String unification bridge: VM string-like normalization (compare/add/method fallbacks) and Interpreter normalization for compare/add; fix lifetime issues; update CURRENT_TASK status

This commit is contained in:
Moe Charm
2025-08-30 23:58:58 +09:00
parent 4ae92cfb56
commit 8d7888f6f1
4 changed files with 122 additions and 27 deletions

View File

@ -34,10 +34,10 @@ How To RunNyash-only
Next Steps優先順・更新 Next Steps優先順・更新
1. String統一ブリッジ完了 1. String統一ブリッジ(実装済・一次完了
- VM: 内部String受けのフォールバックを全パスで拾うlength/isEmpty/charCodeAt/concat/+ - VM: 比較/加算/代表メソッドのフォールバックlength/isEmpty/charCodeAt/concat/+をstring-like正規化で実装
- Interpreter: 同等のフォールバック/正規化(比較・結合・代表メソッド - Interpreter: 比較/加算はstring-like正規化を適用メソッドは後続で最小追補があれば対応
- 混在比較/結合の回帰ケース追加(内部/プラグイン/プリミティブ混在) - 例: encoding_min/regex_min/toml_min/path_min で回帰確認
2. tools/pyc: IR→Nyashの反映強化return/If/Assignを安定化、Strictスイッチ連動 2. tools/pyc: IR→Nyashの反映強化return/If/Assignを安定化、Strictスイッチ連動
3. Strictスイッチ: tools/pycunsupported_nodes非空でErr、envでON/OFF 3. Strictスイッチ: tools/pycunsupported_nodes非空でErr、envでON/OFF
4. CLI隠しフラグ `--pyc`/`--pyc-native`Parser→Compiler→AOTの一本化導線 4. CLI隠しフラグ `--pyc`/`--pyc-native`Parser→Compiler→AOTの一本化導線

View File

@ -1003,6 +1003,34 @@ impl VM {
/// Execute a forced plugin invocation (no builtin fallback) /// Execute a forced plugin invocation (no builtin fallback)
pub(super) fn execute_plugin_invoke(&mut self, dst: Option<ValueId>, box_val: ValueId, method: &str, args: &[ValueId]) -> Result<ControlFlow, VMError> { pub(super) fn execute_plugin_invoke(&mut self, dst: Option<ValueId>, box_val: ValueId, method: &str, args: &[ValueId]) -> Result<ControlFlow, VMError> {
// Helper: extract UTF-8 string from internal StringBox, Result.Ok(String-like), or plugin StringBox via toUtf8
fn extract_string_from_box(bx: &dyn crate::box_trait::NyashBox) -> Option<String> {
// Internal StringBox
if let Some(sb) = bx.as_any().downcast_ref::<crate::box_trait::StringBox>() {
return Some(sb.value.clone());
}
// Result.Ok(inner) → recurse
if let Some(res) = bx.as_any().downcast_ref::<crate::boxes::result::NyashResultBox>() {
if let crate::boxes::result::NyashResultBox::Ok(inner) = res { return extract_string_from_box(inner.as_ref()); }
}
// Plugin StringBox → call toUtf8
if let Some(p) = bx.as_any().downcast_ref::<crate::runtime::plugin_loader_v2::PluginBoxV2>() {
if p.box_type == "StringBox" {
let host = crate::runtime::get_global_plugin_host();
let tmp: Option<String> = if let Ok(ro) = host.read() {
if let Ok(val_opt) = ro.invoke_instance_method("StringBox", "toUtf8", p.inner.instance_id, &[]) {
if let Some(vb) = val_opt {
if let Some(sb2) = vb.as_any().downcast_ref::<crate::box_trait::StringBox>() {
Some(sb2.value.clone())
} else { None }
} else { None }
} else { None }
} else { None };
if tmp.is_some() { return tmp; }
}
}
None
}
let recv = self.get_value(box_val)?; let recv = self.get_value(box_val)?;
// Allow static birth on primitives/builtin boxes to create a plugin instance. // Allow static birth on primitives/builtin boxes to create a plugin instance.
if method == "birth" && !matches!(recv, VMValue::BoxRef(ref b) if b.as_any().downcast_ref::<crate::runtime::plugin_loader_v2::PluginBoxV2>().is_some()) { if method == "birth" && !matches!(recv, VMValue::BoxRef(ref b) if b.as_any().downcast_ref::<crate::runtime::plugin_loader_v2::PluginBoxV2>().is_some()) {
@ -1215,33 +1243,34 @@ impl VM {
return Ok(ControlFlow::Continue); return Ok(ControlFlow::Continue);
} }
} }
// Fallback: support common methods on internal StringBox without requiring PluginBox receiver // Fallback: support common string-like methods without requiring PluginBox receiver
if let VMValue::BoxRef(ref bx) = recv { if let VMValue::BoxRef(ref bx) = recv {
if let Some(sb) = bx.as_any().downcast_ref::<crate::box_trait::StringBox>() { // Try to view receiver as string (internal, plugin, or Result.Ok)
if let Some(s) = extract_string_from_box(bx.as_ref()) {
match method { match method {
"length" => { "length" => {
if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::Integer(sb.value.len() as i64)); } if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::Integer(s.len() as i64)); }
return Ok(ControlFlow::Continue); return Ok(ControlFlow::Continue);
} }
"is_empty" | "isEmpty" => { "is_empty" | "isEmpty" => {
if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::Bool(sb.value.is_empty())); } if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::Bool(s.is_empty())); }
return Ok(ControlFlow::Continue); return Ok(ControlFlow::Continue);
} }
"charCodeAt" => { "charCodeAt" => {
let idx_v = if let Some(a0) = args.get(0) { self.get_value(*a0)? } else { VMValue::Integer(0) }; let idx_v = if let Some(a0) = args.get(0) { self.get_value(*a0)? } else { VMValue::Integer(0) };
let idx = match idx_v { VMValue::Integer(i) => i.max(0) as usize, _ => 0 }; let idx = match idx_v { VMValue::Integer(i) => i.max(0) as usize, _ => 0 };
let code = sb.value.chars().nth(idx).map(|c| c as u32 as i64).unwrap_or(0); let code = s.chars().nth(idx).map(|c| c as u32 as i64).unwrap_or(0);
if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::Integer(code)); } if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::Integer(code)); }
return Ok(ControlFlow::Continue); return Ok(ControlFlow::Continue);
} }
"concat" => { "concat" => {
let rhs_v = if let Some(a0) = args.get(0) { self.get_value(*a0)? } else { VMValue::String(String::new()) }; let rhs_v = if let Some(a0) = args.get(0) { self.get_value(*a0)? } else { VMValue::String(String::new()) };
let rhs_s = match rhs_v { let rhs_s = match rhs_v {
VMValue::String(s) => s, VMValue::String(ss) => ss,
VMValue::BoxRef(br) => br.to_string_box().value, VMValue::BoxRef(br) => extract_string_from_box(br.as_ref()).unwrap_or_else(|| br.to_string_box().value),
_ => rhs_v.to_string(), _ => rhs_v.to_string(),
}; };
let mut new_s = sb.value.clone(); let mut new_s = s.clone();
new_s.push_str(&rhs_s); new_s.push_str(&rhs_s);
let out = Box::new(crate::box_trait::StringBox::new(new_s)); let out = Box::new(crate::box_trait::StringBox::new(new_s));
if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::BoxRef(std::sync::Arc::from(out as Box<dyn crate::box_trait::NyashBox>))); } if let Some(dst_id) = dst { self.set_value(dst_id, VMValue::BoxRef(std::sync::Arc::from(out as Box<dyn crate::box_trait::NyashBox>))); }

View File

@ -11,6 +11,30 @@ use crate::mir::{BinaryOp, CompareOp, UnaryOp};
use super::vm::{VM, VMError, VMValue}; use super::vm::{VM, VMError, VMValue};
impl VM { impl VM {
/// Try to view a BoxRef as a UTF-8 string (internal StringBox, Result.Ok(String-like), or plugin StringBox via toUtf8)
fn try_boxref_to_string(&self, b: &dyn crate::box_trait::NyashBox) -> Option<String> {
// Internal StringBox
if let Some(sb) = b.as_any().downcast_ref::<crate::box_trait::StringBox>() {
return Some(sb.value.clone());
}
// Result.Ok(inner) → recurse
if let Some(res) = b.as_any().downcast_ref::<crate::boxes::result::NyashResultBox>() {
if let crate::boxes::result::NyashResultBox::Ok(inner) = res { return self.try_boxref_to_string(inner.as_ref()); }
}
// Plugin StringBox → call toUtf8
if let Some(pb) = b.as_any().downcast_ref::<crate::runtime::plugin_loader_v2::PluginBoxV2>() {
if pb.box_type == "StringBox" {
let host = crate::runtime::get_global_plugin_host();
let tmp: Option<String> = if let Ok(ro) = host.read() {
if let Ok(val_opt) = ro.invoke_instance_method("StringBox", "toUtf8", pb.inner.instance_id, &[]) {
if let Some(vb) = val_opt { if let Some(sb2) = vb.as_any().downcast_ref::<crate::box_trait::StringBox>() { Some(sb2.value.clone()) } else { None } } else { None }
} else { None }
} else { None };
if tmp.is_some() { return tmp; }
}
}
None
}
/// Execute binary operation /// Execute binary operation
pub(super) fn execute_binary_op(&self, op: &BinaryOp, left: &VMValue, right: &VMValue) -> Result<VMValue, VMError> { pub(super) fn execute_binary_op(&self, op: &BinaryOp, left: &VMValue, right: &VMValue) -> Result<VMValue, VMError> {
let debug_bin = std::env::var("NYASH_VM_DEBUG_BIN").ok().as_deref() == Some("1"); let debug_bin = std::env::var("NYASH_VM_DEBUG_BIN").ok().as_deref() == Some("1");
@ -57,12 +81,14 @@ impl VM {
// String + BoxRef concatenation // String + BoxRef concatenation
(VMValue::String(l), VMValue::BoxRef(r)) => { (VMValue::String(l), VMValue::BoxRef(r)) => {
match op { BinaryOp::Add => Ok(VMValue::String(format!("{}{}", l, r.to_string_box().value))), _ => Err(VMError::TypeError("String-BoxRef operations only support addition".to_string())) } let rs = self.try_boxref_to_string(r.as_ref()).unwrap_or_else(|| r.to_string_box().value);
match op { BinaryOp::Add => Ok(VMValue::String(format!("{}{}", l, rs))), _ => Err(VMError::TypeError("String-BoxRef operations only support addition".to_string())) }
}, },
// BoxRef + String concatenation // BoxRef + String concatenation
(VMValue::BoxRef(l), VMValue::String(r)) => { (VMValue::BoxRef(l), VMValue::String(r)) => {
match op { BinaryOp::Add => Ok(VMValue::String(format!("{}{}", l.to_string_box().value, r))), _ => Err(VMError::TypeError("BoxRef-String operations only support addition".to_string())) } let ls = self.try_boxref_to_string(l.as_ref()).unwrap_or_else(|| l.to_string_box().value);
match op { BinaryOp::Add => Ok(VMValue::String(format!("{}{}", ls, r))), _ => Err(VMError::TypeError("BoxRef-String operations only support addition".to_string())) }
}, },
// Arithmetic with BoxRef(IntegerBox) — support both legacy and new IntegerBox // Arithmetic with BoxRef(IntegerBox) — support both legacy and new IntegerBox
@ -90,6 +116,15 @@ impl VM {
}; };
Ok(VMValue::Integer(res)) Ok(VMValue::Integer(res))
} }
// BoxRef + BoxRef string-like concatenation
(VMValue::BoxRef(li), VMValue::BoxRef(ri)) => {
if matches!(*op, BinaryOp::Add) {
let ls = self.try_boxref_to_string(li.as_ref()).unwrap_or_else(|| li.to_string_box().value);
let rs = self.try_boxref_to_string(ri.as_ref()).unwrap_or_else(|| ri.to_string_box().value);
return Ok(VMValue::String(format!("{}{}", ls, rs)));
}
Err(VMError::TypeError("Unsupported BoxRef+BoxRef operation".to_string()))
}
// Mixed Integer forms // Mixed Integer forms
(VMValue::BoxRef(li), VMValue::Integer(r)) if li.as_any().downcast_ref::<crate::box_trait::IntegerBox>().is_some() (VMValue::BoxRef(li), VMValue::Integer(r)) if li.as_any().downcast_ref::<crate::box_trait::IntegerBox>().is_some()

View File

@ -107,6 +107,33 @@ pub(super) fn try_mod_operation(left: &dyn NyashBox, right: &dyn NyashBox) -> Re
// ======================================================================================== // ========================================================================================
impl NyashInterpreter { impl NyashInterpreter {
/// Try to extract a UTF-8 string from a NyashBox: supports internal StringBox,
/// Result.Ok(String-like), and Plugin StringBox via toUtf8 (when plugins enabled).
fn try_box_to_string(&self, b: &dyn NyashBox) -> Option<String> {
// Internal StringBox
if let Some(sb) = b.as_any().downcast_ref::<StringBox>() { return Some(sb.value.clone()); }
// Result.Ok(inner) → recurse
if let Some(res) = b.as_any().downcast_ref::<crate::boxes::result::NyashResultBox>() {
if let crate::boxes::result::NyashResultBox::Ok(inner) = res { return self.try_box_to_string(inner.as_ref()); }
}
// Plugin StringBox via toUtf8
#[cfg(all(feature = "plugins", not(target_arch = "wasm32")))]
{
if let Some(pb) = b.as_any().downcast_ref::<crate::runtime::plugin_loader_v2::PluginBoxV2>() {
if pb.box_type == "StringBox" {
let host = crate::runtime::get_global_plugin_host();
if let Ok(ro) = host.read() {
if let Ok(val_opt) = ro.invoke_instance_method("StringBox", "toUtf8", pb.inner.instance_id, &[]) {
if let Some(vb) = val_opt {
if let Some(sb2) = vb.as_any().downcast_ref::<StringBox>() { return Some(sb2.value.clone()); }
}
}
}
}
}
}
None
}
/// 二項演算を実行 /// 二項演算を実行
pub(super) fn execute_binary_op(&mut self, op: &BinaryOperator, left: &ASTNode, right: &ASTNode) pub(super) fn execute_binary_op(&mut self, op: &BinaryOperator, left: &ASTNode, right: &ASTNode)
-> Result<Box<dyn NyashBox>, RuntimeError> -> Result<Box<dyn NyashBox>, RuntimeError>
@ -119,6 +146,16 @@ impl NyashInterpreter {
match op { match op {
BinaryOperator::Add => { BinaryOperator::Add => {
// Prefer string-like concatenation when either side is string-like
if let (Some(ls), Some(rs)) = (self.try_box_to_string(left_val), self.try_box_to_string(right_val)) {
return Ok(Box::new(StringBox::new(format!("{}{}", ls, rs))));
}
if let Some(ls) = self.try_box_to_string(left_val) {
return Ok(Box::new(StringBox::new(format!("{}{}", ls, right_val.to_string_box().value))));
}
if let Some(rs) = self.try_box_to_string(right_val) {
return Ok(Box::new(StringBox::new(format!("{}{}", left_val.to_string_box().value, rs))));
}
if let Some(result) = try_add_operation(left_val, right_val) { if let Some(result) = try_add_operation(left_val, right_val) {
Ok(result) Ok(result)
} else { } else {
@ -252,12 +289,9 @@ impl NyashInterpreter {
return Ok(left_int.value == right_int.value); return Ok(left_int.value == right_int.value);
} }
// StringBox comparison // String-like comparison (internal/Result.Ok/plugin StringBox)
if let (Some(left_str), Some(right_str)) = ( if let (Some(ls), Some(rs)) = (self.try_box_to_string(left), self.try_box_to_string(right)) {
left.as_any().downcast_ref::<StringBox>(), return Ok(ls == rs);
right.as_any().downcast_ref::<StringBox>()
) {
return Ok(left_str.value == right_str.value);
} }
// BoolBox comparison // BoolBox comparison
@ -287,12 +321,9 @@ impl NyashInterpreter {
return Ok(left_int.value < right_int.value); return Ok(left_int.value < right_int.value);
} }
// StringBox comparison (lexicographic) // String-like comparison (lexicographic)
if let (Some(left_str), Some(right_str)) = ( if let (Some(ls), Some(rs)) = (self.try_box_to_string(left), self.try_box_to_string(right)) {
left.as_any().downcast_ref::<StringBox>(), return Ok(ls < rs);
right.as_any().downcast_ref::<StringBox>()
) {
return Ok(left_str.value < right_str.value);
} }
Err(RuntimeError::InvalidOperation { Err(RuntimeError::InvalidOperation {