Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 1 | #include "llvm/ADT/STLExtras.h" |
Chandler Carruth | d7cd9ac9 | 2014-01-13 09:53:45 +0000 | [diff] [blame] | 2 | #include "llvm/IR/Verifier.h" |
Chandler Carruth | 005f27a | 2013-01-02 11:56:33 +0000 | [diff] [blame] | 3 | #include "llvm/IR/DerivedTypes.h" |
| 4 | #include "llvm/IR/IRBuilder.h" |
| 5 | #include "llvm/IR/LLVMContext.h" |
| 6 | #include "llvm/IR/Module.h" |
Will Dietz | 981af00 | 2013-10-12 00:55:57 +0000 | [diff] [blame] | 7 | #include <cctype> |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 8 | #include <cstdio> |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 9 | #include <map> |
Chandler Carruth | 605e30e | 2012-12-04 10:16:57 +0000 | [diff] [blame] | 10 | #include <string> |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 11 | #include <vector> |
| 12 | using namespace llvm; |
| 13 | |
| 14 | //===----------------------------------------------------------------------===// |
| 15 | // Lexer |
| 16 | //===----------------------------------------------------------------------===// |
| 17 | |
| 18 | // The lexer returns tokens [0-255] if it is an unknown character, otherwise one |
| 19 | // of these for known things. |
| 20 | enum Token { |
| 21 | tok_eof = -1, |
| 22 | |
| 23 | // commands |
| 24 | tok_def = -2, tok_extern = -3, |
| 25 | |
| 26 | // primary |
| 27 | tok_identifier = -4, tok_number = -5 |
| 28 | }; |
| 29 | |
| 30 | static std::string IdentifierStr; // Filled in if tok_identifier |
| 31 | static double NumVal; // Filled in if tok_number |
| 32 | |
| 33 | /// gettok - Return the next token from standard input. |
| 34 | static int gettok() { |
| 35 | static int LastChar = ' '; |
| 36 | |
| 37 | // Skip any whitespace. |
| 38 | while (isspace(LastChar)) |
| 39 | LastChar = getchar(); |
| 40 | |
| 41 | if (isalpha(LastChar)) { // identifier: [a-zA-Z][a-zA-Z0-9]* |
| 42 | IdentifierStr = LastChar; |
| 43 | while (isalnum((LastChar = getchar()))) |
| 44 | IdentifierStr += LastChar; |
| 45 | |
| 46 | if (IdentifierStr == "def") return tok_def; |
| 47 | if (IdentifierStr == "extern") return tok_extern; |
| 48 | return tok_identifier; |
| 49 | } |
| 50 | |
| 51 | if (isdigit(LastChar) || LastChar == '.') { // Number: [0-9.]+ |
| 52 | std::string NumStr; |
| 53 | do { |
| 54 | NumStr += LastChar; |
| 55 | LastChar = getchar(); |
| 56 | } while (isdigit(LastChar) || LastChar == '.'); |
| 57 | |
| 58 | NumVal = strtod(NumStr.c_str(), 0); |
| 59 | return tok_number; |
| 60 | } |
| 61 | |
| 62 | if (LastChar == '#') { |
| 63 | // Comment until end of line. |
| 64 | do LastChar = getchar(); |
| 65 | while (LastChar != EOF && LastChar != '\n' && LastChar != '\r'); |
| 66 | |
| 67 | if (LastChar != EOF) |
| 68 | return gettok(); |
| 69 | } |
| 70 | |
| 71 | // Check for end of file. Don't eat the EOF. |
| 72 | if (LastChar == EOF) |
| 73 | return tok_eof; |
| 74 | |
| 75 | // Otherwise, just return the character as its ascii value. |
| 76 | int ThisChar = LastChar; |
| 77 | LastChar = getchar(); |
| 78 | return ThisChar; |
| 79 | } |
| 80 | |
| 81 | //===----------------------------------------------------------------------===// |
| 82 | // Abstract Syntax Tree (aka Parse Tree) |
| 83 | //===----------------------------------------------------------------------===// |
Juergen Ributzka | 05c5a93 | 2013-11-19 03:08:35 +0000 | [diff] [blame] | 84 | namespace { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 85 | /// ExprAST - Base class for all expression nodes. |
| 86 | class ExprAST { |
| 87 | public: |
Juergen Ributzka | 05c5a93 | 2013-11-19 03:08:35 +0000 | [diff] [blame] | 88 | virtual ~ExprAST() {} |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 89 | virtual Value *Codegen() = 0; |
| 90 | }; |
| 91 | |
| 92 | /// NumberExprAST - Expression class for numeric literals like "1.0". |
| 93 | class NumberExprAST : public ExprAST { |
| 94 | double Val; |
| 95 | public: |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 96 | NumberExprAST(double Val) : Val(Val) {} |
Alexander Kornienko | f817c1c | 2015-04-11 02:11:45 +0000 | [diff] [blame] | 97 | Value *Codegen() override; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 98 | }; |
| 99 | |
| 100 | /// VariableExprAST - Expression class for referencing a variable, like "a". |
| 101 | class VariableExprAST : public ExprAST { |
| 102 | std::string Name; |
| 103 | public: |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 104 | VariableExprAST(const std::string &Name) : Name(Name) {} |
Alexander Kornienko | f817c1c | 2015-04-11 02:11:45 +0000 | [diff] [blame] | 105 | Value *Codegen() override; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 106 | }; |
| 107 | |
| 108 | /// BinaryExprAST - Expression class for a binary operator. |
| 109 | class BinaryExprAST : public ExprAST { |
| 110 | char Op; |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 111 | std::unique_ptr<ExprAST> LHS, RHS; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 112 | public: |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 113 | BinaryExprAST(char Op, std::unique_ptr<ExprAST> LHS, |
| 114 | std::unique_ptr<ExprAST> RHS) |
| 115 | : Op(Op), LHS(std::move(LHS)), RHS(std::move(RHS)) {} |
Alexander Kornienko | f817c1c | 2015-04-11 02:11:45 +0000 | [diff] [blame] | 116 | Value *Codegen() override; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 117 | }; |
| 118 | |
| 119 | /// CallExprAST - Expression class for function calls. |
| 120 | class CallExprAST : public ExprAST { |
| 121 | std::string Callee; |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 122 | std::vector<std::unique_ptr<ExprAST>> Args; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 123 | public: |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 124 | CallExprAST(const std::string &Callee, |
| 125 | std::vector<std::unique_ptr<ExprAST>> Args) |
| 126 | : Callee(Callee), Args(std::move(Args)) {} |
Alexander Kornienko | f817c1c | 2015-04-11 02:11:45 +0000 | [diff] [blame] | 127 | Value *Codegen() override; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 128 | }; |
| 129 | |
| 130 | /// PrototypeAST - This class represents the "prototype" for a function, |
| 131 | /// which captures its name, and its argument names (thus implicitly the number |
| 132 | /// of arguments the function takes). |
| 133 | class PrototypeAST { |
| 134 | std::string Name; |
| 135 | std::vector<std::string> Args; |
| 136 | public: |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 137 | PrototypeAST(const std::string &Name, std::vector<std::string> Args) |
| 138 | : Name(Name), Args(std::move(Args)) {} |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 139 | |
| 140 | Function *Codegen(); |
| 141 | }; |
| 142 | |
| 143 | /// FunctionAST - This class represents a function definition itself. |
| 144 | class FunctionAST { |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 145 | std::unique_ptr<PrototypeAST> Proto; |
| 146 | std::unique_ptr<ExprAST> Body; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 147 | public: |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 148 | FunctionAST(std::unique_ptr<PrototypeAST> Proto, |
| 149 | std::unique_ptr<ExprAST> Body) |
| 150 | : Proto(std::move(Proto)), Body(std::move(Body)) {} |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 151 | |
| 152 | Function *Codegen(); |
| 153 | }; |
Juergen Ributzka | 05c5a93 | 2013-11-19 03:08:35 +0000 | [diff] [blame] | 154 | } // end anonymous namespace |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 155 | |
| 156 | //===----------------------------------------------------------------------===// |
| 157 | // Parser |
| 158 | //===----------------------------------------------------------------------===// |
| 159 | |
| 160 | /// CurTok/getNextToken - Provide a simple token buffer. CurTok is the current |
| 161 | /// token the parser is looking at. getNextToken reads another token from the |
| 162 | /// lexer and updates CurTok with its results. |
| 163 | static int CurTok; |
| 164 | static int getNextToken() { |
| 165 | return CurTok = gettok(); |
| 166 | } |
| 167 | |
| 168 | /// BinopPrecedence - This holds the precedence for each binary operator that is |
| 169 | /// defined. |
| 170 | static std::map<char, int> BinopPrecedence; |
| 171 | |
| 172 | /// GetTokPrecedence - Get the precedence of the pending binary operator token. |
| 173 | static int GetTokPrecedence() { |
| 174 | if (!isascii(CurTok)) |
| 175 | return -1; |
| 176 | |
| 177 | // Make sure it's a declared binop. |
| 178 | int TokPrec = BinopPrecedence[CurTok]; |
| 179 | if (TokPrec <= 0) return -1; |
| 180 | return TokPrec; |
| 181 | } |
| 182 | |
| 183 | /// Error* - These are little helper functions for error handling. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 184 | std::unique_ptr<ExprAST> Error(const char *Str) { |
| 185 | fprintf(stderr, "Error: %s\n", Str); |
| 186 | return nullptr; |
| 187 | } |
| 188 | std::unique_ptr<PrototypeAST> ErrorP(const char *Str) { |
| 189 | Error(Str); |
| 190 | return nullptr; |
| 191 | } |
| 192 | std::unique_ptr<FunctionAST> ErrorF(const char *Str) { |
| 193 | Error(Str); |
| 194 | return nullptr; |
| 195 | } |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 196 | |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 197 | static std::unique_ptr<ExprAST> ParseExpression(); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 198 | |
| 199 | /// identifierexpr |
| 200 | /// ::= identifier |
| 201 | /// ::= identifier '(' expression* ')' |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 202 | static std::unique_ptr<ExprAST> ParseIdentifierExpr() { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 203 | std::string IdName = IdentifierStr; |
| 204 | |
| 205 | getNextToken(); // eat identifier. |
| 206 | |
| 207 | if (CurTok != '(') // Simple variable ref. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 208 | return llvm::make_unique<VariableExprAST>(IdName); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 209 | |
| 210 | // Call. |
| 211 | getNextToken(); // eat ( |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 212 | std::vector<std::unique_ptr<ExprAST>> Args; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 213 | if (CurTok != ')') { |
| 214 | while (1) { |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 215 | if (auto Arg = ParseExpression()) |
| 216 | Args.push_back(std::move(Arg)); |
| 217 | else |
| 218 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 219 | |
| 220 | if (CurTok == ')') break; |
| 221 | |
| 222 | if (CurTok != ',') |
| 223 | return Error("Expected ')' or ',' in argument list"); |
| 224 | getNextToken(); |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | // Eat the ')'. |
| 229 | getNextToken(); |
| 230 | |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 231 | return llvm::make_unique<CallExprAST>(IdName, std::move(Args)); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 232 | } |
| 233 | |
| 234 | /// numberexpr ::= number |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 235 | static std::unique_ptr<ExprAST> ParseNumberExpr() { |
| 236 | auto Result = llvm::make_unique<NumberExprAST>(NumVal); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 237 | getNextToken(); // consume the number |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 238 | return std::move(Result); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 239 | } |
| 240 | |
| 241 | /// parenexpr ::= '(' expression ')' |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 242 | static std::unique_ptr<ExprAST> ParseParenExpr() { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 243 | getNextToken(); // eat (. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 244 | auto V = ParseExpression(); |
| 245 | if (!V) |
| 246 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 247 | |
| 248 | if (CurTok != ')') |
| 249 | return Error("expected ')'"); |
| 250 | getNextToken(); // eat ). |
| 251 | return V; |
| 252 | } |
| 253 | |
| 254 | /// primary |
| 255 | /// ::= identifierexpr |
| 256 | /// ::= numberexpr |
| 257 | /// ::= parenexpr |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 258 | static std::unique_ptr<ExprAST> ParsePrimary() { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 259 | switch (CurTok) { |
| 260 | default: return Error("unknown token when expecting an expression"); |
| 261 | case tok_identifier: return ParseIdentifierExpr(); |
| 262 | case tok_number: return ParseNumberExpr(); |
| 263 | case '(': return ParseParenExpr(); |
| 264 | } |
| 265 | } |
| 266 | |
| 267 | /// binoprhs |
| 268 | /// ::= ('+' primary)* |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 269 | static std::unique_ptr<ExprAST> ParseBinOpRHS(int ExprPrec, |
| 270 | std::unique_ptr<ExprAST> LHS) { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 271 | // If this is a binop, find its precedence. |
| 272 | while (1) { |
| 273 | int TokPrec = GetTokPrecedence(); |
| 274 | |
| 275 | // If this is a binop that binds at least as tightly as the current binop, |
| 276 | // consume it, otherwise we are done. |
| 277 | if (TokPrec < ExprPrec) |
| 278 | return LHS; |
| 279 | |
| 280 | // Okay, we know this is a binop. |
| 281 | int BinOp = CurTok; |
| 282 | getNextToken(); // eat binop |
| 283 | |
| 284 | // Parse the primary expression after the binary operator. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 285 | auto RHS = ParsePrimary(); |
| 286 | if (!RHS) return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 287 | |
| 288 | // If BinOp binds less tightly with RHS than the operator after RHS, let |
| 289 | // the pending operator take RHS as its LHS. |
| 290 | int NextPrec = GetTokPrecedence(); |
| 291 | if (TokPrec < NextPrec) { |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 292 | RHS = ParseBinOpRHS(TokPrec+1, std::move(RHS)); |
| 293 | if (!RHS) return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 294 | } |
| 295 | |
| 296 | // Merge LHS/RHS. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 297 | LHS = llvm::make_unique<BinaryExprAST>(BinOp, std::move(LHS), |
| 298 | std::move(RHS)); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 299 | } |
| 300 | } |
| 301 | |
| 302 | /// expression |
| 303 | /// ::= primary binoprhs |
| 304 | /// |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 305 | static std::unique_ptr<ExprAST> ParseExpression() { |
| 306 | auto LHS = ParsePrimary(); |
| 307 | if (!LHS) return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 308 | |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 309 | return ParseBinOpRHS(0, std::move(LHS)); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 310 | } |
| 311 | |
| 312 | /// prototype |
| 313 | /// ::= id '(' id* ')' |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 314 | static std::unique_ptr<PrototypeAST> ParsePrototype() { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 315 | if (CurTok != tok_identifier) |
| 316 | return ErrorP("Expected function name in prototype"); |
| 317 | |
| 318 | std::string FnName = IdentifierStr; |
| 319 | getNextToken(); |
| 320 | |
| 321 | if (CurTok != '(') |
| 322 | return ErrorP("Expected '(' in prototype"); |
| 323 | |
| 324 | std::vector<std::string> ArgNames; |
| 325 | while (getNextToken() == tok_identifier) |
| 326 | ArgNames.push_back(IdentifierStr); |
| 327 | if (CurTok != ')') |
| 328 | return ErrorP("Expected ')' in prototype"); |
| 329 | |
| 330 | // success. |
| 331 | getNextToken(); // eat ')'. |
| 332 | |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 333 | return llvm::make_unique<PrototypeAST>(std::move(FnName), |
| 334 | std::move(ArgNames)); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 335 | } |
| 336 | |
| 337 | /// definition ::= 'def' prototype expression |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 338 | static std::unique_ptr<FunctionAST> ParseDefinition() { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 339 | getNextToken(); // eat def. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 340 | auto Proto = ParsePrototype(); |
| 341 | if (!Proto) return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 342 | |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 343 | if (auto E = ParseExpression()) |
| 344 | return llvm::make_unique<FunctionAST>(std::move(Proto), std::move(E)); |
| 345 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 346 | } |
| 347 | |
| 348 | /// toplevelexpr ::= expression |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 349 | static std::unique_ptr<FunctionAST> ParseTopLevelExpr() { |
| 350 | if (auto E = ParseExpression()) { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 351 | // Make an anonymous proto. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 352 | auto Proto = llvm::make_unique<PrototypeAST>("", |
| 353 | std::vector<std::string>()); |
| 354 | return llvm::make_unique<FunctionAST>(std::move(Proto), std::move(E)); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 355 | } |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 356 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 357 | } |
| 358 | |
| 359 | /// external ::= 'extern' prototype |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 360 | static std::unique_ptr<PrototypeAST> ParseExtern() { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 361 | getNextToken(); // eat extern. |
| 362 | return ParsePrototype(); |
| 363 | } |
| 364 | |
| 365 | //===----------------------------------------------------------------------===// |
| 366 | // Code Generation |
| 367 | //===----------------------------------------------------------------------===// |
| 368 | |
| 369 | static Module *TheModule; |
| 370 | static IRBuilder<> Builder(getGlobalContext()); |
| 371 | static std::map<std::string, Value*> NamedValues; |
| 372 | |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 373 | Value *ErrorV(const char *Str) { Error(Str); return nullptr; } |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 374 | |
| 375 | Value *NumberExprAST::Codegen() { |
| 376 | return ConstantFP::get(getGlobalContext(), APFloat(Val)); |
| 377 | } |
| 378 | |
| 379 | Value *VariableExprAST::Codegen() { |
| 380 | // Look this variable up in the function. |
| 381 | Value *V = NamedValues[Name]; |
| 382 | return V ? V : ErrorV("Unknown variable name"); |
| 383 | } |
| 384 | |
| 385 | Value *BinaryExprAST::Codegen() { |
| 386 | Value *L = LHS->Codegen(); |
| 387 | Value *R = RHS->Codegen(); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 388 | if (!L || !R) return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 389 | |
| 390 | switch (Op) { |
Chris Lattner | 26d7950 | 2010-06-21 22:51:14 +0000 | [diff] [blame] | 391 | case '+': return Builder.CreateFAdd(L, R, "addtmp"); |
| 392 | case '-': return Builder.CreateFSub(L, R, "subtmp"); |
| 393 | case '*': return Builder.CreateFMul(L, R, "multmp"); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 394 | case '<': |
| 395 | L = Builder.CreateFCmpULT(L, R, "cmptmp"); |
| 396 | // Convert bool 0/1 to double 0.0 or 1.0 |
| 397 | return Builder.CreateUIToFP(L, Type::getDoubleTy(getGlobalContext()), |
| 398 | "booltmp"); |
| 399 | default: return ErrorV("invalid binary operator"); |
| 400 | } |
| 401 | } |
| 402 | |
| 403 | Value *CallExprAST::Codegen() { |
| 404 | // Look up the name in the global module table. |
| 405 | Function *CalleeF = TheModule->getFunction(Callee); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 406 | if (!CalleeF) |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 407 | return ErrorV("Unknown function referenced"); |
| 408 | |
| 409 | // If argument mismatch error. |
| 410 | if (CalleeF->arg_size() != Args.size()) |
| 411 | return ErrorV("Incorrect # arguments passed"); |
| 412 | |
| 413 | std::vector<Value*> ArgsV; |
| 414 | for (unsigned i = 0, e = Args.size(); i != e; ++i) { |
| 415 | ArgsV.push_back(Args[i]->Codegen()); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 416 | if (!ArgsV.back()) return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 417 | } |
| 418 | |
Francois Pichet | c5d1050 | 2011-07-15 10:59:52 +0000 | [diff] [blame] | 419 | return Builder.CreateCall(CalleeF, ArgsV, "calltmp"); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 420 | } |
| 421 | |
| 422 | Function *PrototypeAST::Codegen() { |
| 423 | // Make the function type: double(double,double) etc. |
John Wiegley | 2a0177b | 2011-07-11 22:39:46 +0000 | [diff] [blame] | 424 | std::vector<Type*> Doubles(Args.size(), |
| 425 | Type::getDoubleTy(getGlobalContext())); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 426 | FunctionType *FT = FunctionType::get(Type::getDoubleTy(getGlobalContext()), |
| 427 | Doubles, false); |
| 428 | |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 429 | Function *F = Function::Create(FT, Function::ExternalLinkage, Name, |
| 430 | TheModule); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 431 | |
| 432 | // If F conflicted, there was already something named 'Name'. If it has a |
| 433 | // body, don't allow redefinition or reextern. |
| 434 | if (F->getName() != Name) { |
| 435 | // Delete the one we just made and get the existing one. |
| 436 | F->eraseFromParent(); |
| 437 | F = TheModule->getFunction(Name); |
| 438 | |
| 439 | // If F already has a body, reject this. |
| 440 | if (!F->empty()) { |
| 441 | ErrorF("redefinition of function"); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 442 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 443 | } |
| 444 | |
| 445 | // If F took a different number of args, reject. |
| 446 | if (F->arg_size() != Args.size()) { |
| 447 | ErrorF("redefinition of function with different # args"); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 448 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 449 | } |
| 450 | } |
| 451 | |
| 452 | // Set names for all arguments. |
| 453 | unsigned Idx = 0; |
| 454 | for (Function::arg_iterator AI = F->arg_begin(); Idx != Args.size(); |
| 455 | ++AI, ++Idx) { |
| 456 | AI->setName(Args[Idx]); |
| 457 | |
| 458 | // Add arguments to variable symbol table. |
| 459 | NamedValues[Args[Idx]] = AI; |
| 460 | } |
| 461 | |
| 462 | return F; |
| 463 | } |
| 464 | |
| 465 | Function *FunctionAST::Codegen() { |
| 466 | NamedValues.clear(); |
| 467 | |
| 468 | Function *TheFunction = Proto->Codegen(); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 469 | if (!TheFunction) |
| 470 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 471 | |
| 472 | // Create a new basic block to start insertion into. |
| 473 | BasicBlock *BB = BasicBlock::Create(getGlobalContext(), "entry", TheFunction); |
| 474 | Builder.SetInsertPoint(BB); |
| 475 | |
| 476 | if (Value *RetVal = Body->Codegen()) { |
| 477 | // Finish off the function. |
| 478 | Builder.CreateRet(RetVal); |
| 479 | |
| 480 | // Validate the generated code, checking for consistency. |
| 481 | verifyFunction(*TheFunction); |
| 482 | |
| 483 | return TheFunction; |
| 484 | } |
| 485 | |
| 486 | // Error reading body, remove function. |
| 487 | TheFunction->eraseFromParent(); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 488 | return nullptr; |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 489 | } |
| 490 | |
| 491 | //===----------------------------------------------------------------------===// |
| 492 | // Top-Level parsing and JIT Driver |
| 493 | //===----------------------------------------------------------------------===// |
| 494 | |
| 495 | static void HandleDefinition() { |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 496 | if (auto FnAST = ParseDefinition()) { |
| 497 | if (auto *FnIR = FnAST->Codegen()) { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 498 | fprintf(stderr, "Read function definition:"); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 499 | FnIR->dump(); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 500 | } |
| 501 | } else { |
| 502 | // Skip token for error recovery. |
| 503 | getNextToken(); |
| 504 | } |
| 505 | } |
| 506 | |
| 507 | static void HandleExtern() { |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 508 | if (auto ProtoAST = ParseExtern()) { |
| 509 | if (auto *FnIR = ProtoAST->Codegen()) { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 510 | fprintf(stderr, "Read extern: "); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 511 | FnIR->dump(); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 512 | } |
| 513 | } else { |
| 514 | // Skip token for error recovery. |
| 515 | getNextToken(); |
| 516 | } |
| 517 | } |
| 518 | |
| 519 | static void HandleTopLevelExpression() { |
| 520 | // Evaluate a top-level expression into an anonymous function. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 521 | if (auto FnAST = ParseTopLevelExpr()) { |
| 522 | if (auto *FnIR = FnAST->Codegen()) { |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 523 | fprintf(stderr, "Read top-level expression:"); |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 524 | FnIR->dump(); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 525 | } |
| 526 | } else { |
| 527 | // Skip token for error recovery. |
| 528 | getNextToken(); |
| 529 | } |
| 530 | } |
| 531 | |
| 532 | /// top ::= definition | external | expression | ';' |
| 533 | static void MainLoop() { |
| 534 | while (1) { |
| 535 | fprintf(stderr, "ready> "); |
| 536 | switch (CurTok) { |
| 537 | case tok_eof: return; |
| 538 | case ';': getNextToken(); break; // ignore top-level semicolons. |
| 539 | case tok_def: HandleDefinition(); break; |
| 540 | case tok_extern: HandleExtern(); break; |
| 541 | default: HandleTopLevelExpression(); break; |
| 542 | } |
| 543 | } |
| 544 | } |
| 545 | |
| 546 | //===----------------------------------------------------------------------===// |
| 547 | // "Library" functions that can be "extern'd" from user code. |
| 548 | //===----------------------------------------------------------------------===// |
| 549 | |
| 550 | /// putchard - putchar that takes a double and returns 0. |
| 551 | extern "C" |
| 552 | double putchard(double X) { |
| 553 | putchar((char)X); |
| 554 | return 0; |
| 555 | } |
| 556 | |
| 557 | //===----------------------------------------------------------------------===// |
| 558 | // Main driver code. |
| 559 | //===----------------------------------------------------------------------===// |
| 560 | |
| 561 | int main() { |
| 562 | LLVMContext &Context = getGlobalContext(); |
| 563 | |
| 564 | // Install standard binary operators. |
| 565 | // 1 is lowest precedence. |
| 566 | BinopPrecedence['<'] = 10; |
| 567 | BinopPrecedence['+'] = 20; |
| 568 | BinopPrecedence['-'] = 20; |
| 569 | BinopPrecedence['*'] = 40; // highest. |
| 570 | |
| 571 | // Prime the first token. |
| 572 | fprintf(stderr, "ready> "); |
| 573 | getNextToken(); |
| 574 | |
| 575 | // Make the module, which holds all the code. |
Lang Hames | 09bf4c1 | 2015-08-18 18:11:06 +0000 | [diff] [blame^] | 576 | std::unique_ptr<Module> Owner = llvm::make_unique<Module>("my cool jit", Context); |
| 577 | TheModule = Owner.get(); |
Erick Tryzelaar | 21e83ea | 2009-09-22 21:15:19 +0000 | [diff] [blame] | 578 | |
| 579 | // Run the main "interpreter loop" now. |
| 580 | MainLoop(); |
| 581 | |
| 582 | // Print out all of the generated code. |
| 583 | TheModule->dump(); |
| 584 | |
| 585 | return 0; |
| 586 | } |