1 #include <cctype>
2 #include <cstdio>
3 #include <map>
4 #include <memory>
5 #include <string>
6 #include <vector>
7 
8 namespace helper {
9 // Cloning make_unique here until it's standard in C++14.
10 // Using a namespace to avoid conflicting with MSVC's std::make_unique (which
11 // ADL can sometimes find in unqualified calls).
12 template <class T, class... Args>
13 static
14     typename std::enable_if<!std::is_array<T>::value, std::unique_ptr<T>>::type
15     make_unique(Args &&... args) {
16   return std::unique_ptr<T>(new T(std::forward<Args>(args)...));
17 }
18 }
19 
20 //===----------------------------------------------------------------------===//
21 // Lexer
22 //===----------------------------------------------------------------------===//
23 
24 // The lexer returns tokens [0-255] if it is an unknown character, otherwise one
25 // of these for known things.
26 enum Token {
27   tok_eof = -1,
28 
29   // commands
30   tok_def = -2,
31   tok_extern = -3,
32 
33   // primary
34   tok_identifier = -4,
35   tok_number = -5
36 };
37 
38 static std::string IdentifierStr; // Filled in if tok_identifier
39 static double NumVal;             // Filled in if tok_number
40 
41 /// gettok - Return the next token from standard input.
42 static int gettok() {
43   static int LastChar = ' ';
44 
45   // Skip any whitespace.
46   while (isspace(LastChar))
47     LastChar = getchar();
48 
49   if (isalpha(LastChar)) { // identifier: [a-zA-Z][a-zA-Z0-9]*
50     IdentifierStr = LastChar;
51     while (isalnum((LastChar = getchar())))
52       IdentifierStr += LastChar;
53 
54     if (IdentifierStr == "def")
55       return tok_def;
56     if (IdentifierStr == "extern")
57       return tok_extern;
58     return tok_identifier;
59   }
60 
61   if (isdigit(LastChar) || LastChar == '.') { // Number: [0-9.]+
62     std::string NumStr;
63     do {
64       NumStr += LastChar;
65       LastChar = getchar();
66     } while (isdigit(LastChar) || LastChar == '.');
67 
68     NumVal = strtod(NumStr.c_str(), nullptr);
69     return tok_number;
70   }
71 
72   if (LastChar == '#') {
73     // Comment until end of line.
74     do
75       LastChar = getchar();
76     while (LastChar != EOF && LastChar != '\n' && LastChar != '\r');
77 
78     if (LastChar != EOF)
79       return gettok();
80   }
81 
82   // Check for end of file.  Don't eat the EOF.
83   if (LastChar == EOF)
84     return tok_eof;
85 
86   // Otherwise, just return the character as its ascii value.
87   int ThisChar = LastChar;
88   LastChar = getchar();
89   return ThisChar;
90 }
91 
92 //===----------------------------------------------------------------------===//
93 // Abstract Syntax Tree (aka Parse Tree)
94 //===----------------------------------------------------------------------===//
95 namespace {
96 /// ExprAST - Base class for all expression nodes.
97 class ExprAST {
98 public:
99   virtual ~ExprAST() {}
100 };
101 
102 /// NumberExprAST - Expression class for numeric literals like "1.0".
103 class NumberExprAST : public ExprAST {
104   double Val;
105 
106 public:
107   NumberExprAST(double Val) : Val(Val) {}
108 };
109 
110 /// VariableExprAST - Expression class for referencing a variable, like "a".
111 class VariableExprAST : public ExprAST {
112   std::string Name;
113 
114 public:
115   VariableExprAST(const std::string &Name) : Name(Name) {}
116 };
117 
118 /// BinaryExprAST - Expression class for a binary operator.
119 class BinaryExprAST : public ExprAST {
120   char Op;
121   std::unique_ptr<ExprAST> LHS, RHS;
122 
123 public:
124   BinaryExprAST(char Op, std::unique_ptr<ExprAST> LHS,
125                 std::unique_ptr<ExprAST> RHS)
126       : Op(Op), LHS(std::move(LHS)), RHS(std::move(RHS)) {}
127 };
128 
129 /// CallExprAST - Expression class for function calls.
130 class CallExprAST : public ExprAST {
131   std::string Callee;
132   std::vector<std::unique_ptr<ExprAST>> Args;
133 
134 public:
135   CallExprAST(const std::string &Callee,
136               std::vector<std::unique_ptr<ExprAST>> Args)
137       : Callee(Callee), Args(std::move(Args)) {}
138 };
139 
140 /// PrototypeAST - This class represents the "prototype" for a function,
141 /// which captures its name, and its argument names (thus implicitly the number
142 /// of arguments the function takes).
143 class PrototypeAST {
144   std::string Name;
145   std::vector<std::string> Args;
146 
147 public:
148   PrototypeAST(const std::string &Name, std::vector<std::string> Args)
149       : Name(Name), Args(std::move(Args)) {}
150 };
151 
152 /// FunctionAST - This class represents a function definition itself.
153 class FunctionAST {
154   std::unique_ptr<PrototypeAST> Proto;
155   std::unique_ptr<ExprAST> Body;
156 
157 public:
158   FunctionAST(std::unique_ptr<PrototypeAST> Proto,
159               std::unique_ptr<ExprAST> Body)
160       : Proto(std::move(Proto)), Body(std::move(Body)) {}
161 };
162 } // end anonymous namespace
163 
164 //===----------------------------------------------------------------------===//
165 // Parser
166 //===----------------------------------------------------------------------===//
167 
168 /// CurTok/getNextToken - Provide a simple token buffer.  CurTok is the current
169 /// token the parser is looking at.  getNextToken reads another token from the
170 /// lexer and updates CurTok with its results.
171 static int CurTok;
172 static int getNextToken() { return CurTok = gettok(); }
173 
174 /// BinopPrecedence - This holds the precedence for each binary operator that is
175 /// defined.
176 static std::map<char, int> BinopPrecedence;
177 
178 /// GetTokPrecedence - Get the precedence of the pending binary operator token.
179 static int GetTokPrecedence() {
180   if (!isascii(CurTok))
181     return -1;
182 
183   // Make sure it's a declared binop.
184   int TokPrec = BinopPrecedence[CurTok];
185   if (TokPrec <= 0)
186     return -1;
187   return TokPrec;
188 }
189 
190 /// Error* - These are little helper functions for error handling.
191 std::unique_ptr<ExprAST> Error(const char *Str) {
192   fprintf(stderr, "Error: %s\n", Str);
193   return nullptr;
194 }
195 std::unique_ptr<PrototypeAST> ErrorP(const char *Str) {
196   Error(Str);
197   return nullptr;
198 }
199 
200 static std::unique_ptr<ExprAST> ParseExpression();
201 
202 /// numberexpr ::= number
203 static std::unique_ptr<ExprAST> ParseNumberExpr() {
204   auto Result = helper::make_unique<NumberExprAST>(NumVal);
205   getNextToken(); // consume the number
206   return std::move(Result);
207 }
208 
209 /// parenexpr ::= '(' expression ')'
210 static std::unique_ptr<ExprAST> ParseParenExpr() {
211   getNextToken(); // eat (.
212   auto V = ParseExpression();
213   if (!V)
214     return nullptr;
215 
216   if (CurTok != ')')
217     return Error("expected ')'");
218   getNextToken(); // eat ).
219   return V;
220 }
221 
222 /// identifierexpr
223 ///   ::= identifier
224 ///   ::= identifier '(' expression* ')'
225 static std::unique_ptr<ExprAST> ParseIdentifierExpr() {
226   std::string IdName = IdentifierStr;
227 
228   getNextToken(); // eat identifier.
229 
230   if (CurTok != '(') // Simple variable ref.
231     return helper::make_unique<VariableExprAST>(IdName);
232 
233   // Call.
234   getNextToken(); // eat (
235   std::vector<std::unique_ptr<ExprAST>> Args;
236   if (CurTok != ')') {
237     while (1) {
238       if (auto Arg = ParseExpression())
239         Args.push_back(std::move(Arg));
240       else
241         return nullptr;
242 
243       if (CurTok == ')')
244         break;
245 
246       if (CurTok != ',')
247         return Error("Expected ')' or ',' in argument list");
248       getNextToken();
249     }
250   }
251 
252   // Eat the ')'.
253   getNextToken();
254 
255   return helper::make_unique<CallExprAST>(IdName, std::move(Args));
256 }
257 
258 /// primary
259 ///   ::= identifierexpr
260 ///   ::= numberexpr
261 ///   ::= parenexpr
262 static std::unique_ptr<ExprAST> ParsePrimary() {
263   switch (CurTok) {
264   default:
265     return Error("unknown token when expecting an expression");
266   case tok_identifier:
267     return ParseIdentifierExpr();
268   case tok_number:
269     return ParseNumberExpr();
270   case '(':
271     return ParseParenExpr();
272   }
273 }
274 
275 /// binoprhs
276 ///   ::= ('+' primary)*
277 static std::unique_ptr<ExprAST> ParseBinOpRHS(int ExprPrec,
278                                               std::unique_ptr<ExprAST> LHS) {
279   // If this is a binop, find its precedence.
280   while (1) {
281     int TokPrec = GetTokPrecedence();
282 
283     // If this is a binop that binds at least as tightly as the current binop,
284     // consume it, otherwise we are done.
285     if (TokPrec < ExprPrec)
286       return LHS;
287 
288     // Okay, we know this is a binop.
289     int BinOp = CurTok;
290     getNextToken(); // eat binop
291 
292     // Parse the primary expression after the binary operator.
293     auto RHS = ParsePrimary();
294     if (!RHS)
295       return nullptr;
296 
297     // If BinOp binds less tightly with RHS than the operator after RHS, let
298     // the pending operator take RHS as its LHS.
299     int NextPrec = GetTokPrecedence();
300     if (TokPrec < NextPrec) {
301       RHS = ParseBinOpRHS(TokPrec + 1, std::move(RHS));
302       if (!RHS)
303         return nullptr;
304     }
305 
306     // Merge LHS/RHS.
307     LHS = helper::make_unique<BinaryExprAST>(BinOp, std::move(LHS),
308                                              std::move(RHS));
309   }
310 }
311 
312 /// expression
313 ///   ::= primary binoprhs
314 ///
315 static std::unique_ptr<ExprAST> ParseExpression() {
316   auto LHS = ParsePrimary();
317   if (!LHS)
318     return nullptr;
319 
320   return ParseBinOpRHS(0, std::move(LHS));
321 }
322 
323 /// prototype
324 ///   ::= id '(' id* ')'
325 static std::unique_ptr<PrototypeAST> ParsePrototype() {
326   if (CurTok != tok_identifier)
327     return ErrorP("Expected function name in prototype");
328 
329   std::string FnName = IdentifierStr;
330   getNextToken();
331 
332   if (CurTok != '(')
333     return ErrorP("Expected '(' in prototype");
334 
335   std::vector<std::string> ArgNames;
336   while (getNextToken() == tok_identifier)
337     ArgNames.push_back(IdentifierStr);
338   if (CurTok != ')')
339     return ErrorP("Expected ')' in prototype");
340 
341   // success.
342   getNextToken(); // eat ')'.
343 
344   return helper::make_unique<PrototypeAST>(FnName, std::move(ArgNames));
345 }
346 
347 /// definition ::= 'def' prototype expression
348 static std::unique_ptr<FunctionAST> ParseDefinition() {
349   getNextToken(); // eat def.
350   auto Proto = ParsePrototype();
351   if (!Proto)
352     return nullptr;
353 
354   if (auto E = ParseExpression())
355     return helper::make_unique<FunctionAST>(std::move(Proto), std::move(E));
356   return nullptr;
357 }
358 
359 /// toplevelexpr ::= expression
360 static std::unique_ptr<FunctionAST> ParseTopLevelExpr() {
361   if (auto E = ParseExpression()) {
362     // Make an anonymous proto.
363     auto Proto = helper::make_unique<PrototypeAST>("__anon_expr",
364                                                    std::vector<std::string>());
365     return helper::make_unique<FunctionAST>(std::move(Proto), std::move(E));
366   }
367   return nullptr;
368 }
369 
370 /// external ::= 'extern' prototype
371 static std::unique_ptr<PrototypeAST> ParseExtern() {
372   getNextToken(); // eat extern.
373   return ParsePrototype();
374 }
375 
376 //===----------------------------------------------------------------------===//
377 // Top-Level parsing
378 //===----------------------------------------------------------------------===//
379 
380 static void HandleDefinition() {
381   if (ParseDefinition()) {
382     fprintf(stderr, "Parsed a function definition.\n");
383   } else {
384     // Skip token for error recovery.
385     getNextToken();
386   }
387 }
388 
389 static void HandleExtern() {
390   if (ParseExtern()) {
391     fprintf(stderr, "Parsed an extern\n");
392   } else {
393     // Skip token for error recovery.
394     getNextToken();
395   }
396 }
397 
398 static void HandleTopLevelExpression() {
399   // Evaluate a top-level expression into an anonymous function.
400   if (ParseTopLevelExpr()) {
401     fprintf(stderr, "Parsed a top-level expr\n");
402   } else {
403     // Skip token for error recovery.
404     getNextToken();
405   }
406 }
407 
408 /// top ::= definition | external | expression | ';'
409 static void MainLoop() {
410   while (1) {
411     fprintf(stderr, "ready> ");
412     switch (CurTok) {
413     case tok_eof:
414       return;
415     case ';': // ignore top-level semicolons.
416       getNextToken();
417       break;
418     case tok_def:
419       HandleDefinition();
420       break;
421     case tok_extern:
422       HandleExtern();
423       break;
424     default:
425       HandleTopLevelExpression();
426       break;
427     }
428   }
429 }
430 
431 //===----------------------------------------------------------------------===//
432 // Main driver code.
433 //===----------------------------------------------------------------------===//
434 
435 int main() {
436   // Install standard binary operators.
437   // 1 is lowest precedence.
438   BinopPrecedence['<'] = 10;
439   BinopPrecedence['+'] = 20;
440   BinopPrecedence['-'] = 20;
441   BinopPrecedence['*'] = 40; // highest.
442 
443   // Prime the first token.
444   fprintf(stderr, "ready> ");
445   getNextToken();
446 
447   // Run the main "interpreter loop" now.
448   MainLoop();
449 
450   return 0;
451 }
452