feat(codegen): overload global functions and procedures

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

8 files changed, 113 insertions, 54 deletions

diff --git a/lib/CodeGen/CodeGen.cc b/lib/CodeGen/CodeGen.cc
index b81fe2b..9f59a4d 100644
--- a/lib/CodeGen/CodeGen.cc
+++ b/lib/CodeGen/CodeGen.cc
@@ -160,7 +160,7 @@ bool CodeGen::isVisible(size_t Owner) const {
 void CodeGen::selectUnit(size_t Unit) {
     CurrentUnit = Unit;
     DiagnosticSource = Units->size() > 1 ? (*Units)[Unit].Path : "";
-    FuncGen->setCurrentUnit(Unit);
+    FuncGen->setCurrentUnit(Unit, (*Units)[Unit].Path);
 }
 
 void CodeGen::importSymbols() {
@@ -187,15 +187,17 @@ void CodeGen::compile(const std::vector<SourceUnit> &Sources) {
     });
     FuncGen->setVisibility([this](size_t Owner) { return isVisible(Owner); });
 
-    IdentifierMap<size_t> Definitions;
+    struct Definition { size_t Unit; bool IsFunction; bool IsOverload; };
+    IdentifierMap<Definition> Definitions;
     std::optional<size_t> CodeUnit;
     FunctionDefAST *Entry = nullptr;
     size_t EntryUnit = 0;
-    auto Reserve = [&](const std::string &Name) {
-        auto [It, Added] = Definitions.emplace(Name, CurrentUnit);
-        if (!Added && It->second != CurrentUnit)
+    auto Reserve = [&](const std::string &Name, bool IsFunction = false, bool IsOverload = false) {
+        auto [It, Added] = Definitions.emplace(Name, Definition{CurrentUnit, IsFunction, IsOverload});
+        bool Compatible = IsFunction && IsOverload && It->second.IsFunction && It->second.IsOverload;
+        if (!Added && It->second.Unit != CurrentUnit && !Compatible)
             reportError("Duplicate definition %s in %s and %s", Name.c_str(),
-                        Sources[It->second].Path.c_str(), Sources[CurrentUnit].Path.c_str());
+                        Sources[It->second.Unit].Path.c_str(), Sources[CurrentUnit].Path.c_str());
     };
     for (size_t I = 0; I < Sources.size(); ++I) {
         selectUnit(I);
@@ -203,7 +205,7 @@ void CodeGen::compile(const std::vector<SourceUnit> &Sources) {
             SaveAndRestore<int> Location(DiagnosticLine, S->getLine());
             std::string TypeName;
             if (auto *F = dynamic_cast<FunctionDefAST*>(S.get())) {
-                Reserve(F->getProto()->getName());
+                Reserve(F->getProto()->getName(), true, F->getProto()->isOverload());
                 if (sameIdentifier(F->getProto()->getName(), "MAIN")) {
                     if (Entry) reportError("Multiple MAIN definitions in %s and %s",
                                            Sources[EntryUnit].Path.c_str(), Sources[I].Path.c_str());
@@ -326,7 +328,7 @@ void CodeGen::compile(const std::vector<SourceUnit> &Sources) {
 
     if (!Builder->GetInsertBlock()->hasTerminator()) {
         Value *Status = ConstantInt::get(*TheContext, APInt(32, 0));
-        if (Entry) {
+        if (Entry && !HadError) {
             Value *Result = Exceptions->emitCall(TheModule->getFunction("__cps_main"), {}, "main_result");
             Status = Builder->CreateTrunc(Result, Type::getInt32Ty(*TheContext));
         }
@@ -356,16 +358,13 @@ void CodeGen::print(raw_ostream &Output) {
 // Collects the arguments for a call to a user FUNCTION/PROCEDURE, dispatching
 // on the *declared* signature: BYREF parameters take the variable's alloca,
 // BYVAL parameters are evaluated and coerced to the declared parameter type.
-bool CodeGen::marshalCallArgs(const std::string &Callee,
+bool CodeGen::marshalCallArgs(const FuncSig &Signature,
                               const std::vector<std::unique_ptr<ExprAST>> &ArgExprs,
                               int Line,
                               std::vector<Value*> &Out) {
     SaveAndRestore<int> Location(DiagnosticLine, Line ? Line : DiagnosticLine);
-    const FuncSig *Sig = FuncGen->getSignature(Callee);
-    if (!Sig) {
-        reportError("Call to undefined function %s", Callee.c_str());
-        return false;
-    }
+    const FuncSig *Sig = &Signature;
+    const std::string &Callee = Sig->Name;
     if (ArgExprs.size() != Sig->Params.size()) {
         reportError("Incorrect # arguments passed to %s", Callee.c_str());
         return false;
@@ -424,3 +423,45 @@ bool CodeGen::marshalCallArgs(const std::string &Callee,
     }
     return true;
 }
+
+// Resolve using static source types only; argument evaluation belongs to marshalling.
+const FuncSig *CodeGen::resolveCall(const std::string &Name,
+                                  const std::vector<std::unique_ptr<ExprAST>> &Args) {
+    auto Overloads = FuncGen->getOverloads(Name);
+    if (Overloads.empty()) {
+        reportError("Call to undefined function %s", Name.c_str());
+        return nullptr;
+    }
+    if (Overloads.size() == 1 && !Overloads.front()->IsOverload) return Overloads.front();
+    struct ArgumentType { const TypeInfo *Type; int Rank; };
+    std::vector<ArgumentType> Actual;
+    std::string Key;
+    for (const auto &Arg : Args) {
+        const ArrayMetadata *Array = nullptr;
+        if (auto *Var = dynamic_cast<VariableExprAST *>(Arg.get())) {
+            const SymbolInfo *Symbol = getSymbolInfo(Var->getName());
+            if (Symbol && Symbol->IsArray) Array = Arrays->getMetadata(Var->getName());
+        }
+        const TypeInfo *Type = Array ? Types->resolve(Array->ElementTypeName) : getExprTypeInfo(Arg.get());
+        int Rank = Array ? Array->Rank : 0;
+        Actual.push_back({Type, Rank});
+        if (!Key.empty()) Key += ", ";
+        if (Rank) Key += "ARRAY" + std::to_string(Rank) + " OF ";
+        Key += Type ? Type->Name : "?";
+    }
+    for (const FuncSig *Sig : Overloads) {
+        if (Sig->Params.size() != Actual.size()) continue;
+        bool Match = true;
+        for (size_t I = 0; I < Actual.size(); ++I) {
+            const ParamSig &P = Sig->Params[I];
+            if (Actual[I].Rank != (P.IsArray ? P.Rank : 0) || !Actual[I].Type ||
+                !Actual[I].Type->sameType(Types->resolve(P.TypeName))) {
+                Match = false;
+                break;
+            }
+        }
+        if (Match) return Sig;
+    }
+    reportError("No matching overload for %s(%s)", Name.c_str(), Key.c_str());
+    return nullptr;
+}
diff --git a/lib/CodeGen/ConstantGen.cc b/lib/CodeGen/ConstantGen.cc
index 6e1ec57..fc5d178 100644
--- a/lib/CodeGen/ConstantGen.cc
+++ b/lib/CodeGen/ConstantGen.cc
@@ -60,7 +60,7 @@ void CodeGen::emitConstantStmt(ConstantStmtAST *Stmt) {
         return;
     }
     if (ConstantScopes.back().ID == 0 &&
-        (FuncGen->getSignature(Name) || isBuiltinName(Name))) {
+        (FuncGen->hasFunction(Name) || isBuiltinName(Name))) {
         reportError("Constant %s conflicts with a function name", Name.c_str());
         return;
     }
diff --git a/lib/CodeGen/Designator.cc b/lib/CodeGen/Designator.cc
index b31492b..540059d 100644
--- a/lib/CodeGen/Designator.cc
+++ b/lib/CodeGen/Designator.cc
@@ -97,8 +97,9 @@ Value *CodeGen::emitMethodCall(const std::string &Class, Value *Receiver,
         return nullptr;
     }
     std::vector<Value*> Args = {Receiver};
-    if (!marshalCallArgs(Method->getLinkName(), ArgExprs, Line, Args)) return nullptr;
-    return FuncGen->emitCall(Method->getLinkName(), Args);
+    const FuncSig *Sig = resolveCall(Method->getLinkName(), ArgExprs);
+    if (!Sig || !marshalCallArgs(*Sig, ArgExprs, Line, Args)) return nullptr;
+    return FuncGen->emitCall(Sig->LinkName, Args);
 }
 
 // Resolves a designator (or legacy variable/array-access node) to the address
diff --git a/lib/CodeGen/ExpressionGen.cc b/lib/CodeGen/ExpressionGen.cc
index ba5e329..b667e6a 100644
--- a/lib/CodeGen/ExpressionGen.cc
+++ b/lib/CodeGen/ExpressionGen.cc
@@ -299,8 +299,9 @@ Value *CodeGen::emitCall(const std::string &Callee,
     }
 
     std::vector<Value*> Args;
-    if (!marshalCallArgs(Name, ArgExprs, Line, Args)) return nullptr;
-    return FuncGen->emitCall(Name, Args);
+    const FuncSig *Sig = resolveCall(Name, ArgExprs);
+    if (!Sig || !marshalCallArgs(*Sig, ArgExprs, Line, Args)) return nullptr;
+    return FuncGen->emitCall(Sig->LinkName, Args);
 }
 
 Value *CodeGen::emitAllocationSize(ExprAST *Expr) {
diff --git a/lib/CodeGen/FunctionGen.cc b/lib/CodeGen/FunctionGen.cc
index d985bce..b4b37d3 100644
--- a/lib/CodeGen/FunctionGen.cc
+++ b/lib/CodeGen/FunctionGen.cc
@@ -104,13 +104,28 @@ void FunctionGen::createArgumentAllocas(Function *F, const std::vector<ParamDecl
     }
 }
 
-const FuncSig *FunctionGen::getSignature(const std::string &Name) const {
+std::vector<const FuncSig *> FunctionGen::getOverloads(const std::string &Name) const {
+    std::vector<const FuncSig *> Result;
     auto It = Signatures.find(sameIdentifier(Name, "MAIN") ? "__cps_main" : Name);
-    return It == Signatures.end() || (Visible && !Visible(It->second.OwnerUnit))
-               ? nullptr : &It->second;
+    if (It != Signatures.end())
+        for (const FuncSig &Sig : It->second)
+            if (!Visible || Visible(Sig.OwnerUnit)) Result.push_back(&Sig);
+    return Result;
+}
+
+const FuncSig *FunctionGen::getSignatureByLink(const std::string &Name) const {
+    auto It = LinkIndex.find(Name);
+    if (It == LinkIndex.end()) return nullptr;
+    return &Signatures.at(It->second.first)[It->second.second];
 }
 
 Function *FunctionGen::emitPrototype(PrototypeAST *Proto) {
+    auto Registered = PrototypeLinks.find(Proto);
+    if (Registered != PrototypeLinks.end()) return Module.getFunction(Registered->second);
+    if (Proto->isOverload() && Proto->getOwnerClass().empty() && sameIdentifier(Proto->getName(), "MAIN")) {
+        reportError("MAIN cannot be declared OVERLOAD");
+        return nullptr;
+    }
     if (Proto->getOwnerClass().empty()) {
         if (isReservedRuntimeName(Proto->getName())) {
             reportError("'%s' is a reserved runtime symbol and cannot be a FUNCTION/PROCEDURE name",
@@ -137,6 +152,8 @@ Function *FunctionGen::emitPrototype(PrototypeAST *Proto) {
     FuncSig Sig;
     Sig.OwnerUnit = CurrentUnit;
     Sig.Name = Proto->getLinkName();
+    Sig.IsOverload = Proto->isOverload();
+    Sig.Prototype = Proto;
     Sig.ReturnTypeName = Proto->getReturnType();
 
     std::vector<Type*> ArgTypes;
@@ -167,29 +184,38 @@ Function *FunctionGen::emitPrototype(PrototypeAST *Proto) {
             }
             ArgTypes.push_back(T);
         }
+        if (!Sig.ParamKey.empty()) Sig.ParamKey += ",";
+        if (P.IsArray) Sig.ParamKey += "ARRAY" + std::to_string(P.Rank) + " OF ";
+        Sig.ParamKey += Info ? (Info->isC() ? "C:" + Info->CIdentity : Info->Name) : P.TypeName;
         Sig.Params.push_back(std::move(PS));
     }
 
     Type *RetType = getLLVMType(Proto->getReturnType());
     FunctionType *FT = FunctionType::get(RetType, ArgTypes, false);
-    const FuncSig *Existing = getSignature(Proto->getLinkName());
-    Function *F = Module.getFunction(Existing ? Existing->Name : Proto->getLinkName());
-    if (F && F->getFunctionType() != FT) {
-        reportError("Function %s conflicts with an existing declaration of a different type",
-                Proto->getName().c_str());
-        return nullptr;
-    }
-    if (!F) {
-        F = Function::Create(FT,
-            Proto->getOwnerClass().empty() && sameIdentifier(Proto->getName(), "MAIN")
-                ? Function::InternalLinkage : Function::ExternalLinkage,
-            Proto->getLinkName(), &Module);
+    auto &Overloads = Signatures[Sig.Name];
+    for (const FuncSig &Existing : Overloads) {
+        if (!Sig.IsOverload || !Existing.IsOverload) {
+            reportError("Function %s cannot be redefined. Multiple definition of %s; every overload must be declared OVERLOAD",
+                        Proto->getName().c_str(), Proto->getName().c_str());
+            return nullptr;
+        }
+        if (sameIdentifier(Existing.ParamKey, Sig.ParamKey)) {
+            if (Existing.OwnerUnit != CurrentUnit)
+                reportError("Duplicate definition %s in %s and %s", Sig.Name.c_str(),
+                            UnitPaths[Existing.OwnerUnit].c_str(), UnitPaths[CurrentUnit].c_str());
+            else reportError("Function %s cannot be redefined", Proto->getName().c_str());
+            return nullptr;
+        }
     }
+    Sig.LinkName = Sig.Name + (Sig.IsOverload ? "(" + Sig.ParamKey + ")" : "");
+    Function *F = Function::Create(FT,
+        Proto->getOwnerClass().empty() && sameIdentifier(Proto->getName(), "MAIN")
+            ? Function::InternalLinkage : Function::ExternalLinkage,
+        Sig.LinkName, &Module);
     Exceptions.prepareFunction(F);
-
-    // Keep the first registration: on a duplicate definition the first one
-    // wins and the second is rejected by emitFunctionDef.
-    Signatures.emplace(Proto->getLinkName(), std::move(Sig));
+    PrototypeLinks.emplace(Proto, Sig.LinkName);
+    LinkIndex.emplace(Sig.LinkName, std::make_pair(Sig.Name, Overloads.size()));
+    Overloads.push_back(std::move(Sig));
 
     Function::arg_iterator AI = F->arg_begin();
     if (!Proto->getOwnerClass().empty()) (AI++)->setName("this");
@@ -280,8 +306,7 @@ static Value *GenerateCall(llvm::Module &Module,
 }
 
 llvm::Value *FunctionGen::emitCall(const std::string &Callee, const std::vector<llvm::Value*> &Args) {
-    const FuncSig *Sig = getSignature(Callee);
-    return GenerateCall(Module, Exceptions, Sig ? Sig->Name : Callee, Args, HadError, DiagnosticLine);
+    return GenerateCall(Module, Exceptions, Callee, Args, HadError, DiagnosticLine);
 }
 
 void FunctionGen::emitReturn(ReturnStmtAST *Ret, llvm::Value *RetVal) {
diff --git a/lib/CodeGen/GlobalGen.cc b/lib/CodeGen/GlobalGen.cc
index 659ae85..46b6f0c 100644
--- a/lib/CodeGen/GlobalGen.cc
+++ b/lib/CodeGen/GlobalGen.cc
@@ -34,7 +34,7 @@ void CodeGen::runGlobalPrePass(const SourceUnit &Unit, int Phase) {
                     continue;
                 }
                 if (Exports[CurrentUnit].count(Name) || Types->resolve(Name) ||
-                    Types->lookupEnumConstant(Name) || FuncGen->getSignature(Name) ||
+                    Types->lookupEnumConstant(Name) || FuncGen->hasFunction(Name) ||
                     isBuiltinName(Name)) {
                     reportError("Global %s conflicts with an existing definition", Name.c_str());
                     continue;
diff --git a/lib/CodeGen/StatementGen.cc b/lib/CodeGen/StatementGen.cc
index 1a9c285..8783cde 100644
--- a/lib/CodeGen/StatementGen.cc
+++ b/lib/CodeGen/StatementGen.cc
@@ -100,7 +100,7 @@ void CodeGen::emitStmt(StmtAST *Stmt) {
     if (auto *Ret = dynamic_cast<ReturnStmtAST*>(Stmt)) {
         Function *F = Builder->GetInsertBlock()->getParent();
         const FuncSig *Sig = F->getName() == "main" ? nullptr
-            : FuncGen->getSignature(F->getName().str());
+            : FuncGen->getSignatureByLink(F->getName().str());
         // The generated wrapper has no signature; user MAIN uses __cps_main.
         if (!Sig) {
             reportError("RETURN is only allowed inside a FUNCTION or PROCEDURE");
diff --git a/lib/CodeGen/ValueTypeGen.cc b/lib/CodeGen/ValueTypeGen.cc
index 5e926cf..d2bba03 100644
--- a/lib/CodeGen/ValueTypeGen.cc
+++ b/lib/CodeGen/ValueTypeGen.cc
@@ -8,7 +8,7 @@ using namespace cps;
 
 using namespace cps::detail;
 
-const TypeInfo *CodeGen::getExprTypeInfo(ExprAST *Expr) const {
+const TypeInfo *CodeGen::getExprTypeInfo(ExprAST *Expr) {
     if (!Expr) return nullptr;
 
     if (dynamic_cast<IntegerExprAST*>(Expr)) return resolveType("INTEGER");
@@ -129,20 +129,11 @@ const TypeInfo *CodeGen::getExprTypeInfo(ExprAST *Expr) const {
         if (const BuiltinInfo *B = findBuiltin(Name)) {
             return resolveType(B->ReturnTypeName);
         }
-        if (const FuncSig *Sig = FuncGen->getSignature(Name)) {
+        if (const FuncSig *Sig = resolveCall(Name, Call->getArgs())) {
             return resolveType(Sig->ReturnTypeName);
         }
 
-        Function *CalleeF = TheModule->getFunction(Name);
-        if (!CalleeF) return resolveType("INTEGER");
-
-        Type *RetTy = CalleeF->getReturnType();
-        if (RetTy->isVoidTy()) return resolveType("VOID");
-        if (RetTy->isDoubleTy()) return resolveType("REAL");
-        if (RetTy->isIntegerTy(1)) return resolveType("BOOLEAN");
-        if (RetTy->isIntegerTy(8)) return resolveType("CHAR");
-        if (RetTy->isPointerTy()) return resolveType("STRING");
-        return resolveType("INTEGER");
+        return nullptr;
     }
 
     return nullptr;