添加数据流分析类，实现前向后向分析的模板动作，实现活跃变量分析，基本借鉴学长代码，后续可优化实现

src/SysYIRAnalyser.cpp

@@ -254,6 +254,267 @@ void ControlFlowAnalysis::computeDomFrontierAllBlk() {
   }
 }
 
+// ==========================
+// dataflow analysis utils
+// ==========================
+
+// 先引用学长的代码
+// TODO: Worklist 增加逆后序遍历机制
+void DataFlowAnalysisUtils::forwardAnalyze(Module *pModule){
+  std::map<DataFlowAnalysis *, bool> workAnalysis;
+  for (auto &dataflow : forwardAnalysisList) {
+    dataflow->init(pModule);
+  }
+
+  for (const auto &function : pModule->getFunctions()) {
+    for (auto &dataflow : forwardAnalysisList) {
+      workAnalysis.emplace(dataflow, false);
+    }
+    while (!workAnalysis.empty()) {
+      for (const auto &block : function.second->getBasicBlocks()) {
+        for (auto &elem : workAnalysis) {
+          if (elem.first->analyze(pModule, block.get())) {
+            elem.second = true;
+          }
+        }
+      }
+      std::map<DataFlowAnalysis *, bool> tmp;
+      std::remove_copy_if(workAnalysis.begin(), workAnalysis.end(), std::inserter(tmp, tmp.end()),
+                          [](const std::pair<DataFlowAnalysis *, bool> &elem) -> bool { return !elem.second; });
+      workAnalysis.swap(tmp);
+
+      for (auto &elem : workAnalysis) {
+        elem.second = false;
+      }
+    }
+  }
+}
+
+void DataFlowAnalysisUtils::backwardAnalyze(Module *pModule) {
+  std::map<DataFlowAnalysis *, bool> workAnalysis;
+  for (auto &dataflow : backwardAnalysisList) {
+    dataflow->init(pModule);
+  }
+
+  for (const auto &function : pModule->getFunctions()) {
+    for (auto &dataflow : backwardAnalysisList) {
+      workAnalysis.emplace(dataflow, false);
+    }
+    while (!workAnalysis.empty()) {
+      for (const auto &block : function.second->getBasicBlocks()) {
+        for (auto &elem : workAnalysis) {
+          if (elem.first->analyze(pModule, block.get())) {
+            elem.second = true;
+          }
+        }
+      }
+      std::map<DataFlowAnalysis *, bool> tmp;
+      std::remove_copy_if(workAnalysis.begin(), workAnalysis.end(), std::inserter(tmp, tmp.end()),
+                          [](const std::pair<DataFlowAnalysis *, bool> &elem) -> bool { return !elem.second; });
+      workAnalysis.swap(tmp);
+
+      for (auto &elem : workAnalysis) {
+        elem.second = false;
+      }
+    }
+  }
+}
+
+
+std::set<User *> ActiveVarAnalysis::getUsedSet(Instruction *inst) {
+  using Kind = Instruction::Kind;
+  std::vector<User *> operands;
+  for (const auto &operand : inst->getOperands()) {
+    operands.emplace_back(dynamic_cast<User *>(operand->getValue()));
+  }
+  std::set<User *> result;
+  switch (inst->getKind()) {
+    // phi op
+    case Kind::kPhi:
+    case Kind::kCall:
+      result.insert(std::next(operands.begin()), operands.end());
+      break;
+    case Kind::kCondBr:
+      result.insert(operands[0]);
+      break;
+    case Kind::kBr:
+    case Kind::kAlloca:
+      break;
+    // mem op
+    case Kind::kStore:
+      // StoreInst 的第一个操作数是被存储的值，第二个操作数是存储的变量
+      // 后续的是可能的数组维度
+      result.insert(operands[0]);
+      result.insert(operands.begin() + 2, operands.end());
+      break;
+    case Kind::kLoad:
+    case Kind::kLa: {
+      auto variable = dynamic_cast<AllocaInst *>(operands[0]);
+      auto global = dynamic_cast<GlobalValue *>(operands[0]);
+      auto constArray = dynamic_cast<ConstantVariable *>(operands[0]);
+      if ((variable != nullptr && variable->getNumDims() == 0) || (global != nullptr && global->getNumDims() == 0) ||
+          (constArray != nullptr && constArray->getNumDims() == 0)) {
+        result.insert(operands[0]);
+      }
+      result.insert(std::next(operands.begin()), operands.end());
+      break;
+    }
+    case Kind::kGetSubArray: {
+      for (unsigned i = 2; i < operands.size(); i++) {
+        // 数组的维度信息
+        result.insert(operands[i]);
+      }
+      break;
+    }
+    case Kind::kMemset: {
+      result.insert(std::next(operands.begin()), operands.end());
+      break;
+    }
+    case Kind::kInvalid:
+    // Binary
+    case Kind::kAdd:
+    case Kind::kSub:
+    case Kind::kMul:
+    case Kind::kDiv:
+    case Kind::kRem:
+    case Kind::kICmpEQ:
+    case Kind::kICmpNE:
+    case Kind::kICmpLT:
+    case Kind::kICmpLE:
+    case Kind::kICmpGT:
+    case Kind::kICmpGE:
+    case Kind::kFAdd:
+    case Kind::kFSub:
+    case Kind::kFMul:
+    case Kind::kFDiv:
+    case Kind::kFCmpEQ:
+    case Kind::kFCmpNE:
+    case Kind::kFCmpLT:
+    case Kind::kFCmpLE:
+    case Kind::kFCmpGT:
+    case Kind::kFCmpGE:
+    case Kind::kAnd:
+    case Kind::kOr:
+    // Unary
+    case Kind::kNeg:
+    case Kind::kNot:
+    case Kind::kFNot:
+    case Kind::kFNeg:
+    case Kind::kFtoI:
+    case Kind::kItoF:
+    // terminator
+    case Kind::kReturn:
+      result.insert(operands.begin(), operands.end());
+      break;
+    default:
+      assert(false);
+      break;
+  }
+  result.erase(nullptr);
+  return result;
+}
+
+User * ActiveVarAnalysis::getDefine(Instruction *inst) {
+  User *result = nullptr;
+  if (inst->isStore()) {
+    StoreInst* store = dynamic_cast<StoreInst *>(inst);
+    auto operand = store->getPointer();
+    AllocaInst* variable = dynamic_cast<AllocaInst *>(operand);
+    GlobalValue* global = dynamic_cast<GlobalValue *>(operand);
+    if ((variable != nullptr && variable->getNumDims() != 0) || (global != nullptr && global->getNumDims() != 0)) {
+      // 如果是数组变量或者全局变量，则不返回定义
+      // TODO：兼容数组变量
+      result = nullptr;
+    } else {
+      result = dynamic_cast<User *>(operand);
+    }
+  } else if (inst->isPhi()) {
+    result = dynamic_cast<User *>(inst->getOperand(0));
+  } else if (inst->isBinary() || inst->isUnary() || inst->isCall() ||
+             inst->isLoad() || inst->isLa()) {
+    result = dynamic_cast<User *>(inst);
+  }
+  return result;
+}
+
+void ActiveVarAnalysis::init(Module *pModule) {
+  for (const auto &function : pModule->getFunctions()) {
+    for (const auto &block : function.second->getBasicBlocks()) {
+      activeTable.emplace(block.get(), std::vector<std::set<User *>>{});
+      for (unsigned i = 0; i < block->getNumInstructions() + 1; i++)
+        activeTable.at(block.get()).emplace_back();
+    }
+  }
+}
+
+// 活跃变量分析公式 每个块内的分析动作供分析器调用
+bool ActiveVarAnalysis::analyze(Module *pModule, BasicBlock *block) {
+  bool changed = false;  // 标记数据流结果是否有变化
+  std::set<User *> activeSet{};  // 当前计算的活跃变量集合
+
+  // 步骤1: 计算基本块出口的活跃变量集 (OUT[B])
+  // 公式: OUT[B] = ∪_{S ∈ succ(B)} IN[S]
+  for (const auto &succ : block->getSuccessors()) {
+    // 获取后继块入口的活跃变量集 (IN[S])
+    auto succActiveSet = activeTable.at(succ).front();
+    // 合并所有后继块的入口活跃变量
+    activeSet.insert(succActiveSet.begin(), succActiveSet.end());
+  }
+
+  // 步骤2: 处理基本块出口处的活跃变量集
+  const auto &instructions = block->getInstructions();
+  const auto numInstructions = instructions.size();
+  
+  // 获取旧的出口活跃变量集 (block出口对应索引numInstructions)
+  const auto &oldEndActiveSet = activeTable.at(block)[numInstructions];
+  
+  // 检查出口活跃变量集是否有变化
+  if (!std::equal(activeSet.begin(), activeSet.end(), 
+                 oldEndActiveSet.begin(), oldEndActiveSet.end())) 
+  {
+    changed = true;  // 标记变化
+    activeTable.at(block)[numInstructions] = activeSet;  // 更新出口活跃变量集
+  }
+
+  // 步骤3: 逆序遍历基本块中的指令
+  // 从最后一条指令开始向前计算每个程序点的活跃变量
+  auto instructionIter = instructions.end();
+  instructionIter--;  // 指向最后一条指令
+  
+  // 从出口向入口遍历 (索引从numInstructions递减到1)
+  for (unsigned i = numInstructions; i > 0; i--) {
+    auto inst = instructionIter->get();  // 当前指令
+    
+    auto used = getUsedSet(inst);
+    User *defined = getDefine(inst);
+    
+    // 步骤3.3: 计算指令入口的活跃变量 (IN[i])
+    // 公式: IN[i] = use_i ∪ (OUT[i] - def_i)
+    activeSet.erase(defined);    // 移除被定义的变量 (OUT[i] - def_i)
+    activeSet.insert(used.begin(), used.end());  // 添加使用的变量
+    
+    // 获取旧的入口活跃变量集 (位置i-1对应当前指令的入口)
+    const auto &oldActiveSet = activeTable.at(block)[i - 1];
+    
+    // 检查活跃变量集是否有变化
+    if (!std::equal(activeSet.begin(), activeSet.end(), 
+                   oldActiveSet.begin(), oldActiveSet.end())) 
+    {
+      changed = true;  // 标记变化
+      activeTable.at(block)[i - 1] = activeSet;  // 更新入口活跃变量集
+    }
+    
+    instructionIter--;  // 移动到前一条指令
+  }
+
+  return changed;  // 返回数据流结果是否变化
+}
+
+
+auto ActiveVarAnalysis::getActiveTable() const -> const std::map<BasicBlock *, std::vector<std::set<User *>>> & {
+  return activeTable;
+}
+
 
 } // namespace sysy
 

src/include/SysYIRAnalyser.h

@@ -252,6 +252,7 @@ public:
   bool isParallelable() const { return parallelable; }
 };
 
+// 控制流分析类
 class ControlFlowAnalysis {
 private:
   Module *pModule;  ///< 模块
@@ -287,8 +288,111 @@ private:
   BasicBlock* findCommonDominator(BasicBlock *a, BasicBlock *b);  // 查找两个基本块的共同支配结点
 };
 
+// 数据流分析类
+// 该类为抽象类，具体的数据流分析器需要继承此类
+// 因为每个数据流分析器的分析动作都不一样，所以需要继承并实现analyze方法
+class DataFlowAnalysis {
+ public:
+  virtual ~DataFlowAnalysis() = default;
 
-// 分析管理器（整合版）
+ public:
+  virtual void init(Module *pModule) {}                                              ///< 分析器初始化
+  virtual auto analyze(Module *pModule, BasicBlock *block) -> bool { return true; }  ///< 分析动作，若完成则返回true;
+  virtual void clear() {}                                                            ///< 清空
+};
+
+// 数据流分析工具类
+// 该类用于管理多个数据流分析器，提供统一的前向与后向分析接口
+class DataFlowAnalysisUtils {
+private:
+  std::vector<DataFlowAnalysis *> forwardAnalysisList;   ///< 前向分析器列表
+  std::vector<DataFlowAnalysis *> backwardAnalysisList;  ///< 后向分析器列表
+
+public:
+  DataFlowAnalysisUtils() = default;
+  
+  // 统一构造
+  DataFlowAnalysisUtils(
+    std::vector<DataFlowAnalysis *> forwardList = {},
+    std::vector<DataFlowAnalysis *> backwardList = {})
+    : forwardAnalysisList(std::move(forwardList)),
+      backwardAnalysisList(std::move(backwardList)) {}
+  
+  // 统一添加接口
+  void addAnalyzers(
+    std::vector<DataFlowAnalysis *> forwardList,
+    std::vector<DataFlowAnalysis *> backwardList = {}) 
+  {
+    forwardAnalysisList.insert(
+      forwardAnalysisList.end(), 
+      forwardList.begin(), 
+      forwardList.end());
+    
+    backwardAnalysisList.insert(
+      backwardAnalysisList.end(),
+      backwardList.begin(),
+      backwardList.end());
+  }
+  
+  // 单独添加接口
+  void addForwardAnalyzer(DataFlowAnalysis *analyzer) {
+    forwardAnalysisList.push_back(analyzer);
+  }
+  
+  void addBackwardAnalyzer(DataFlowAnalysis *analyzer) {
+    backwardAnalysisList.push_back(analyzer);
+  }
+
+  // 设置分析器列表
+  void setAnalyzers(
+    std::vector<DataFlowAnalysis *> forwardList,
+    std::vector<DataFlowAnalysis *> backwardList)
+  {
+    forwardAnalysisList = std::move(forwardList);
+    backwardAnalysisList = std::move(backwardList);
+  }
+
+  // 清空列表
+  void clear() {
+    forwardAnalysisList.clear();
+    backwardAnalysisList.clear();
+  }
+
+  // 访问器
+  const auto& getForwardAnalyzers() const { return forwardAnalysisList; }
+  const auto& getBackwardAnalyzers() const { return backwardAnalysisList; }
+
+public:
+  void forwardAnalyze(Module *pModule);   ///< 执行前向分析
+  void backwardAnalyze(Module *pModule);  ///< 执行后向分析
+};
+
+// 活跃变量分析类
+// 提供def - use分析
+// 未兼容数组变量但是考虑了维度的use信息
+class ActiveVarAnalysis : public DataFlowAnalysis {
+ private:
+  std::map<BasicBlock *, std::vector<std::set<User *>>> activeTable;  ///< 活跃信息表，存储每个基本块内的的活跃变量信息
+
+ public:
+  ActiveVarAnalysis() = default;
+  ~ActiveVarAnalysis() override = default;
+
+ public:
+  static std::set<User*> getUsedSet(Instruction *inst);
+  static User* getDefine(Instruction *inst);
+
+ public:
+  void init(Module *pModule) override;
+  bool analyze(Module *pModule, BasicBlock *block) override;
+  // 外部活跃信息表访问器
+  const std::map<BasicBlock *, std::vector<std::set<User *>>> &getActiveTable() const;
+  void clear() override {
+    activeTable.clear();  // 清空活跃信息表
+  }
+};
+
+// 分析管理器
 class AnalysisManager {
 
 };