阶段 B 步骤 5:类型元数据配置化(TypeInfo)。

- TypeInfo.h/cpp:TypeMeta(配置别名 + 内建基元解析),查询大小写不敏感;
  提供 width/is_signed/is_float/tag/has_range/value_in_range
- Codegen:layout_data 初值按 type_meta 的基元宽度/浮点写(删除手写宽度分支);
  const_id 的 tag 改查 type_meta
- machine_test:+10 断言(BOOL 1B/range[0,1]/tag0、INT 2B 有符号、TIME 8B、大小写不敏感、未定义空)
- 验收:line1 224 字节 hash 0xc3fe4ae74ad45900 不变、ctest 12/12
This commit is contained in:
2026-08-21 22:16:48 +08:00
parent 447a8d0119
commit e187ed7b03
5 changed files with 127 additions and 11 deletions
+2 -1
View File
@@ -17,7 +17,8 @@ add_library(compiler STATIC
./src/Codegen.cpp ./src/Codegen.cpp
./src/MachineConfig.cpp ./src/MachineConfig.cpp
./src/Stb.cpp ./src/Stb.cpp
./src/Codec.cpp) ./src/Codec.cpp
./src/TypeInfo.cpp)
target_include_directories(compiler PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include) target_include_directories(compiler PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include)
target_include_directories(compiler PRIVATE ${CMAKE_SOURCE_DIR}/third_party/tomlplusplus/include) target_include_directories(compiler PRIVATE ${CMAKE_SOURCE_DIR}/third_party/tomlplusplus/include)
+41
View File
@@ -0,0 +1,41 @@
/**
* @file TypeInfo.h
* @brief 语言类型元数据(machine.toml 别名 + 内建基元解析)
* @author
* @date 2026-08-21
*
* @details 类型定义数据化(配置),运算/类型检查语义代码化:
* - 元数据(宽度/符号/浮点/base/range/tag)全部来自 machine.toml + 内建基元表
* - 查询大小写不敏感(配置名大写,Linker type_name 小写)
*/
#pragma once
#include <cstdint>
#include <string>
#include "compiler/MachineConfig.h"
namespace compiler {
// 语言类型元数据
struct TypeMeta {
const ConfigType* config = nullptr; // 配置类型行(别名 + range + tag
const PrimType* prim = nullptr; // 内建基元(宽度/符号/浮点)
uint32_t width() const { return prim ? prim->width : 0; }
bool is_signed() const { return prim ? prim->is_signed : false; }
bool is_float() const { return prim ? prim->is_float : false; }
uint32_t tag() const { return config ? config->tag : 0; }
bool has_range() const { return config ? config->has_range : false; }
bool value_in_range(int64_t v) const {
return !config || !config->has_range ||
(v >= config->range_min && v <= config->range_max);
}
explicit operator bool() const { return config != nullptr && prim != nullptr; }
};
// 按语言类型名查元数据(大小写不敏感;未找到返回空 TypeMeta)
TypeMeta type_meta(const MachineConfig& cfg, const std::string& name);
}
+10 -10
View File
@@ -42,6 +42,7 @@
#include "compiler/Codec.h" #include "compiler/Codec.h"
#include "compiler/MachineConfig.h" #include "compiler/MachineConfig.h"
#include "compiler/Stb.h" #include "compiler/Stb.h"
#include "compiler/TypeInfo.h"
namespace compiler { namespace compiler {
namespace { namespace {
@@ -589,20 +590,19 @@ namespace {
const int64_t v = has_init ? init : 0; const int64_t v = has_init ? init : 0;
const size_t off = static_cast<size_t>(s.address) * 8; const size_t off = static_cast<size_t>(s.address) * 8;
data_.resize(off + 8, 0); data_.resize(off + 8, 0);
// 初值按基元宽度写(类型元数据来自配置别名 → base // 初值按类型元数据写(配置别名 → 基元宽度/浮点
const ConfigType* ct = cfg_.find_type(uppercase_of(s.type_name)); const TypeMeta tm = type_meta(cfg_, s.type_name);
if (ct == nullptr) { if (!tm) {
return fail("no type in machine.toml for '" + s.type_name + "'"); return fail("no type in machine.toml for '" + s.type_name + "'");
} }
const PrimType* prim = MachineConfig::find_prim(ct->base); if (tm.is_float()) {
if (prim == nullptr) { return fail("float initializer not supported yet");
return fail("no prim '" + ct->base + "' for type '" + ct->name + "'");
} }
if (prim->width == 2 && !prim->is_float) { if (tm.width() == 2) {
const uint16_t iv = static_cast<uint16_t>(v); const uint16_t iv = static_cast<uint16_t>(v);
data_[off] = static_cast<uint8_t>(iv & 0xFFu); data_[off] = static_cast<uint8_t>(iv & 0xFFu);
data_[off + 1] = static_cast<uint8_t>((iv >> 8) & 0xFFu); data_[off + 1] = static_cast<uint8_t>((iv >> 8) & 0xFFu);
} else if (prim->width >= 8 && !prim->is_float) { } else if (tm.width() == 8) {
const uint64_t tv = static_cast<uint64_t>(v); const uint64_t tv = static_cast<uint64_t>(v);
for (int i = 0; i < 8; ++i) { for (int i = 0; i < 8; ++i) {
data_[off + i] = static_cast<uint8_t>((tv >> (8 * i)) & 0xFFu); data_[off + i] = static_cast<uint8_t>((tv >> (8 * i)) & 0xFFu);
@@ -672,8 +672,8 @@ namespace {
* @return const_idu16 * @return const_idu16
*/ */
uint16_t const_id(const char* type_name, int64_t value) { uint16_t const_id(const char* type_name, int64_t value) {
const ConfigType* ct = cfg_.find_type(type_name); const TypeMeta tm = type_meta(cfg_, type_name);
const uint32_t tag = ct ? ct->tag : 0; const uint32_t tag = tm ? tm.tag() : 0;
for (size_t i = 0; i < consts_.size(); ++i) { for (size_t i = 0; i < consts_.size(); ++i) {
if (consts_[i].tag == tag && consts_[i].value == static_cast<uint64_t>(value)) { if (consts_[i].tag == tag && consts_[i].value == static_cast<uint64_t>(value)) {
return static_cast<uint16_t>(i); return static_cast<uint16_t>(i);
+41
View File
@@ -0,0 +1,41 @@
/**
* @file TypeInfo.cpp
* @brief 语言类型元数据(machine.toml 别名 + 内建基元解析)
* @author
* @date 2026-08-21
*/
#include "compiler/TypeInfo.h"
#include <cctype>
namespace compiler {
namespace {
std::string lower(const std::string& s) {
std::string out = s;
for (char& ch : out) {
ch = static_cast<char>(std::tolower(static_cast<unsigned char>(ch)));
}
return out;
}
} // namespace
TypeMeta type_meta(const MachineConfig& cfg, const std::string& name) {
TypeMeta m;
const std::string key = lower(name);
for (const ConfigType& t : cfg.types()) {
if (lower(t.name) == key) {
m.config = &t;
break;
}
}
if (m.config != nullptr) {
m.prim = MachineConfig::find_prim(m.config->base);
}
return m;
}
} // namespace compiler
+33
View File
@@ -10,6 +10,7 @@
#include <string> #include <string>
#include "compiler/MachineConfig.h" #include "compiler/MachineConfig.h"
#include "compiler/TypeInfo.h"
#ifndef REPO_ROOT #ifndef REPO_ROOT
#define REPO_ROOT "." #define REPO_ROOT "."
@@ -252,10 +253,42 @@ static bool test_positive_min() {
return true; return true;
} }
// ---- 4. 类型元数据(TypeInfo----
static bool test_type_meta() {
compiler::MachineConfig cfg;
std::string err;
CHECK(cfg.load(std::string(REPO_ROOT) + "/compiler/machine.toml", &err));
// BOOLbase=uint8 → 1 字节、无符号、非浮点、tag 0、range [0,1]
const compiler::TypeMeta b = compiler::type_meta(cfg, "BOOL");
CHECK(b && b.width() == 1 && !b.is_signed() && !b.is_float() && b.tag() == 0);
CHECK(b.has_range() && b.value_in_range(0) && b.value_in_range(1));
CHECK(!b.value_in_range(2) && !b.value_in_range(-1));
// 大小写不敏感(Linker type_name 是小写)
const compiler::TypeMeta b2 = compiler::type_meta(cfg, "bool");
CHECK(b2 && b2.tag() == 0 && b2.width() == 1);
// INTint16 → 2 字节、有符号、tag 1、无 range
const compiler::TypeMeta i = compiler::type_meta(cfg, "int");
CHECK(i && i.width() == 2 && i.is_signed() && !i.is_float() && i.tag() == 1);
CHECK(!i.has_range() || i.value_in_range(30000));
// TIMEint64 → 8 字节、有符号、tag 2
const compiler::TypeMeta t = compiler::type_meta(cfg, "TIME");
CHECK(t && t.width() == 8 && t.is_signed() && !t.is_float() && t.tag() == 2);
// 未定义类型 → 空
CHECK(!compiler::type_meta(cfg, "REAL"));
CHECK(!compiler::type_meta(cfg, ""));
return true;
}
int main() { int main() {
if (!test_positive()) return 1; if (!test_positive()) return 1;
if (!test_negative()) return 1; if (!test_negative()) return 1;
if (!test_positive_min()) return 1; if (!test_positive_min()) return 1;
if (!test_type_meta()) return 1;
std::printf("machine_test: %d checks passed\n", g_checks); std::printf("machine_test: %d checks passed\n", g_checks);
return 0; return 0;
} }