阶段 A 步骤 3：MachineConfig 加载层（toml++ + 强校验 5 项）。
- MachineConfig.h/cpp：解析 machine.toml → meta/types/ops/fbs 表 - 强校验：全属性必填、base 命中内建基元、range min≤max、tag 契约（BOOL=0/INT=1/TIME=2 唯一）、 opcode 唯一 0..255、类别-格式互锁（instance↔CAL）、params 数量与 format 一致、 fb.opcode 匹配 instance op、fb 字段类型命中 type 表 - 基元内置表（bit/int8..uint64/float32/64，宽度/符号/浮点） - machine_test：38 断言（正例仓库表 + 9 类负例 + 最小配置），ctest 12/12
This commit is contained in:
@@ -13,7 +13,8 @@ add_library(compiler STATIC
|
||||
./src/Parser.cpp
|
||||
./src/Linker.cpp
|
||||
./src/Typecheck.cpp
|
||||
./src/Codegen.cpp)
|
||||
./src/Codegen.cpp
|
||||
./src/MachineConfig.cpp)
|
||||
|
||||
target_include_directories(compiler PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include)
|
||||
target_include_directories(compiler PRIVATE ${CMAKE_SOURCE_DIR}/third_party/tomlplusplus/include)
|
||||
|
||||
@@ -0,0 +1,92 @@
|
||||
/**
|
||||
* @file MachineConfig.h
|
||||
* @brief 机器定义（machine.toml）加载与强校验
|
||||
* @author
|
||||
* @date 2026-08-21
|
||||
*
|
||||
* @details 见 Doc/isa/指令配置.md：
|
||||
* - 配置只给编译器（toml++ 运行时加载）；vm 不读配置（内建全指令集）
|
||||
* - 强校验 5 项：字段必填 / type.base 命中内建基元·range 合法·tag 契约 /
|
||||
* op opcode 唯一·类别-格式互锁 / fb.opcode 与 op 一致·字段类型命中 type 表
|
||||
* - 基元内置（bit/int8..uint64/float32/float64，元数据在代码）
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
|
||||
namespace compiler {
|
||||
|
||||
// 内建基元（编译器内建，C 语义）
|
||||
struct PrimType {
|
||||
const char* name;
|
||||
uint32_t width; // 字节
|
||||
bool is_signed;
|
||||
bool is_float;
|
||||
};
|
||||
|
||||
// 配置类型 = 基元别名（+ 值域约束 + .stb 契约 tag）
|
||||
struct ConfigType {
|
||||
std::string name;
|
||||
std::string base;
|
||||
bool has_range = false;
|
||||
int64_t range_min = 0;
|
||||
int64_t range_max = 0;
|
||||
uint32_t tag = 0;
|
||||
};
|
||||
|
||||
// 配置指令
|
||||
struct ConfigOp {
|
||||
std::string name;
|
||||
uint32_t opcode = 0;
|
||||
bool is_instance = false; // class: instance / plain
|
||||
std::string format; // RR/RRR/IMM/SLOT/JMP/JC/CALL/CAL/NONE
|
||||
std::vector<std::string> params;
|
||||
bool enabled = false;
|
||||
};
|
||||
|
||||
struct ConfigFbField {
|
||||
std::string name;
|
||||
std::string type; // 配置类型名（BOOL/INT/TIME）
|
||||
};
|
||||
|
||||
struct ConfigFb {
|
||||
std::string name;
|
||||
uint32_t opcode = 0;
|
||||
std::vector<ConfigFbField> fields;
|
||||
};
|
||||
|
||||
class MachineConfig {
|
||||
public:
|
||||
// 加载 + 强校验；失败返回 false 并写 err（前缀 "machine error"）
|
||||
bool load(const std::string& path, std::string* err);
|
||||
|
||||
bool ok() const { return ok_; }
|
||||
const std::string& model_name() const { return model_name_; }
|
||||
uint32_t version() const { return version_; }
|
||||
|
||||
const std::vector<ConfigType>& types() const { return types_; }
|
||||
const std::vector<ConfigOp>& ops() const { return ops_; }
|
||||
const std::vector<ConfigFb>& fbs() const { return fbs_; }
|
||||
|
||||
const ConfigType* find_type(const std::string& name) const;
|
||||
const ConfigOp* find_op(const std::string& name) const;
|
||||
const ConfigOp* find_op_by_code(uint32_t opcode) const;
|
||||
bool op_enabled(uint32_t opcode) const;
|
||||
const ConfigFb* find_fb(const std::string& name) const;
|
||||
|
||||
// 内建基元表
|
||||
static const std::vector<PrimType>& prims();
|
||||
static const PrimType* find_prim(const std::string& name);
|
||||
|
||||
private:
|
||||
bool ok_ = false;
|
||||
std::string model_name_;
|
||||
uint32_t version_ = 0;
|
||||
std::vector<ConfigType> types_;
|
||||
std::vector<ConfigOp> ops_;
|
||||
std::vector<ConfigFb> fbs_;
|
||||
};
|
||||
}
|
||||
@@ -0,0 +1,396 @@
|
||||
/**
|
||||
* @file MachineConfig.cpp
|
||||
* @brief 机器定义（machine.toml）加载与强校验
|
||||
* @author
|
||||
* @date 2026-08-21
|
||||
*
|
||||
* @details 校验项（见 Doc/isa/指令配置.md 强校验清单 1~5）：
|
||||
* 1. TOML 语法与字段完整性（全属性必填）
|
||||
* 2. type.base 命中内建基元；range 合法（min ≤ max）
|
||||
* 3. tag 与 .stb 常量表契约一致（BOOL=0、INT=1、TIME=2，唯一）
|
||||
* 4. op.opcode 唯一、0..255；class/format 合法；类别-格式互锁（INSTANCE ↔ CAL）；
|
||||
* params 数量与 format 一致
|
||||
* 5. fb.opcode 与 op 行（instance 类）一致；字段类型命中配置类型表；字段名唯一
|
||||
*/
|
||||
|
||||
#include "compiler/MachineConfig.h"
|
||||
|
||||
#include <toml++/toml.hpp>
|
||||
|
||||
#include <map>
|
||||
#include <string>
|
||||
|
||||
namespace compiler {
|
||||
|
||||
const std::vector<PrimType>& MachineConfig::prims() {
|
||||
static const std::vector<PrimType> kPrims = {
|
||||
{"bit", 1, false, false},
|
||||
{"int8", 1, true, false},
|
||||
{"int16", 2, true, false},
|
||||
{"int32", 4, true, false},
|
||||
{"int64", 8, true, false},
|
||||
{"uint8", 1, false, false},
|
||||
{"uint16", 2, false, false},
|
||||
{"uint32", 4, false, false},
|
||||
{"uint64", 8, false, false},
|
||||
{"float32", 4, true, true},
|
||||
{"float64", 8, true, true},
|
||||
};
|
||||
return kPrims;
|
||||
}
|
||||
|
||||
const PrimType* MachineConfig::find_prim(const std::string& name) {
|
||||
for (const PrimType& p : prims()) {
|
||||
if (p.name == name) {
|
||||
return &p;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
namespace {
|
||||
|
||||
// 稳定前缀
|
||||
std::string fail(std::string* err, const std::string& msg) {
|
||||
if (err) {
|
||||
*err = "machine error: " + msg;
|
||||
}
|
||||
return msg;
|
||||
}
|
||||
|
||||
// format → 参数数量（解析时校验）
|
||||
int params_of(const std::string& format) {
|
||||
if (format == "RR") return 2;
|
||||
if (format == "RRR") return 3;
|
||||
if (format == "IMM") return 2;
|
||||
if (format == "SLOT") return 2;
|
||||
if (format == "JMP") return 1;
|
||||
if (format == "JC") return 2;
|
||||
if (format == "CALL") return 1;
|
||||
if (format == "CAL") return 1;
|
||||
if (format == "NONE") return 0;
|
||||
return -1;
|
||||
}
|
||||
|
||||
bool is_known_format(const std::string& f) {
|
||||
return params_of(f) >= 0;
|
||||
}
|
||||
|
||||
bool is_known_tag(uint32_t tag) {
|
||||
return tag <= 2; // .stb 常量表契约：0=BOOL 1=INT 2=TIME
|
||||
}
|
||||
|
||||
// 必填字符串
|
||||
bool req_string(const toml::table& t, const char* what, const std::string& key,
|
||||
std::string* out, std::string* err) {
|
||||
const auto nv = t[key];
|
||||
if (!nv || !nv.is_string()) {
|
||||
fail(err, std::string("missing field '") + key + "' in " + what);
|
||||
return false;
|
||||
}
|
||||
*out = nv.value_or(std::string());
|
||||
return true;
|
||||
}
|
||||
|
||||
// 必填整数
|
||||
bool req_int(const toml::table& t, const char* what, const std::string& key,
|
||||
int64_t* out, std::string* err) {
|
||||
const auto nv = t[key];
|
||||
if (!nv || !nv.is_integer()) {
|
||||
fail(err, std::string("missing field '") + key + "' in " + what);
|
||||
return false;
|
||||
}
|
||||
*out = nv.value_or<int64_t>(0);
|
||||
return true;
|
||||
}
|
||||
|
||||
// 必填布尔
|
||||
bool req_bool(const toml::table& t, const char* what, const std::string& key,
|
||||
bool* out, std::string* err) {
|
||||
const auto nv = t[key];
|
||||
if (!nv || !nv.is_boolean()) {
|
||||
fail(err, std::string("missing field '") + key + "' in " + what);
|
||||
return false;
|
||||
}
|
||||
*out = nv.value_or<bool>(false);
|
||||
return true;
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
bool MachineConfig::load(const std::string& path, std::string* err) {
|
||||
ok_ = false;
|
||||
model_name_.clear();
|
||||
version_ = 0;
|
||||
types_.clear();
|
||||
ops_.clear();
|
||||
fbs_.clear();
|
||||
|
||||
toml::table root;
|
||||
try {
|
||||
root = toml::parse_file(path);
|
||||
} catch (const toml::parse_error& e) {
|
||||
fail(err, std::string("parse error: ") + std::string(e.description()));
|
||||
return false;
|
||||
}
|
||||
|
||||
// ---- meta ----
|
||||
const toml::table* meta = root["meta"].as_table();
|
||||
if (!meta) {
|
||||
fail(err, "missing table 'meta'");
|
||||
return false;
|
||||
}
|
||||
if (!req_string(*meta, "[meta]", "name", &model_name_, err)) return false;
|
||||
int64_t ver = 0;
|
||||
if (!req_int(*meta, "[meta]", "version", &ver, err)) return false;
|
||||
if (ver < 0 || ver > 0xFFFFFFFFLL) {
|
||||
fail(err, "invalid version in [meta]");
|
||||
return false;
|
||||
}
|
||||
version_ = static_cast<uint32_t>(ver);
|
||||
|
||||
// ---- type ----
|
||||
{
|
||||
const toml::array* arr = root["type"].as_array();
|
||||
if (!arr || arr->empty()) {
|
||||
fail(err, "missing 'type' table");
|
||||
return false;
|
||||
}
|
||||
std::map<uint32_t, std::string> tag_owner; // tag → 类型名
|
||||
for (const auto& el : *arr) {
|
||||
const toml::table* t = el.as_table();
|
||||
if (!t) {
|
||||
fail(err, "bad entry in type");
|
||||
return false;
|
||||
}
|
||||
ConfigType ct;
|
||||
if (!req_string(*t, "type", "name", &ct.name, err)) return false;
|
||||
if (!req_string(*t, "type", "base", &ct.base, err)) return false;
|
||||
if (find_prim(ct.base) == nullptr) {
|
||||
fail(err, "type '" + ct.name + "': unknown base '" + ct.base + "'");
|
||||
return false;
|
||||
}
|
||||
int64_t tag = 0;
|
||||
if (!req_int(*t, "type", "tag", &tag, err)) return false;
|
||||
if (tag < 0 || !is_known_tag(static_cast<uint32_t>(tag))) {
|
||||
fail(err, "type '" + ct.name + "': tag out of contract (0..2)");
|
||||
return false;
|
||||
}
|
||||
ct.tag = static_cast<uint32_t>(tag);
|
||||
if (tag_owner.count(ct.tag)) {
|
||||
fail(err, "duplicate tag " + std::to_string(ct.tag));
|
||||
return false;
|
||||
}
|
||||
tag_owner[ct.tag] = ct.name;
|
||||
if (t->contains("range")) {
|
||||
const toml::array* r = (*t)["range"].as_array();
|
||||
if (!r || r->size() != 2) {
|
||||
fail(err, "type '" + ct.name + "': range must be [min, max]");
|
||||
return false;
|
||||
}
|
||||
ct.has_range = true;
|
||||
ct.range_min = (*r)[0].value_or<int64_t>(0);
|
||||
ct.range_max = (*r)[1].value_or<int64_t>(0);
|
||||
if (ct.range_min > ct.range_max) {
|
||||
fail(err, "type '" + ct.name + "': range min > max");
|
||||
return false;
|
||||
}
|
||||
}
|
||||
types_.push_back(ct);
|
||||
}
|
||||
// .stb 契约：BOOL/INT/TIME 必须存在且 tag 为 0/1/2
|
||||
const std::string need[3] = {"BOOL", "INT", "TIME"};
|
||||
for (uint32_t i = 0; i < 3; ++i) {
|
||||
const auto it = tag_owner.find(i);
|
||||
if (it == tag_owner.end() || it->second != need[i]) {
|
||||
fail(err, std::string("type contract broken: tag ") + std::to_string(i) +
|
||||
" must be " + need[i]);
|
||||
return false;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// ---- op ----
|
||||
{
|
||||
const toml::array* arr = root["op"].as_array();
|
||||
if (!arr || arr->empty()) {
|
||||
fail(err, "missing 'op' table");
|
||||
return false;
|
||||
}
|
||||
std::map<uint32_t, std::string> code_owner;
|
||||
std::map<std::string, uint32_t> name_owner;
|
||||
for (const auto& el : *arr) {
|
||||
const toml::table* t = el.as_table();
|
||||
if (!t) {
|
||||
fail(err, "bad entry in op");
|
||||
return false;
|
||||
}
|
||||
ConfigOp op;
|
||||
if (!req_string(*t, "op", "name", &op.name, err)) return false;
|
||||
int64_t code = 0;
|
||||
if (!req_int(*t, "op", "opcode", &code, err)) return false;
|
||||
if (code < 0 || code > 255) {
|
||||
fail(err, "op '" + op.name + "': opcode out of range");
|
||||
return false;
|
||||
}
|
||||
op.opcode = static_cast<uint32_t>(code);
|
||||
if (code_owner.count(op.opcode)) {
|
||||
fail(err, "duplicate opcode " + std::to_string(op.opcode));
|
||||
return false;
|
||||
}
|
||||
code_owner[op.opcode] = op.name;
|
||||
if (name_owner.count(op.name)) {
|
||||
fail(err, "duplicate op name '" + op.name + "'");
|
||||
return false;
|
||||
}
|
||||
name_owner[op.name] = op.opcode;
|
||||
std::string cls;
|
||||
if (!req_string(*t, "op", "class", &cls, err)) return false;
|
||||
if (cls == "instance") {
|
||||
op.is_instance = true;
|
||||
} else if (cls == "plain") {
|
||||
op.is_instance = false;
|
||||
} else {
|
||||
fail(err, "op '" + op.name + "': bad class '" + cls + "'");
|
||||
return false;
|
||||
}
|
||||
if (!req_string(*t, "op", "format", &op.format, err)) return false;
|
||||
if (!is_known_format(op.format)) {
|
||||
fail(err, "op '" + op.name + "': bad format '" + op.format + "'");
|
||||
return false;
|
||||
}
|
||||
// 类别-格式互锁：INSTANCE ↔ CAL
|
||||
const bool is_cal = op.format == "CAL";
|
||||
if (op.is_instance != is_cal) {
|
||||
fail(err, "op '" + op.name + "': class-format mismatch (instance <-> CAL)");
|
||||
return false;
|
||||
}
|
||||
if (!req_bool(*t, "op", "enabled", &op.enabled, err)) return false;
|
||||
const auto pv = (*t)["params"];
|
||||
if (!pv || !pv.is_array()) {
|
||||
fail(err, "op '" + op.name + "': missing 'params'");
|
||||
return false;
|
||||
}
|
||||
const int want = params_of(op.format);
|
||||
for (const auto& p : *pv.as_array()) {
|
||||
if (!p.is_string()) {
|
||||
fail(err, "op '" + op.name + "': params must be strings");
|
||||
return false;
|
||||
}
|
||||
op.params.push_back(p.value_or(std::string()));
|
||||
}
|
||||
if (static_cast<int>(op.params.size()) != want) {
|
||||
fail(err, "op '" + op.name + "': params count " +
|
||||
std::to_string(op.params.size()) + " != format expects " +
|
||||
std::to_string(want));
|
||||
return false;
|
||||
}
|
||||
ops_.push_back(op);
|
||||
}
|
||||
}
|
||||
|
||||
// ---- fb ----
|
||||
{
|
||||
const toml::array* arr = root["fb"].as_array();
|
||||
if (!arr || arr->empty()) {
|
||||
fail(err, "missing 'fb' table");
|
||||
return false;
|
||||
}
|
||||
for (const auto& el : *arr) {
|
||||
const toml::table* t = el.as_table();
|
||||
if (!t) {
|
||||
fail(err, "bad entry in fb");
|
||||
return false;
|
||||
}
|
||||
ConfigFb fb;
|
||||
if (!req_string(*t, "fb", "name", &fb.name, err)) return false;
|
||||
int64_t code = 0;
|
||||
if (!req_int(*t, "fb", "opcode", &code, err)) return false;
|
||||
fb.opcode = static_cast<uint32_t>(code);
|
||||
// opcode 必须命中 instance 类 op 行
|
||||
const ConfigOp* op = find_op_by_code(fb.opcode);
|
||||
if (op == nullptr || !op->is_instance) {
|
||||
fail(err, "fb '" + fb.name + "': opcode " + std::to_string(code) +
|
||||
" must match an instance op");
|
||||
return false;
|
||||
}
|
||||
const auto fv = (*t)["fields"];
|
||||
if (!fv || !fv.is_array()) {
|
||||
fail(err, "fb '" + fb.name + "': missing 'fields'");
|
||||
return false;
|
||||
}
|
||||
std::map<std::string, bool> fnames;
|
||||
for (const auto& f : *fv.as_array()) {
|
||||
const toml::array* pair = f.as_array();
|
||||
if (!pair || pair->size() != 2) {
|
||||
fail(err, "fb '" + fb.name + "': field must be [name, type]");
|
||||
return false;
|
||||
}
|
||||
ConfigFbField field;
|
||||
field.name = (*pair)[0].value_or(std::string());
|
||||
field.type = (*pair)[1].value_or(std::string());
|
||||
if (field.name.empty() || field.type.empty()) {
|
||||
fail(err, "fb '" + fb.name + "': empty field name/type");
|
||||
return false;
|
||||
}
|
||||
if (fnames.count(field.name)) {
|
||||
fail(err, "fb '" + fb.name + "': duplicate field '" + field.name + "'");
|
||||
return false;
|
||||
}
|
||||
fnames[field.name] = true;
|
||||
if (find_type(field.type) == nullptr) {
|
||||
fail(err, "fb '" + fb.name + "': unknown field type '" + field.type + "'");
|
||||
return false;
|
||||
}
|
||||
fb.fields.push_back(field);
|
||||
}
|
||||
fbs_.push_back(fb);
|
||||
}
|
||||
}
|
||||
|
||||
ok_ = true;
|
||||
return true;
|
||||
}
|
||||
|
||||
const ConfigType* MachineConfig::find_type(const std::string& name) const {
|
||||
for (const ConfigType& t : types_) {
|
||||
if (t.name == name) {
|
||||
return &t;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
const ConfigOp* MachineConfig::find_op(const std::string& name) const {
|
||||
for (const ConfigOp& o : ops_) {
|
||||
if (o.name == name) {
|
||||
return &o;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
const ConfigOp* MachineConfig::find_op_by_code(uint32_t opcode) const {
|
||||
for (const ConfigOp& o : ops_) {
|
||||
if (o.opcode == opcode) {
|
||||
return &o;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
bool MachineConfig::op_enabled(uint32_t opcode) const {
|
||||
const ConfigOp* op = find_op_by_code(opcode);
|
||||
return op != nullptr && op->enabled;
|
||||
}
|
||||
|
||||
const ConfigFb* MachineConfig::find_fb(const std::string& name) const {
|
||||
for (const ConfigFb& f : fbs_) {
|
||||
if (f.name == name) {
|
||||
return &f;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
} // namespace compiler
|
||||
Reference in New Issue
Block a user