test: 新增 emit 顶层与模块链接 IR 集成测试网

This commit is contained in:
2026-09-15 21:02:02 +00:00
parent 50cfad32ed
commit 536a97eb5e
2 changed files with 423 additions and 0 deletions
+258
View File
@@ -0,0 +1,258 @@
//! 脚本顶层形态的 IR 级结构断言:parse → `emit_program` 为止,钉死现状 emit 行为。
//!
//! 断言顺序无关:只表达 存在 / 计数 / 名字集 / 键名集。GDI 序言对组与 var 入口
//! MAKE_CELL 组的指令序逐进程轮换,这两组只断言键名集;断言不依赖寄存器号
//! 与 `n_registers` 绝对值。
use std::collections::HashMap;
use std::collections::HashSet;
use oxide_bytecode::opcode::OpCode;
use oxide_emit::{Constant, Emitter};
use oxide_ir::inst::Inst;
use oxide_ir::operand::Operand;
use oxide_ir::IRFunction;
/// 夹具:parse + 脚本口径 emit,失败 panic 带源串。
fn emit_ir(src: &str) -> IRFunction {
let alloc = oxide_parser::Allocator::default();
let program = oxide_parser::parse(&alloc, src).unwrap_or_else(|errs| panic!("parse 失败 {src:?}: {errs:?}"));
Emitter::new()
.emit_program(&program, false, false)
.unwrap_or_else(|e| panic!("emit 失败 {src:?}: {e}"))
}
fn count_op(insts: &[Inst], op: OpCode) -> usize {
insts.iter().filter(|i| i.op == op).count()
}
fn tree_count(ir: &IRFunction, op: OpCode) -> usize {
let mut n = count_op(&ir.insts, op);
for sub in &ir.nested {
n += tree_count(sub, op);
}
n
}
fn pool_strings(ir: &IRFunction) -> HashSet<String> {
ir.constants
.iter()
.filter_map(|c| if let Constant::String(s) = c { Some(s.clone()) } else { None })
.collect()
}
/// 寄存器 → 其被 LOAD_CONST 装入的常量(vreg 号单调不复用,映射唯一)。
fn loaded_consts(ir: &IRFunction) -> HashMap<u32, &Constant> {
let mut m: HashMap<u32, &Constant> = HashMap::new();
for i in &ir.insts {
if i.op == OpCode::LOAD_CONST {
if let (Operand::Reg(r), Operand::Const(c)) = (&i.rd, &i.a) {
m.insert(*r, &ir.constants[*c as usize]);
}
}
}
m
}
/// 指定 opcode 的 b 槽键寄存器回溯其 LOAD_CONST 装入的池字符串名集。
/// 用于 GDI 序言对组(`DEFINE_GLOBAL_PROP_IF_ABSENT`)与顶层 A 侧全局写
/// (`DEFINE_GLOBAL_PROP`)的键名集断言(对序逐进程轮换,只按名集断言)。
fn key_names_of(ir: &IRFunction, op: OpCode) -> HashSet<String> {
let loads = loaded_consts(ir);
let mut keys = HashSet::new();
for i in &ir.insts {
if i.op == op {
if let Operand::Reg(r) = &i.b {
if let Some(Constant::String(s)) = loads.get(r) {
keys.insert(s.clone());
}
}
}
}
keys
}
fn gdi_key_names(ir: &IRFunction) -> HashSet<String> {
key_names_of(ir, OpCode::DEFINE_GLOBAL_PROP_IF_ABSENT)
}
fn define_global_prop_keys(ir: &IRFunction) -> HashSet<String> {
key_names_of(ir, OpCode::DEFINE_GLOBAL_PROP)
}
fn class_flags_scan(f: &IRFunction, base: &mut bool, derived: &mut bool, home: &mut bool) {
if f.is_class_constructor && !f.is_derived_constructor {
*base = true;
}
if f.is_class_constructor && f.is_derived_constructor {
*derived = true;
}
if f.needs_home_object {
*home = true;
}
for n in &f.nested {
class_flags_scan(n, base, derived, home);
}
}
/// 面 1:GDI 序言只覆盖顶层 var 名,lexical(let/const)不进序言。
#[test]
fn gdi_prologue_covers_var_not_lexical() {
let ir = emit_ir("var a = 1; var b = 2; let c = 3;");
assert_eq!(
gdi_key_names(&ir),
["a", "b"].map(String::from).into_iter().collect(),
"GDI 序言键名集应恰为顶层 var 名"
);
}
/// 面 2:顶层函数声明的 GDI 检查、全局绑定与闭包物化形态。
#[test]
fn top_level_function_declaration() {
let ir = emit_ir("function f(n) { return n + 1; }");
assert_eq!(count_op(&ir.insts, OpCode::CAN_DECLARE_GLOBAL_FUNC), 1);
assert_eq!(count_op(&ir.insts, OpCode::DEFINE_GLOBAL_FUNC_BIND), 1);
assert_eq!(count_op(&ir.insts, OpCode::CREATE_CLOSURE), 1);
assert_eq!(ir.nested.len(), 1);
assert!(ir.is_top_level);
let f = &ir.nested[0];
assert_eq!(f.function_name.as_deref(), Some("f"));
assert_eq!(f.function_length, 1);
assert!(!f.is_top_level);
}
/// 面 3:块级函数入口物化——每块函数全流恰 1 个 CREATE_CLOSURE(声明点不重编),
/// 外层 var 写回存在;三常量名(NaN)被守卫:无 GDI 对、无 A 侧全局写。
#[test]
fn block_fn_entry_materialized_once() {
let src = "{ function g() { return 2; } var y = g(); }\n\
label: { function h() { return 3; } }\n\
for (var i = 0; i < 1; i++) { function f() { return i; } }\n\
label3: { function NaN() { return 0; } }";
let ir = emit_ir(src);
assert_eq!(tree_count(&ir, OpCode::CREATE_CLOSURE), 4, "块函数数 = 闭包物化数");
assert_eq!(
gdi_key_names(&ir),
["f", "g", "h", "i", "y"].map(String::from).into_iter().collect(),
"GDI 键名集应含外层 var 与块函数泄漏名,不含三常量名 NaN"
);
assert_eq!(
define_global_prop_keys(&ir),
["f", "g", "h", "i"].map(String::from).into_iter().collect(),
"A 侧全局写键名集不含 NaN(不可写内置守卫)"
);
assert!(count_op(&ir.insts, OpCode::STORE_VAR) >= 1, "外层 var 写回 STORE_VAR 应存在");
}
/// 面 4:三级闭包链的 cell 分配与 upvalue 捕获(cell 下标按名排序分配)。
#[test]
fn closure_captures_cell_and_upvalue() {
let src = "function f() { var x = 1; var s = 0; return function g() { x = x + 1; \
return function h() { s = s + x; return x; }; }; }";
let ir = emit_ir(src);
let f = &ir.nested[0];
assert!(f.cells_needed >= 1);
assert!(count_op(&f.insts, OpCode::MAKE_CELL) >= 1);
let g = &f.nested[0];
let g_cells: HashMap<&str, u8> = g.upvalue_captures.iter().map(|c| (c.name.as_str(), c.cell_idx)).collect();
let mut expected: HashMap<&str, u8> = HashMap::new();
expected.insert("s", 0);
expected.insert("x", 1);
assert_eq!(g_cells, expected, "g 的捕获应按名排序分配 cell 下标");
assert!(count_op(&g.insts, OpCode::LOAD_UPVALUE) >= 1);
assert!(count_op(&g.insts, OpCode::STORE_UPVALUE) >= 1);
let h = &g.nested[0];
let h_names: HashSet<&str> = h.upvalue_captures.iter().map(|c| c.name.as_str()).collect();
let mut expected_names: HashSet<&str> = HashSet::new();
expected_names.insert("s");
expected_names.insert("x");
assert_eq!(h_names, expected_names);
assert!(
h.upvalue_captures.iter().all(|c| c.parent_uv_idx.is_some()),
"链式捕获应经父闭包 upvalue 数组取 cell"
);
}
/// 面 5:对象 rest 展开与数组 holes 的现状形态:REST_OBJECT 的 ext 键是排除键
/// (显式绑定的键名);解构赋值经 for-of 迭代面(FOR_OF_INIT)逐元素绑定,
/// 空洞元素被消费但不绑定;常量池含 Undefined。
#[test]
fn destructured_rest_and_holes() {
let src = "var {a, ...r} = {a: 1, b: 2}; var [x, , z] = [1, 2, 3]; \
var p = [[\"k\", 1]]; for (var [k, v] of p) {}";
let ir = emit_ir(src);
let rest = ir
.insts
.iter()
.find(|i| i.op == OpCode::REST_OBJECT)
.expect("REST_OBJECT 应存在");
assert_eq!(rest.ext.len(), 1);
assert_eq!(
ir.constants[rest.ext[0] as usize],
Constant::String("a".to_string()),
"REST_OBJECT 排除键应为显式绑定的键名"
);
assert!(count_op(&ir.insts, OpCode::FOR_OF_INIT) >= 1);
assert!(ir.constants.iter().any(|c| matches!(c, Constant::Undefined)), "常量池应含 Undefined");
}
/// 面 6:模板字面量的 TEMPLATE_STR 面与 quasis 池键(键经 string_forge 编码,
/// 纯 ASCII 键按值还原即原形态)。
#[test]
fn template_literal_pool_keys() {
let src = "\"use strict\";\nconst t = `a${1}b`;\nconst tag = (s) => s[0];\ntag`x${2}y`";
let ir = emit_ir(src);
assert!(count_op(&ir.insts, OpCode::TEMPLATE_STR) >= 1);
let pool = pool_strings(&ir);
assert!(pool.contains("a") && pool.contains("b"), "quasis 池键按值应含 a/b: {pool:?}");
}
/// 面 7:类构造器/派生标志、home object 与私有名指令面。
#[test]
fn class_nested_flags() {
let src = "class B { m() { return 1; } } class A extends B { #p = 2; static s = 3; \
m() { return this.#p; } }";
let ir = emit_ir(src);
let (mut base, mut derived, mut home) = (false, false, false);
class_flags_scan(&ir, &mut base, &mut derived, &mut home);
assert!(base, "基类构造器应标 is_class_constructor");
assert!(derived, "extends 构造器应标 is_derived_constructor");
assert!(home, "原型方法应标 needs_home_object");
assert!(tree_count(&ir, OpCode::INIT_PRIVATE) >= 1, "私有字段初始化面应存在");
assert!(tree_count(&ir, OpCode::GET_PRIVATE) >= 1, "私有成员读面应存在");
}
/// 面 8:严格模式标志在函数体内与脚本顶层的继承口径。
#[test]
fn strict_flag_inherited() {
let ir = emit_ir("function f() { \"use strict\"; } function g() { return 1; }");
let by_name = |n: &str| -> &IRFunction {
ir.nested
.iter()
.find(|f| f.function_name.as_deref() == Some(n))
.unwrap_or_else(|| panic!("nested 缺 {n}"))
};
assert!(by_name("f").is_strict, "函数体内 use strict 应置位");
assert!(!by_name("g").is_strict, "无 directive 函数不应继承严格");
let top = emit_ir("\"use strict\"; var x = 1;");
assert!(top.is_strict, "脚本顶层 use strict 应置位");
}
/// 面 9:source_encoded 口径下正则源池键为 marker 形态(与静态口径的明文键不同形)。
#[test]
fn source_encoded_regex_pool_key() {
let src = "var r = /\\ud800/g;";
let encode_ir = {
let alloc = oxide_parser::Allocator::default();
let program = oxide_parser::parse(&alloc, src).unwrap();
Emitter::new()
.with_source_encoded(true)
.emit_program(&program, false, false)
.unwrap()
};
let static_ir = emit_ir(src);
let enc = pool_strings(&encode_ir);
assert!(enc.contains("\u{FFFD}d800"), "编码源池键应为 FFFD+hex marker 形态: {enc:?}");
assert!(!enc.contains("\\ud800"));
assert!(pool_strings(&static_ir).contains("\\ud800"), "静态源池键为明文转义文本");
}
@@ -0,0 +1,165 @@
//! 模块链接面的 IR 级断言:内存版 `ModuleSourceLoader` +
//! `emit_program_module`,钉死 import prelude / export 就地注册 / star 再导出 /
//! 循环导入编译期报错 / 数据模块的现状行为。
//!
//! 路径口径:加载器 `path` 回填表键原样字符串,与入口 `module_path` 入参逐字符
//! 一致(虚拟路径 canonicalize 失败保留原路径),自导入身份比较与循环检测依赖
//! 此口径。断言不依赖寄存器号与指令序。
use std::collections::HashMap;
use std::collections::HashSet;
use oxide_bytecode::opcode::OpCode;
use oxide_emit::module::{ModuleKind, ModuleSourceLoader, ResolvedModule};
use oxide_emit::{Constant, Emitter};
use oxide_ir::IRFunction;
/// 内存版加载器:specifier 精确查表;`path` 原样回填表键。
/// `resolved` 记录每次 resolve 的 (specifier, 属性对),供链接期观察。
struct MemLoader {
table: HashMap<String, (ModuleKind, String)>,
resolved: Vec<(String, Vec<(String, String)>)>,
}
impl ModuleSourceLoader for MemLoader {
fn resolve(
&mut self, _base_dir: &str, specifier: &str, attributes: &[(&str, &str)],
) -> Result<ResolvedModule, String> {
self.resolved.push((
specifier.to_string(),
attributes.iter().map(|(k, v)| (k.to_string(), v.to_string())).collect(),
));
self.table
.get(specifier)
.cloned()
.map(|(kind, source)| ResolvedModule {
source,
path: specifier.to_string(),
kind,
})
.ok_or_else(|| format!("MemLoader 无 {specifier}"))
}
}
/// 夹具:parse_module + emit_program_module,返回 IR 与加载器。
fn emit_module(src: &str, path: &str, deps: &[(&str, ModuleKind, &str)]) -> Result<(IRFunction, MemLoader), String> {
let alloc = oxide_parser::Allocator::default();
let program =
oxide_parser::parse_module(&alloc, src).map_err(|errs| format!("parse_module 失败 {src:?}: {errs:?}"))?;
let mut loader = MemLoader {
table: deps.iter().map(|(s, k, t)| (s.to_string(), (*k, t.to_string()))).collect(),
resolved: Vec::new(),
};
let ir = Emitter::new().emit_program_module(&program, path, &mut loader)?;
Ok((ir, loader))
}
fn builtin_names(ir: &IRFunction) -> HashSet<String> {
ir.builtin_reg_map.iter().map(|(n, _)| n.clone()).collect()
}
fn builtin_reg(ir: &IRFunction, name: &str) -> Option<u32> {
ir.builtin_reg_map.iter().find(|(n, _)| n == name).map(|(_, r)| *r)
}
/// 以某 built-in 为 callee(rd 槽)的 CALL_NATIVE 计数。
fn native_calls_to(ir: &IRFunction, name: &str) -> usize {
let Some(reg) = builtin_reg(ir, name) else {
return 0;
};
ir.insts
.iter()
.filter(|i| i.op == OpCode::CALL_NATIVE && matches!(i.rd, oxide_ir::operand::Operand::Reg(r) if r == reg))
.count()
}
fn pool_strings(ir: &IRFunction) -> HashSet<String> {
ir.constants
.iter()
.filter_map(|c| if let Constant::String(s) = c { Some(s.clone()) } else { None })
.collect()
}
/// 面 1:import 绑定 prelude 经 `__moduleLinkGet` 链接,依赖模块编入子模块树。
#[test]
fn import_binding_prelude() {
let (ir, _) = emit_module(
"import { x } from \"./dep.js\"; var y = x;",
"./entry.js",
&[("./dep.js", ModuleKind::Js, "export var x = 42;")],
)
.expect("模块编译应成功");
assert!(builtin_names(&ir).contains("__moduleLinkGet"), "prelude 应链接 __moduleLinkGet");
assert!(native_calls_to(&ir, "__moduleLinkGet") >= 1, "命名导入应发 __moduleLinkGet 调用");
assert_eq!(ir.nested.len(), 1, "依赖模块应编入子模块树");
let dep = &ir.nested[0];
assert!(!dep.insts.is_empty());
assert!(builtin_names(dep).contains("__moduleSet"), "依赖的 export 应就地注册");
}
/// 面 2:export 声明就地注册导出值到命名空间(`__moduleSet` 逐名成对)。
#[test]
fn export_in_place_register() {
let (ir, _) = emit_module("export function f() {} export const c = 2;", "./entry.js", &[]).expect("模块编译应成功");
assert!(builtin_names(&ir).contains("__moduleSet"));
assert_eq!(native_calls_to(&ir, "__moduleSet"), 2, "f 与 c 各注册一次");
let pool = pool_strings(&ir);
assert!(pool.contains("f") && pool.contains("c"), "导出名应成对入池: {pool:?}");
}
/// 面 3:`export * from` 匿名 star 再导出经 `__moduleStar` 整体转发(现状口径)。
#[test]
fn export_all_from() {
let (ir, _) = emit_module(
"export * from \"./dep.js\";",
"./entry.js",
&[("./dep.js", ModuleKind::Js, "export var z = 1;")],
)
.expect("star 再导出应编译通过");
assert!(builtin_names(&ir).contains("__moduleStar"), "匿名 star 应走 __moduleStar");
assert_eq!(ir.nested.len(), 1);
}
/// 面 4:循环导入在编译期报错,错误文本即循环检测点产物。
#[test]
fn circular_import_compile_error() {
let src_a = "import { y } from \"./b.js\"; export var x = 1;";
let result = emit_module(
src_a,
"./a.js",
&[
("./a.js", ModuleKind::Js, src_a),
("./b.js", ModuleKind::Js, "import { x } from \"./a.js\"; export var y = 2;"),
],
);
assert_eq!(
result.err().as_deref(),
Some("circular module import not supported: ./a.js"),
"Err 应来自循环检测点(自导入身份比较命中)"
);
}
/// 面 5:json 数据模块经 `__moduleData` 物化,数据源文本入子模块常量池。
#[test]
fn json_data_module() {
let json = r#"{"k":1}"#;
let (ir, loader) = emit_module(
"import data from \"./d.json\" assert { type: \"json\" }; data;",
"./entry.js",
&[("./d.json", ModuleKind::Json, json)],
)
.expect("数据模块编译应成功");
// import 属性应原样传给加载器。
let attrs = loader
.resolved
.iter()
.find(|(s, _)| s == "./d.json")
.map(|(_, a)| a.clone())
.expect("d.json 应被 resolve");
assert_eq!(attrs, vec![("type".to_string(), "json".to_string())]);
assert_eq!(ir.nested.len(), 1);
let data = &ir.nested[0];
assert!(builtin_names(data).contains("__moduleData"), "数据模块应链接 __moduleData");
let pool = pool_strings(data);
assert!(pool.contains("json") && pool.contains(json), "json 源文本应入池: {pool:?}");
}