Implement unions and largest-element allocation

include/Data.h

@@ -20,7 +20,8 @@ extern_ struct SymbolTableEntry* Globals, * GlobalsEnd;
 extern_ struct SymbolTableEntry* Locals, * LocalsEnd;
 extern_ struct SymbolTableEntry* Params, * ParamsEnd;
 extern_ struct SymbolTableEntry* Structs, * StructsEnd;
-extern_ struct SymbolTableEntry* StructMembers, * StructMembersEnd;
+
+extern_ struct SymbolTableEntry* CompositeMembers, * CompositeMembersEnd;
 
 extern_ struct SymbolTableEntry* Unions, * UnionsEnd;
 extern_ struct SymbolTableEntry* Enums, * EnumsEnd;

include/Defs.h

@@ -96,7 +96,8 @@ enum TokenTypes {
     KW_WHILE,
     KW_FOR,
     KW_RETURN,
-    KW_STRUCT
+    KW_STRUCT,
+    KW_UNION
 };
 
 /*
@@ -219,6 +220,7 @@ struct SymbolTableEntry {
 enum StorageScope {
     SC_GLOBAL = 1,  // Global Scope
     SC_STRUCT,      // Struct Definitions
+    SC_UNION,       // Union Definitions
     SC_ENUM,        // Enum Definitions
     SC_MEMBER,      // The members of Structs or Enums
     //SC_CLASS,       // Class-local definitions
@@ -348,7 +350,7 @@ struct ASTNode* ParseFunction(int Type);
 
 struct ASTNode* ParseCompound();
 
-struct SymbolTableEntry* BeginStructDeclaration();
+struct SymbolTableEntry* BeginCompositeDeclaration(int Type);
 
 struct ASTNode* GetExpressionList();
 
@@ -387,6 +389,8 @@ struct SymbolTableEntry* FindGlobal(char* Symbol);
 
 struct SymbolTableEntry* FindStruct(char* Symbol);
 
+struct SymbolTableEntry* FindUnion(char* Symbol);
+
 struct SymbolTableEntry* FindMember(char* Symbol);
 
 void AppendSymbol(struct SymbolTableEntry** Head, struct SymbolTableEntry** Tail, struct SymbolTableEntry* Node);

src/Assembler.c

@@ -799,7 +799,7 @@ int AsDeref(int Reg, int Type) {
     printf("\tDereferencing %s\n", Registers[Reg]);
     switch (DestSize) {
         case 1:
-            fprintf(OutputFile, "\tmovzbq\t(%s), %s\n", Registers[Reg], ByteRegisters[Reg]);
+            fprintf(OutputFile, "\tmovzbq\t(%s), %s\n", Registers[Reg], Registers[Reg]);
             break;
         case 2:
             fprintf(OutputFile, "\tmovslq\t(%s), %s\n", Registers[Reg], DoubleRegisters[Reg]);

src/Lexer.c

@@ -338,6 +338,11 @@ static int ReadKeyword(char* Str) {
                 return KW_STRUCT;
             break;
 
+        case 'u':
+            if(!strcmp(Str, "union"))
+                return KW_UNION;
+            break;
+
         case 'v':
             if (!strcmp(Str, "void"))
                 return TY_VOID;

src/Main.c

@@ -87,6 +87,7 @@ char* ScopeNames[] = {
         "INVALID",
         "GLOBAL",
         "STRUCT",
+        "UNION",
         "ENUM",
         "MEMBER",
         "PARAMETER",

src/Parser.c

@@ -534,7 +534,7 @@ void ParseGlobals() {
 
         // Structs are parsed fully in ParseOptionalPointer
         // TODO: FIX THAT!!
-        if (Type == DAT_STRUCT && CurrentToken.type == LI_SEMIC) {
+        if ((Type == DAT_STRUCT || Type == DAT_UNION) && CurrentToken.type == LI_SEMIC) {
             Tokenise();
             continue;
         }

src/Pointers.c

@@ -94,7 +94,11 @@ int ParseOptionalPointer(struct SymbolTableEntry** Composite) {
             break;
         case KW_STRUCT:
             Type = DAT_STRUCT;
-            *Composite = BeginStructDeclaration();
+            *Composite = BeginCompositeDeclaration(Type);
+            break;
+        case KW_UNION:
+            Type = DAT_UNION;
+            *Composite = BeginCompositeDeclaration(Type);
             break;
         default:
             DieDecimal("Illegal type for pointerisation", CurrentToken.type);
@@ -175,13 +179,13 @@ struct ASTNode* AccessMember(bool Deref) {
 
     if ((CompositeVar = FindSymbol(CurrentIdentifier)) == NULL)
         DieMessage("Undeclared variable", CurrentIdentifier);
-    if (Deref && CompositeVar->Type != PointerTo(DAT_STRUCT))
+    if (Deref && (CompositeVar->Type != PointerTo(DAT_STRUCT) && CompositeVar->Type != PointerTo(DAT_UNION)))
         DieMessage("Undeclared struct", CurrentIdentifier);
-    if (!Deref && CompositeVar->Type != DAT_STRUCT)
+    if (!Deref && (CompositeVar->Type != DAT_STRUCT && CompositeVar->Type != DAT_UNION))
         DieMessage("Undeclared struct", CurrentIdentifier);
 
     if (Deref)
-        LeftNode = ConstructASTLeaf(REF_IDENT, PointerTo(DAT_STRUCT), CompositeVar, 0);
+        LeftNode = ConstructASTLeaf(REF_IDENT, PointerTo(CompositeVar->Type), CompositeVar, 0);
     else
         LeftNode = ConstructASTLeaf(OP_ADDRESS, CompositeVar->Type, CompositeVar, 0);
 
@@ -193,14 +197,14 @@ struct ASTNode* AccessMember(bool Deref) {
     VerifyToken(TY_IDENTIFIER, "identifier");
 
     for (Member = TypePtr->Start; Member != NULL; Member = Member->NextSymbol) {
-        printf("\tComparing struct entry %s with the wanted %s. Index %d.\r\n", Member->Name, CurrentIdentifier,
+        printf("\tComparing composite entry %s with the wanted %s. Index %d.\r\n", Member->Name, CurrentIdentifier,
                Member->SinkOffset);
         if (!strcmp(Member->Name, CurrentIdentifier))
             break;
     }
 
     if (Member == NULL)
-        DieMessage("Invalid struct member", CurrentIdentifier);
+        DieMessage("Invalid composite member", CurrentIdentifier);
 
     RightNode = ConstructASTLeaf(TERM_INTLITERAL, RET_INT, NULL, Member->SinkOffset);
 

src/Statements.c

@@ -62,25 +62,26 @@ static int ReadDeclarationList(struct SymbolTableEntry* FunctionSymbol, int Stor
 }
 
 /*
- * Handles the declaration of a new struct.
+ * Handles the declaration of a new composite type.
+ * For example, a struct is a composite of multiple different named positions:
  *  struct thisStct { int x, int y, int z };
  * 
  * Verifies that the current identifier is not used,
  *  verifies that this is not a redefinition (excluding
  *   the case where there is a declaration but no definition)
- *  and then saves it into the Structs symbol table.
+ *  and then saves it into the appropriate symbol table.
  * 
- * @return the Symbol Table entry of this new struct.
+ * @return the Symbol Table entry of this new composite.
  */
 
-struct SymbolTableEntry* BeginStructDeclaration() {
+struct SymbolTableEntry* BeginCompositeDeclaration(int Type) {
     struct SymbolTableEntry* Composite = NULL, * Member;
-    int Offset;
+    int Offset = 0, Largest = 0;
 
     Tokenise();
 
     if (CurrentToken.type == TY_IDENTIFIER) {
-        Composite = FindStruct(CurrentIdentifier);
+        Composite = Type == DAT_STRUCT ? FindStruct(CurrentIdentifier) : FindUnion(CurrentIdentifier);
         Tokenise();
     }
 
@@ -91,29 +92,34 @@ struct SymbolTableEntry* BeginStructDeclaration() {
     }
 
     if (Composite)
-        DieMessage("Redefinition of struct", CurrentIdentifier);
+        DieMessage("Redefinition of composite", CurrentIdentifier);
 
-    Composite = AddSymbol(CurrentIdentifier, DAT_STRUCT, 0, SC_STRUCT, 0, 0, NULL);
+    Composite = AddSymbol(CurrentIdentifier, Type, 0, Type == DAT_STRUCT ? SC_STRUCT : SC_UNION, 0, 0, NULL);
     Tokenise();
-    printf("Reading a struct declaration..\n");
+    printf("Reading a composite declaration.. Type is %s\n", Type == DAT_STRUCT ? "struct" : "union");
     ReadDeclarationList(NULL, SC_MEMBER, LI_RBRAC);
     VerifyToken(LI_RBRAC, "}");
 
-    Composite->Start = StructMembers;
-    StructMembers = StructMembersEnd = NULL;
+    Composite->Start = CompositeMembers;
+    CompositeMembers = CompositeMembersEnd = NULL;
 
     Member = Composite->Start;
-    printf("\tSetting first entry in struct to %s\r\n", Member->Name);
+    printf("\tSetting first entry in composite to %s\r\n", Member->Name);
     Member->SinkOffset = 0;
     Offset = TypeSize(Member->Type, Member->CompositeType);
 
     for (Member = Member->NextSymbol; Member != NULL; Member = Member->NextSymbol) {
-        Member->SinkOffset = AsAlignMemory(Member->Type, Offset, 1);
+        if (Type == DAT_STRUCT)
+            Member->SinkOffset = AsAlignMemory(Member->Type, Offset, 1);
+        else
+            Member->SinkOffset = 0;
 
-        Offset += TypeSize(Member->Type, Member->CompositeType);
+        int CurrentSize = TypeSize(Member->Type, Member->CompositeType);
+        Offset += CurrentSize;
+        Largest = CurrentSize > Largest ? CurrentSize : Largest;
     }
 
-    Composite->Length = Offset;
+    Composite->Length = Type == DAT_STRUCT ? Offset : Largest;
     return Composite;
 }
 

src/Symbols.c

@@ -120,6 +120,16 @@ struct SymbolTableEntry* FindGlobal(char* Symbol) {
 struct SymbolTableEntry* FindStruct(char* Symbol) {
     return SearchList(Symbol, Structs);
 }
+/*
+ * An override for FindSymbol.
+ * Searches only the defined Unions.
+ *  @param Symbol: The string name of the symbol to search for.
+ *  @return a pointer to the node if found, else NULL
+ *
+ */
+struct SymbolTableEntry* FindUnion(char* Symbol) {
+    return SearchList(Symbol, Unions);
+}
 
 /*
  * An override for FindSymbol.
@@ -129,7 +139,7 @@ struct SymbolTableEntry* FindStruct(char* Symbol) {
  * 
  */
 struct SymbolTableEntry* FindMember(char* Symbol) {
-    return SearchList(Symbol, StructMembers);
+    return SearchList(Symbol, CompositeMembers);
 }
 
 /*
@@ -177,7 +187,7 @@ void ClearTables() {
     Globals = GlobalsEnd = NULL;
     Locals = LocalsEnd = NULL;
     Params = ParamsEnd = NULL;
-    StructMembers = StructMembersEnd = NULL;
+    CompositeMembers = CompositeMembersEnd = NULL;
     Structs = StructsEnd = NULL;
 }
 
@@ -218,8 +228,12 @@ struct SymbolTableEntry* AddSymbol(char* Name, int Type, int Structure, int Stor
         case SC_STRUCT:
             AppendSymbol(&Structs, &StructsEnd, Node);
             break;
+        case SC_UNION:
+            AppendSymbol(&Unions, &UnionsEnd, Node);
+            break;
         case SC_MEMBER:
-            AppendSymbol(&StructMembers, &StructMembersEnd, Node);
+            AppendSymbol(&CompositeMembers, &CompositeMembersEnd, Node);
+            break;
         case SC_LOCAL:
             AppendSymbol(&Locals, &LocalsEnd, Node);
             break;

src/Types.c

@@ -64,7 +64,7 @@ int PrimitiveSize(int Type) {
  */
 
 int TypeSize(int Type, struct SymbolTableEntry* Composite) {
-    if (Type == DAT_STRUCT) return Composite->Length;
+    if (Type == DAT_STRUCT || Type == DAT_UNION) return Composite->Length;
     return PrimitiveSize(Type);
 }
 

tests/union.er

@@ -0,0 +1,21 @@
+int :: printf(char* format);
+
+union fruit {
+    char cherry,
+    int apple,
+    int orange,
+    long banana
+};
+
+union fruit basket;
+
+int :: main() {
+    basket.apple = 55;
+    printf("1 %d b\n", basket.cherry);
+    printf("2 %d a\n", basket.orange);
+    basket.orange = 100;
+    printf("3 %d c\n", basket.banana);
+    printf("4 %d d\n", basket.apple);
+
+    return (0);
+}