if pattern.starts_with('/') && pattern.len() > 2 && pattern.ends_with('/') {
let pat = &pattern[1..pattern.len() - 1];
if let Ok(re) = Regex::new(pat) {
let matches: Vec<String> = re.find_iter(&s).map(|m| m.as_str().to_string()).collect();
return Ok(make_string_array(vm, &matches));
}
} else if let Some(pos) = s.find(&pattern) {
return Ok(make_string_array(
vm,
&[s[pos..pos + pattern.len()].to_string()],
));
}
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));
}
let pattern = as_string(vm, vm.reg(args[1]));
if pattern.starts_with('/') && pattern.len() > 2 && pattern.ends_with('/') {
let pat = &pattern[1..pattern.len() - 1];
if let Ok(re) = Regex::new(pat) {
if let Some(m) = re.find(&s) {
return Ok(JsValue::int(m.start() as i32));
}
}
} else if let Some(pos) = s.find(&pattern) {
return Ok(JsValue::int(pos as i32));
}
Ok(JsValue::int(-1))
}
+53 -2
View File
@@ -8,7 +8,7 @@ use oxide_compiler::opcode::{self, OpCode};
use crate::coercion;
use crate::native::NativeFn;
use oxide_kernel::builtin::{ArrayMethods, ErrorMethods, ObjectMethods};
use oxide_kernel::builtin::{ArrayMethods, ErrorMethods, ObjectMethods, StringMethods};
use oxide_kernel::kernel::{KernelConfig, OxideKernel};
use oxide_kernel::prop_forge::PropTemplate;
use oxide_kernel::shape_forge::EMPTY_SHAPE_ID;
@@ -131,6 +131,45 @@ pub fn init_kernel_builtins(kernel: &Arc<OxideKernel>) {
let bump3 = bumpalo::Bump::new();
global.set_prop_expand(cur_count, err_val, &bump3);
global.bump_generation();
let string_methods = StringMethods {
index_of: crate::builtins::string::string_index_of as *const (),
includes: crate::builtins::string::string_includes as *const (),
char_at: crate::builtins::string::string_char_at as *const (),
char_code_at: crate::builtins::string::string_char_code_at as *const (),
concat: crate::builtins::string::string_concat as *const (),
slice: crate::builtins::string::string_slice as *const (),
substring: crate::builtins::string::string_substring as *const (),
to_upper_case: crate::builtins::string::string_to_upper_case as *const (),
to_lower_case: crate::builtins::string::string_to_lower_case as *const (),
trim: crate::builtins::string::string_trim as *const (),
repeat: crate::builtins::string::string_repeat as *const (),
pad_start: crate::builtins::string::string_pad_start as *const (),
pad_end: crate::builtins::string::string_pad_end as *const (),
starts_with: crate::builtins::string::string_starts_with as *const (),
ends_with: crate::builtins::string::string_ends_with as *const (),
split: crate::builtins::string::string_split as *const (),
replace: crate::builtins::string::string_replace as *const (),
match_fn: crate::builtins::string::string_match_fn as *const (),
search: crate::builtins::string::string_search as *const (),
};
kernel.builtin_world().bind_string_methods(
&string_methods,
kernel.string_forge().as_ref(),
kernel.shape_forge().as_ref(),
);
let si_str = kernel.string_forge().intern("String").0;
let str_shape = kernel.shape_forge().make_shape(global.shape_id(), si_str);
let str_val = JsValue::from_js_object(
kernel.builtin_world().string_constructor.as_ptr() as *mut JsObject
);
let cur_count = global.prop_count();
global.set_shape_id(str_shape);
global.set_prop_count(cur_count + 1);
let bump4 = bumpalo::Bump::new();
global.set_prop_expand(cur_count, str_val, &bump4);
global.bump_generation();
}
impl Vm {
@@ -552,8 +591,20 @@ impl Vm {
}
OpCode::IC_GET_PROP => {
let obj_ptr = self.regs[a].as_object_ptr() as *mut JsObject;
let val = self.regs[a];
let obj_ptr = val.as_object_ptr() as *mut JsObject;
if obj_ptr.is_null() {
if val.is_string() {
let proto_ptr =
self.kernel.builtin_world().string_proto.as_ptr() as *mut JsObject;
let proto = unsafe { &*proto_ptr };
let prop_name_si = self.regs[b].as_string_index();
if let Some(resolved) = self.resolve_property(proto, prop_name_si) {
self.regs[a] = resolved;
self.pc += 1;
continue;
}
}
return Err("IC_GET_PROP on non-object".into());
}
let obj = unsafe { &*obj_ptr };
@@ -0,0 +1,167 @@
use oxide_compiler::compiler::Compiler;
use oxide_types::value::JsValue;
use oxide_vm::vm::Vm;
fn eval(vm: &mut Vm, source: &str) -> Result<JsValue, String> {
let allocator = oxide_parser::Allocator::default();
let program =
oxide_parser::parse(&allocator, source).map_err(|e| format!("Parse error: {:?}", e))?;
let module = Compiler::new()
.compile(&program)
.map_err(|e| format!("Compile error: {}", e))?;
vm.run(&module)
}
fn to_str(vm: &Vm, val: JsValue) -> String {
vm.kernel()
.string_forge()
.lookup(val.as_string_index())
.unwrap_or_default()
}
#[test]
fn string_index_of_found() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.indexOf('e')").unwrap();
assert_eq!(result.as_int(), 1);
}
#[test]
fn string_index_of_not_found() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.indexOf('x')").unwrap();
assert_eq!(result.as_int(), -1);
}
#[test]
fn string_includes_true() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.includes('ell')").unwrap();
assert!(result.as_bool());
}
#[test]
fn string_includes_false() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.includes('xyz')").unwrap();
assert!(!result.as_bool());
}
#[test]
fn string_char_at() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'hello'.charAt(1)").unwrap();
assert_eq!(to_str(&vm, s), "e");
}
#[test]
fn string_char_code_at() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.charCodeAt(0)").unwrap();
assert_eq!(result.as_int(), 104);
}
#[test]
fn string_concat() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'hello'.concat(' world')").unwrap();
assert_eq!(to_str(&vm, s), "hello world");
}
#[test]
fn string_slice() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'hello'.slice(1, 4)").unwrap();
assert_eq!(to_str(&vm, s), "ell");
}
#[test]
fn string_substring() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'hello'.substring(0, 2)").unwrap();
assert_eq!(to_str(&vm, s), "he");
}
#[test]
fn string_to_upper_case() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'hello'.toUpperCase()").unwrap();
assert_eq!(to_str(&vm, s), "HELLO");
}
#[test]
fn string_to_lower_case() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'HELLO'.toLowerCase()").unwrap();
assert_eq!(to_str(&vm, s), "hello");
}
#[test]
fn string_trim() {
let mut vm = Vm::new();
let s = eval(&mut vm, "' hi '.trim()").unwrap();
assert_eq!(to_str(&vm, s), "hi");
}
#[test]
fn string_repeat() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'abc'.repeat(3)").unwrap();
assert_eq!(to_str(&vm, s), "abcabcabc");
}
#[test]
fn string_pad_start() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'5'.padStart(4, '0')").unwrap();
assert_eq!(to_str(&vm, s), "0005");
}
#[test]
fn string_pad_end() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'hi'.padEnd(4)").unwrap();
assert_eq!(to_str(&vm, s), "hi ");
}
#[test]
fn string_starts_with() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.startsWith('hel')").unwrap();
assert!(result.as_bool());
}
#[test]
fn string_ends_with() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.endsWith('lo')").unwrap();
assert!(result.as_bool());
}
#[test]
fn string_split_comma() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'a,b,c'.split(',')").unwrap();
assert!(result.is_object());
}
#[test]
fn string_replace_first() {
let mut vm = Vm::new();
let s = eval(&mut vm, "'hello'.replace('l', 'L')").unwrap();
assert_eq!(to_str(&vm, s), "heLlo");
}
#[test]
fn string_search_found() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.search('ll')").unwrap();
assert_eq!(result.as_int(), 2);
}
#[test]
fn string_search_not_found() {
let mut vm = Vm::new();
let result = eval(&mut vm, "'hello'.search('x')").unwrap();
assert_eq!(result.as_int(), -1);
}