fix GetVariableOffset in cases of multi-level and multiple inheritance

Author: Vipul-Cariappa

include/clang/Interpreter/CppInterOp.h

@@ -446,7 +446,8 @@ namespace Cpp {
 
   /// Gets the address of the variable, you can use it to get the
   /// value stored in the variable.
-  CPPINTEROP_API intptr_t GetVariableOffset(TCppScope_t var);
+  CPPINTEROP_API intptr_t GetVariableOffset(TCppScope_t var,
+                                            TCppScope_t parent = nullptr);
 
   /// Checks if the provided variable is a 'Public' variable.
   CPPINTEROP_API bool IsPublicVariable(TCppScope_t var);

lib/Interpreter/CppInterOp.cpp

@@ -1222,20 +1222,22 @@ namespace Cpp {
     return 0;
   }
 
-  intptr_t GetVariableOffset(compat::Interpreter& I, Decl* D) {
+  intptr_t GetVariableOffset(compat::Interpreter& I, Decl* D,
+                             CXXRecordDecl* BaseCXXRD) {
     if (!D)
       return 0;
 
     auto& C = I.getSema().getASTContext();
 
     if (auto* FD = llvm::dyn_cast<FieldDecl>(D)) {
-      const clang::RecordDecl* RD = FD->getParent();
+      clang::RecordDecl* FieldParentRecordDecl = FD->getParent();
       intptr_t offset =
           C.toCharUnitsFromBits(C.getFieldOffset(FD)).getQuantity();
-      while (RD->isAnonymousStructOrUnion()) {
-        const clang::RecordDecl* anon = RD;
-        RD = llvm::dyn_cast<RecordDecl>(anon->getParent());
-        for (auto F = RD->field_begin(); F != RD->field_end(); ++F) {
+      while (FieldParentRecordDecl->isAnonymousStructOrUnion()) {
+        clang::RecordDecl* anon = FieldParentRecordDecl;
+        FieldParentRecordDecl = llvm::dyn_cast<RecordDecl>(anon->getParent());
+        for (auto F = FieldParentRecordDecl->field_begin();
+             F != FieldParentRecordDecl->field_end(); ++F) {
           const auto* RT = F->getType()->getAs<RecordType>();
           if (!RT)
             continue;
@@ -1246,6 +1248,46 @@ namespace Cpp {
         }
         offset += C.toCharUnitsFromBits(C.getFieldOffset(FD)).getQuantity();
       }
+      if (BaseCXXRD && BaseCXXRD != FieldParentRecordDecl) {
+        // FieldDecl FD belongs to some class C, but the base class BaseCXXRD is
+        // not C. That means BaseCXXRD derives from C. Offset needs to be
+        // calculated for Derived class
+
+        // Depth first Search is performed to the class that declears FD from
+        // the base class
+        std::vector<CXXRecordDecl*> stack;
+        std::map<CXXRecordDecl*, CXXRecordDecl*> direction;
+        stack.push_back(BaseCXXRD);
+        while (stack.size()) {
+          CXXRecordDecl* RD = stack.back();
+          stack.pop_back();
+          size_t num_bases = GetNumBases(RD);
+          bool flag = false;
+          for (size_t i = 0; i < num_bases; i++) {
+            auto* CRD = static_cast<CXXRecordDecl*>(GetBaseClass(RD, i));
+            direction[CRD] = RD;
+            if (CRD == FieldParentRecordDecl) {
+              flag = true;
+              break;
+            }
+            stack.push_back(CRD);
+          }
+          if (flag)
+            break;
+        }
+        if (auto* RD = llvm::dyn_cast<CXXRecordDecl>(FieldParentRecordDecl)) {
+          // add in the offsets for the (multi level) base classes
+          while (BaseCXXRD != RD) {
+            CXXRecordDecl* Parent = direction.at(RD);
+            offset += C.getASTRecordLayout(Parent)
+                          .getBaseClassOffset(RD)
+                          .getQuantity();
+            RD = Parent;
+          }
+        } else {
+          assert(false && "Unreachable");
+        }
+      }
       return offset;
     }
 
@@ -1321,9 +1363,11 @@ namespace Cpp {
     return 0;
   }
 
-  intptr_t GetVariableOffset(TCppScope_t var) {
+  intptr_t GetVariableOffset(TCppScope_t var, TCppScope_t parent) {
     auto* D = static_cast<Decl*>(var);
-    return GetVariableOffset(getInterp(), D);
+    auto* RD =
+        llvm::dyn_cast_or_null<CXXRecordDecl>(static_cast<Decl*>(parent));
+    return GetVariableOffset(getInterp(), D, RD);
   }
 
   // Check if the Access Specifier of the variable matches the provided value.

unittests/CppInterOp/VariableReflectionTest.cpp

@@ -258,6 +258,81 @@ TEST(VariableReflectionTest, GetVariableOffset) {
   EXPECT_TRUE((bool) Cpp::GetVariableOffset(VD_C_s_a));
 }
 
+#define CODE                                                                   \
+  class BaseA {                                                                \
+  public:                                                                      \
+    int a;                                                                     \
+    BaseA(int a) : a(a) {}                                                     \
+  };                                                                           \
+                                                                               \
+  class BaseB : public BaseA {                                                 \
+  public:                                                                      \
+    std::string b;                                                             \
+    BaseB(int x, std::string b) : BaseA(x), b(b) {}                            \
+  };                                                                           \
+                                                                               \
+  class Base1 {                                                                \
+  public:                                                                      \
+    int i;                                                                     \
+    std::string s;                                                             \
+    Base1(int i, std::string s) : i(i), s(s) {}                                \
+  };                                                                           \
+                                                                               \
+  class MyKlass : public BaseB, public Base1 {                                 \
+  public:                                                                      \
+    int k;                                                                     \
+    MyKlass(int k, int i, int x, std::string b, std::string s)                 \
+        : BaseB(x, b), Base1(i, s), k(k) {}                                    \
+  } my_k(5, 4, 3, "Cpp", "Python");
+
+CODE
+
+TEST(VariableReflectionTest, VariableOffsetsWithInheritance) {
+  Cpp::Declare("#include<string>");
+
+#define Stringify(s) Stringifyx(s)
+#define Stringifyx(...) #__VA_ARGS__
+  Cpp::Declare(Stringify(CODE));
+#undef Stringifyx
+#undef Stringify
+#undef CODE
+
+  Cpp::TCppScope_t myklass = Cpp::GetNamed("MyKlass");
+  EXPECT_TRUE(myklass);
+
+  size_t num_bases = Cpp::GetNumBases(myklass);
+  EXPECT_EQ(num_bases, 2);
+
+  std::vector<Cpp::TCppScope_t> datamembers;
+  Cpp::GetDatamembers(myklass, datamembers);
+  for (size_t i = 0; i < num_bases; i++) {
+    Cpp::TCppScope_t base = Cpp::GetBaseClass(myklass, i);
+    EXPECT_TRUE(base);
+    for (size_t i = 0; i < Cpp::GetNumBases(base); i++) {
+      Cpp::TCppScope_t bbase = Cpp::GetBaseClass(base, i);
+      EXPECT_TRUE(base);
+      Cpp::GetDatamembers(bbase, datamembers);
+    }
+    Cpp::GetDatamembers(base, datamembers);
+  }
+  EXPECT_EQ(datamembers.size(), 5);
+
+  EXPECT_EQ(Cpp::GetVariableOffset(datamembers[0], myklass),
+            ((intptr_t)&(my_k.k)) - ((intptr_t)&(my_k)));
+
+  EXPECT_EQ(Cpp::GetVariableOffset(datamembers[1], myklass),
+            ((intptr_t)&(my_k.a)) - ((intptr_t)&(my_k)));
+
+  EXPECT_EQ(Cpp::GetVariableOffset(datamembers[2], myklass),
+            ((intptr_t)&(my_k.b)) - ((intptr_t)&(my_k)));
+
+  EXPECT_EQ(Cpp::GetVariableOffset(datamembers[3], myklass),
+            ((intptr_t)&(my_k.i)) - ((intptr_t)&(my_k)));
+
+  EXPECT_EQ(Cpp::GetVariableOffset(datamembers[4], myklass),
+            ((intptr_t)&(my_k.s)) - ((intptr_t)&(my_k)));
+}
+
 TEST(VariableReflectionTest, IsPublicVariable) {
   std::vector<Decl *> Decls, SubDecls;
   std::string code = R"(