#include "cps/Diagnostics.h" #include "cps/Parser.h" #include "cps/FunctionAST.h" #include "cps/Operator.h" using namespace cps; // Array-parameter bounds live in the signature, not the AST, so they must be // integer literals (an optional leading '-' is allowed). bool Parser::ParseSignedIntBound(int64_t &Out) { bool Negative = false; if (CurTok == '-') { Negative = true; getNextToken(); } if (CurTok != tok_number_int) { cps::sourceError(Lex.getLine(), "Array parameter bounds must be integer literals"); return false; } return ParseIntegerValue(Negative, Out); } // ( [BYREF|BYVAL] : // | : ARRAY OF -- rank 1, bounds from the argument // | : ARRAY[:,...] OF , ... ) // The pass mode is sticky across parameters (guide §8.3: "the BYVAL or BYREF // keyword need not be repeated"); unannotated parameters default to BYVAL // (§8.1). BYREF is procedure-only (§8.3), gated by AllowByRef. std::optional> Parser::ParsePrototypeArgs(bool AllowByRef) { std::vector Params; if (CurTok != '(') return Params; getNextToken(); PassMode Current = PassMode::ByVal; while (CurTok != ')') { if (CurTok == tok_byref) { if (!AllowByRef) { cps::sourceError(Lex.getLine(), "BYREF parameters are not allowed in a FUNCTION (procedures only)"); return std::nullopt; } Current = PassMode::ByRef; getNextToken(); } else if (CurTok == tok_byval) { Current = PassMode::ByVal; getNextToken(); } if (CurTok != tok_identifier) { cps::sourceError(Lex.getLine(), "Expected argument name"); return std::nullopt; } ParamDecl P; P.Line = Lex.getLine(); P.Name = Lex.IdentifierStr; P.Mode = Current; getNextToken(); if (CurTok != tok_colon) { cps::sourceError(Lex.getLine(), "Expected ':' after argument name"); return std::nullopt; } getNextToken(); if (CurTok == tok_array) { getNextToken(); P.IsArray = true; if (CurTok == '[') { getNextToken(); while (true) { int64_t Lo = 0, Hi = 0; if (!ParseSignedIntBound(Lo)) return std::nullopt; if (CurTok != tok_colon) { cps::sourceError(Lex.getLine(), "Expected ':' between array parameter bounds"); return std::nullopt; } getNextToken(); if (!ParseSignedIntBound(Hi)) return std::nullopt; P.DeclaredBounds.emplace_back(Lo, Hi); if (CurTok == ']') break; if (CurTok != ',') { cps::sourceError(Lex.getLine(), "Expected ',' or ']' in array parameter bounds"); return std::nullopt; } getNextToken(); } getNextToken(); // eat ']' P.Rank = static_cast(P.DeclaredBounds.size()); } else { P.Rank = 1; // ARRAY OF T: bounds travel with the argument } if (CurTok != tok_of) { cps::sourceError(Lex.getLine(), "Expected OF after ARRAY in parameter '%s'", P.Name.c_str()); return std::nullopt; } getNextToken(); } P.TypeName = ParseTypeName(false); if (P.TypeName.empty()) { cps::sourceError(Lex.getLine(), "Expected argument type for '%s'", P.Name.c_str()); return std::nullopt; } if (CurTok == tok_ellipsis) { if (P.IsArray || !isTemplatePack(P.TypeName)) { sourceError(P.Line, "A pack parameter must use the current template pack and cannot be ARRAY"); return std::nullopt; } P.IsPack = true; getNextToken(); if (CurTok != ')') { sourceError(P.Line, "A pack parameter must be last"); return std::nullopt; } } else if (isTemplatePack(P.TypeName)) { sourceError(P.Line, "A template parameter pack requires '...'"); return std::nullopt; } for (const ParamDecl &Prev : Params) { if (sameIdentifier(Prev.Name, P.Name)) { cps::sourceError(Lex.getLine(), "Duplicate parameter name '%s'", P.Name.c_str()); return std::nullopt; } } Params.push_back(std::move(P)); if (CurTok == ')') break; if (CurTok != ',') { cps::sourceError(Lex.getLine(), "Expected ',' or ')'"); return std::nullopt; } getNextToken(); if (CurTok == ')') { cps::sourceError(Lex.getLine(), "Expected another parameter after ','"); return std::nullopt; } } getNextToken(); return Params; } std::string Parser::ParseOperatorName(int *OpOut) { getNextToken(); // OPERATOR int Op = CurTok; if (OpOut) *OpOut = Op; if (operatorSpelling(Op).empty()) { sourceError(Lex.getLine(), "Expected a supported operator after OPERATOR"); getNextToken(); return {}; } getNextToken(); if (CurTok != '(') { sourceError(Lex.getLine(), "Expected '(' after operator name"); return {}; } return operatorName(Op); } // A callee is an identifier or OPERATOR ; false means a diagnosed failure. bool Parser::ParseCalleeName(std::string &Name) { if (CurTok == tok_operator) { Name = ParseOperatorName(); return !Name.empty(); } Name = Lex.IdentifierStr; getNextToken(); return true; } std::unique_ptr Parser::ParseFunction(const std::string &OwnerClass, bool IsPublic, bool IsOverload) { int Line = Lex.getLine(); getNextToken(); if (CurTok == tok_delete || (CurTok == tok_identifier && sameIdentifier(Lex.IdentifierStr, "DELETE"))) { cps::sourceError(Line, "destructor DELETE must be a PROCEDURE"); return nullptr; } if (CurTok == tok_new || (CurTok == tok_identifier && sameIdentifier(Lex.IdentifierStr, "NEW"))) { cps::sourceError(Line, "constructor NEW must be a PROCEDURE"); return nullptr; } if (CurTok != tok_identifier && CurTok != tok_operator) { cps::sourceError(Lex.getLine(), "Expected function name"); return nullptr; } int Operator = 0; std::string Name; if (CurTok == tok_operator) { Name = ParseOperatorName(&Operator); if (Name.empty()) return nullptr; if (!IsOverload) { sourceError(Line, "An operator function must be declared OVERLOAD"); return nullptr; } } else { Name = Lex.IdentifierStr; getNextToken(); } auto Args = ParsePrototypeArgs(/*AllowByRef=*/false); if (!Args) return nullptr; std::string RetType = "INTEGER"; if (CurTok == tok_returns) { getNextToken(); if (CurTok == tok_array) { cps::sourceError(Lex.getLine(), "A FUNCTION cannot RETURN an array; use a BYREF PROCEDURE parameter instead"); return nullptr; } RetType = ParseTypeName(true); if (RetType.empty()) { cps::sourceError(Lex.getLine(), "Expected return type for function '%s'", Name.c_str()); return nullptr; } } auto Proto = std::make_unique(Name, std::move(*Args), RetType, false, OwnerClass, IsPublic, IsOverload); Proto->setOperator(Operator); return ParseFunctionBody(std::move(Proto), tok_endfunction, "expected ENDFUNCTION", Line); } std::unique_ptr Parser::ParseProcedure(const std::string &OwnerClass, bool IsPublic, bool IsOverload) { int Line = Lex.getLine(); getNextToken(); if (CurTok != tok_identifier && CurTok != tok_new && CurTok != tok_delete && CurTok != tok_operator) { cps::sourceError(Lex.getLine(), "Expected procedure name"); return nullptr; } int Operator = 0; bool IsOperator = CurTok == tok_operator; std::string OperatorName; if (IsOperator) { OperatorName = ParseOperatorName(&Operator); if (OperatorName.empty()) return nullptr; if (Operator != tok_assign || OwnerClass.empty() || !IsOverload) { sourceError(Line, "Assignment operator must be an OVERLOAD member PROCEDURE"); return nullptr; } } std::string Name = IsOperator ? OperatorName : CurTok == tok_new ? "NEW" : CurTok == tok_delete ? "DELETE" : Lex.IdentifierStr; if (OwnerClass.empty() && sameIdentifier(Name, "DELETE")) { cps::sourceError(Line, "destructor DELETE is only allowed in a CLASS"); return nullptr; } if (OwnerClass.empty() && sameIdentifier(Name, "NEW")) { cps::sourceError(Line, "constructor NEW is only allowed in a CLASS"); return nullptr; } if (!IsOperator) getNextToken(); auto Args = ParsePrototypeArgs(/*AllowByRef=*/true); if (!Args) return nullptr; if (sameIdentifier(Name, "DELETE") && !Args->empty()) { cps::sourceError(Line, "destructor DELETE must have no parameters"); return nullptr; } auto Proto = std::make_unique(Name, std::move(*Args), "VOID", false, OwnerClass, IsPublic, IsOverload); Proto->setOperator(Operator); return ParseFunctionBody(std::move(Proto), tok_endprocedure, "expected ENDPROCEDURE", Line); } // Shared FUNCTION/PROCEDURE tail: the body (with the prototype's pack // parameters in scope) up to EndToken, which must be present. std::unique_ptr Parser::ParseFunctionBody( std::unique_ptr Proto, int EndToken, const char *EndMessage, int Line) { auto SavedPacks = std::move(PackNames); PackNames.clear(); for (const auto &P : Proto->getParams()) if (P.IsPack) PackNames.insert(P.Name); std::vector> Body = ParseBlock({EndToken}); PackNames = std::move(SavedPacks); if (CurTok != EndToken) { cps::sourceError(Lex.getLine(), "%s", EndMessage); return nullptr; } getNextToken(); auto Def = std::make_unique(std::move(Proto), std::move(Body)); Def->setLine(Line); return Def; } // EXTERN "C" EXTERN_FUNCTION (, ... [, ...]) RETURNS // EXTERN "C" EXTERN_PROCEDURE (, ... [, ...]) // Parameters are unnamed; a trailing ... makes the prototype variadic. std::unique_ptr Parser::ParseExternDecl() { int Line = Lex.getLine(); getNextToken(); if (CurTok != tok_string_literal || Lex.StringVal != "C") { cps::sourceError(Lex.getLine(), "EXTERN must be followed by \"C\""); return nullptr; } getNextToken(); if (CurTok != tok_extern_function && CurTok != tok_extern_procedure) { cps::sourceError(Lex.getLine(), "Expected EXTERN_FUNCTION or EXTERN_PROCEDURE after EXTERN \"C\""); return nullptr; } bool IsFunction = CurTok == tok_extern_function; getNextToken(); if (CurTok != tok_identifier) { cps::sourceError(Lex.getLine(), "Expected C function name"); return nullptr; } std::string Name = Lex.IdentifierStr; getNextToken(); std::vector Params; bool IsVariadic = false; if (CurTok != '(') { cps::sourceError(Lex.getLine(), "Expected '(' after C function name '%s'", Name.c_str()); return nullptr; } getNextToken(); while (CurTok != ')') { if (CurTok == tok_ellipsis) { IsVariadic = true; getNextToken(); if (CurTok != ')') { cps::sourceError(Lex.getLine(), "'...' must be the last parameter"); return nullptr; } break; } if (CurTok == tok_byref || CurTok == tok_byval) { cps::sourceError(Lex.getLine(), "EXTERN parameters cannot be BYREF or BYVAL; use a C pointer type"); return nullptr; } ParamDecl P; P.Line = Lex.getLine(); P.TypeName = ParseTypeName(false); if (P.TypeName.empty()) return nullptr; P.Name = "arg" + std::to_string(Params.size() + 1); Params.push_back(std::move(P)); if (CurTok == ')') break; if (CurTok != ',') { cps::sourceError(Lex.getLine(), "Expected ',' or ')'"); return nullptr; } getNextToken(); if (CurTok == ')') { cps::sourceError(Lex.getLine(), "Expected another parameter after ','"); return nullptr; } } getNextToken(); std::string RetType = "VOID"; if (CurTok == tok_returns) { if (!IsFunction) { cps::sourceError(Lex.getLine(), "EXTERN_PROCEDURE cannot have RETURNS"); return nullptr; } getNextToken(); RetType = ParseTypeName(false); if (RetType.empty()) return nullptr; } else if (IsFunction) { cps::sourceError(Line, "EXTERN_FUNCTION '%s' requires RETURNS", Name.c_str()); return nullptr; } auto Proto = std::make_unique(Name, std::move(Params), RetType, /*IsExternal=*/true); Proto->setVariadic(IsVariadic); auto Decl = std::make_unique(std::move(Proto)); Decl->setLine(Line); return Decl; } std::unique_ptr Parser::ParseCallStmt() { int Line = Lex.getLine(); getNextToken(); if (CurTok != tok_identifier && CurTok != tok_operator) { cps::sourceError(Lex.getLine(), "Expected callee name after CALL"); return nullptr; } std::string Callee; if (!ParseCalleeName(Callee)) return nullptr; std::vector TemplateArgs; if (CurTok == '<') { auto Parsed = ParseTemplateArgs(); if (!Parsed) return nullptr; TemplateArgs = std::move(*Parsed); if (CurTok != '(') { sourceError(Line, "Expected '(' after explicit template arguments"); return nullptr; } } std::vector Accesses; if (!ParseDesignatorSuffix(Accesses)) return nullptr; if (!Accesses.empty()) { auto Call = ParseMemberCall(Callee, std::move(Accesses), Line); if (!Call) return nullptr; return std::make_unique(std::move(Call)); } std::vector> Args; if (CurTok == '(') { getNextToken(); auto Parsed = ParseExprList(')', true); if (!Parsed) return nullptr; Args = std::move(*Parsed); } if (CurTok == '.') { auto Result = ParseResultMethods(std::make_unique(Callee, std::move(Args), Line, std::move(TemplateArgs)), Line); if (!Result) return nullptr; return std::make_unique( std::unique_ptr(static_cast(Result.release()))); } return std::make_unique(Callee, std::move(Args), Line, std::move(TemplateArgs)); } std::unique_ptr Parser::ParseReturnStmt() { getNextToken(); std::unique_ptr Expr = nullptr; // A bare RETURN is recognized right before any block terminator. if (!atCaseBoundary() && CurTok != tok_endcase && CurTok != tok_otherwise && CurTok != tok_endif && CurTok != tok_else && CurTok != tok_endfunction && CurTok != tok_endprocedure && CurTok != tok_endwhile && CurTok != tok_until && CurTok != tok_next && CurTok != tok_catch && CurTok != tok_endtry && CurTok != tok_eof) { Expr = ParseExpression(); if (!Expr) return nullptr; } return std::make_unique(std::move(Expr)); }