fix: 闭包捕获/upvalue 分析纳入参数默认值表达式（含嵌套函数声明）

2026-10-23 18:21:13 +08:00 · 2026-08-07 19:04:25 +08:00
parent 9981086424
commit e3d8092088
2 changed files with 99 additions and 4 deletions
@@ -123,6 +123,7 @@ impl Emitter {
                let mut inner = shadow.clone();
                inner.extend(self.collect_fn_param_names(&fd.params));
                inner.extend(self.collect_own_binding_names(&[], body));
+                self.collect_fn_default_names(&fd.params, ref_set, shadow, out);
                self.collect_capture_names_shadowed(body, ref_set, &inner, out);
            }
            Statement::IfStatement(is) => {
@@ -240,6 +241,20 @@ impl Emitter {
        }
    }

+    /// 收集函数参数默认值表达式里的引用（子层 upvalue 判定；参数名遮蔽）。
+    fn collect_fn_default_names(
+        &self, params: &oxide_parser::FormalParameters, ref_set: &HashSet<String>, shadow: &HashSet<String>,
+        out: &mut HashSet<String>,
+    ) {
+        for p in &params.items {
+            if let Some(init) = &p.initializer {
+                self.collect_capture_names_expr(init, ref_set, shadow, out);
+            }
+            let mut param_shadow = shadow.clone();
+            self.collect_capture_names_binding_pattern(&p.pattern, ref_set, shadow, out, &mut param_shadow);
+        }
+    }
+
    fn collect_capture_names_expr(
        &self, expr: &Expression, ref_set: &HashSet<String>, shadow: &HashSet<String>, out: &mut HashSet<String>,
    ) {
@@ -272,12 +287,14 @@ impl Emitter {
                let mut inner = shadow.clone();
                inner.extend(self.collect_fn_param_names(&fe.params));
                inner.extend(self.collect_own_binding_names(&[], body));
+                self.collect_fn_default_names(&fe.params, ref_set, shadow, out);
                self.collect_capture_names_shadowed(body, ref_set, &inner, out);
            }
            Expression::ArrowFunctionExpression(ae) => {
                let mut inner = shadow.clone();
                inner.extend(self.collect_fn_param_names(&ae.params));
                inner.extend(self.collect_own_binding_names(&[], &ae.body.statements));
+                self.collect_fn_default_names(&ae.params, ref_set, shadow, out);
                self.collect_capture_names_shadowed(&ae.body.statements, ref_set, &inner, out);
            }
            Expression::BinaryExpression(be) => {
@@ -382,21 +399,41 @@ impl Emitter {
    }

    /// 分析本函数：哪些绑定被任意深度嵌套函数捕获 → captured_bindings。
-    pub(crate) fn collect_captured_bindings(&self, stmts: &[Statement], own: &HashSet<String>) -> BTreeMap<String, u8> {
+    pub(crate) fn collect_captured_bindings(
+        &self, stmts: &[Statement], extra_exprs: &[&oxide_parser::Expression], own: &HashSet<String>,
+    ) -> BTreeMap<String, u8> {
        let mut names = HashSet::new();
        for stmt in stmts {
            self.collect_captured_stmt(stmt, own, &mut names);
        }
+        // 参数默认值表达式（不在 body_stmts 内）里的嵌套函数引用也要纳入捕获，
+        // 否则默认值内 IIFE 引用全局/外层变量走 LOAD_VAR 读寄存器残留（B022 扩展）。
+        for expr in extra_exprs {
+            self.collect_captured_expr(expr, own, &mut names);
+        }
        // 名字排序分配 cell_idx（稳定跨 run，父 MAKE_CELL 与子 upvalue 统一引用）
        let mut sorted: Vec<String> = names.into_iter().collect();
        sorted.sort();
        sorted.into_iter().enumerate().map(|(i, n)| (n, i as u8)).collect()
    }

+    /// 收集函数参数默认值表达式里的嵌套函数引用（父层 captured/子层 upvalue 判定）。
+    fn collect_fn_default_captured(
+        &self, params: &oxide_parser::FormalParameters, own: &HashSet<String>, out: &mut HashSet<String>,
+    ) {
+        for p in &params.items {
+            if let Some(init) = &p.initializer {
+                self.collect_captured_expr(init, own, out);
+            }
+            self.collect_captured_binding_pattern(&p.pattern, own, out);
+        }
+    }
+
    /// 只从嵌套函数节点进入扫描（本函数直接引用不算捕获）。
    fn collect_captured_stmt(&self, stmt: &Statement, own: &HashSet<String>, out: &mut HashSet<String>) {        match stmt {
            Statement::FunctionDeclaration(fd) => {
                let body: &[Statement] = fd.body.as_ref().map(|b| &b.statements[..]).unwrap_or(&[]);
+                self.collect_fn_default_captured(&fd.params, own, out);
                self.collect_capture_names(body, own, out);
            }
            Statement::ExpressionStatement(es) => self.collect_captured_expr(&es.expression, own, out),
@@ -523,11 +560,13 @@ impl Emitter {
                let body: &[Statement] = fe.body.as_ref().map(|b| &b.statements[..]).unwrap_or(&[]);
                let mut inner = self.collect_fn_param_names(&fe.params);
                inner.extend(self.collect_own_binding_names(&[], body));
+                self.collect_fn_default_captured(&fe.params, own, out);
                self.collect_capture_names_shadowed(body, own, &inner, out);
            }
            Expression::ArrowFunctionExpression(ae) => {
                let mut inner = self.collect_fn_param_names(&ae.params);
                inner.extend(self.collect_own_binding_names(&[], &ae.body.statements));
+                self.collect_fn_default_captured(&ae.params, own, out);
                self.collect_capture_names_shadowed(&ae.body.statements, own, &inner, out);
            }
            Expression::CallExpression(ce) => {
@@ -634,7 +673,7 @@ impl Emitter {
    /// 若在父 `upvalue_captures`（父自身从更外层捕获）则链式标记 `parent_uv_idx`，
    /// 运行时从父闭包 upvalues 取 cell。enclosing_reg 由父 emit 完成后填充。
    pub(crate) fn collect_upvalue_names(
-        &self, body_stmts: &[Statement], parent_captured: &BTreeMap<String, u8>,
+        &self, body_stmts: &[Statement], extra_exprs: &[&oxide_parser::Expression], parent_captured: &BTreeMap<String, u8>,
        parent_upvalues: &[UpvalueCapture], sub_own: &HashSet<String>,
    ) -> Vec<UpvalueCapture> {
        let mut parent_names: HashSet<String> = parent_captured.keys().cloned().collect();
@@ -643,6 +682,10 @@ impl Emitter {
        }
        let mut names = HashSet::new();
        self.collect_capture_names_shadowed(body_stmts, &parent_names, sub_own, &mut names);
+        // 参数默认值表达式（不在 body_stmts）里的嵌套函数引用也会捕获父变量。
+        for expr in extra_exprs {
+            self.collect_capture_names_expr(expr, &parent_names, sub_own, &mut names);
+        }
        // HashSet 迭代序带随机种子（进程级非确定），必须排序使 upvalue_captures 的顺序与
        // 父 captured_bindings 的 cell_idx（BTreeMap 名字序）对齐——否则 LOAD_UPVALUE 的
        // a 槽 uv_idx 编码与 cell_idx 错位，读错 upvalue。
@@ -523,6 +523,39 @@ impl Emitter {
        Self
    }

+    /// 遍历解构 pattern 收集内嵌默认值表达式（AssignmentPattern.right）。
+    fn collect_pattern_default_exprs<'a>(
+        &self, pattern: &'a oxide_parser::BindingPattern<'a>, out: &mut Vec<&'a oxide_parser::Expression<'a>>,
+    ) {
+        use oxide_parser::BindingPattern;
+        match pattern {
+            BindingPattern::AssignmentPattern(ap) => {
+                out.push(&ap.right);
+                self.collect_pattern_default_exprs(&ap.left, out);
+            }
+            BindingPattern::ArrayPattern(ap) => {
+                for elem in &ap.elements {
+                    if let Some(p) = elem {
+                        self.collect_pattern_default_exprs(p, out);
+                    }
+                }
+                if let Some(rest) = &ap.rest {
+                    self.collect_pattern_default_exprs(&rest.argument, out);
+                }
+            }
+            BindingPattern::ObjectPattern(op) => {
+                for prop in &op.properties {
+                    self.collect_pattern_default_exprs(&prop.value, out);
+                }
+                if let Some(rest) = &op.rest {
+                    self.collect_pattern_default_exprs(&rest.argument, out);
+                }
+            }
+            _ => {}
+        }
+    }
+
+
    // ── 闭包捕获分析（AST 级，时序无关）──

    /// 收集函数参数的绑定名（BindingIdentifier 形态）。
@@ -759,8 +792,26 @@ impl Emitter {

        // 闭包捕获分析（AST 级，emit 前确定，时序无关）
        let param_names: Vec<&str> = param_specs.iter().map(|s| s.register_name()).collect();
+        let mut param_defaults: Vec<&oxide_parser::Expression> = Vec::new();
+        for spec in param_specs {
+            match spec {
+                ParamSpec::Identifier { initializer, .. } => {
+                    if let Some(init) = initializer {
+                        param_defaults.push(init);
+                    }
+                }
+                ParamSpec::Pattern { pattern, initializer, .. } => {
+                    if let Some(init) = initializer {
+                        param_defaults.push(init);
+                    }
+                    // 模式内嵌默认值（`[x = expr]` 的 AssignmentPattern.right）也会被
+                    // 内部闭包引用，需纳入捕获分析。
+                    self.collect_pattern_default_exprs(pattern, &mut param_defaults);
+                }
+            }
+        }
        ctx.own_bindings = self.collect_own_binding_names(&param_names, body_stmts);
-        ctx.captured_bindings = self.collect_captured_bindings(body_stmts, &ctx.own_bindings);
+        ctx.captured_bindings = self.collect_captured_bindings(body_stmts, &param_defaults, &ctx.own_bindings);

        // 被捕获的参数也必须建 cell（MAKE_CELL）：否则子函数经 lazy upvalue 路径读
        // 自身寄存器（依赖调用者寄存器残留），vreg 化/RegAlloc 移动寄存器后读到垃圾。
@@ -782,6 +833,7 @@ impl Emitter {
        if matches!(body_context, FunctionBodyContext::Ordinary | FunctionBodyContext::Arrow) {
            ctx.current_upvalue_captures = self.collect_upvalue_names(
                body_stmts,
+                &param_defaults,
                &parent_ctx.captured_bindings,
                &parent_ctx.current_upvalue_captures,
                &ctx.own_bindings,
@@ -888,7 +940,7 @@ impl Emitter {

        // 闭包捕获分析（AST 级，emit 前确定）
        ctx.own_bindings = self.collect_own_binding_names(&[], &program.body);
-        ctx.captured_bindings = self.collect_captured_bindings(&program.body, &ctx.own_bindings);
+        ctx.captured_bindings = self.collect_captured_bindings(&program.body, &[], &ctx.own_bindings);

        // 首个 sub-pass：发函数声明（hoisting），保证任何代码运行前函数对象已就绪。
        for stmt in &program.body {