Lexer.h revision 448cec4c1c3705f6f49ffdefb58a7329942a2dd8
1//===--- Lexer.h - C Language Family Lexer ----------------------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file was developed by Chris Lattner and is distributed under 6// the University of Illinois Open Source License. See LICENSE.TXT for details. 7// 8//===----------------------------------------------------------------------===// 9// 10// This file defines the Lexer interface. 11// 12//===----------------------------------------------------------------------===// 13 14#ifndef LLVM_CLANG_LEXER_H 15#define LLVM_CLANG_LEXER_H 16 17#include "clang/Lex/Token.h" 18#include "clang/Lex/MultipleIncludeOpt.h" 19#include "clang/Basic/LangOptions.h" 20#include <string> 21#include <vector> 22#include <cassert> 23 24namespace clang { 25class Diagnostic; 26class Preprocessor; 27 28/// Lexer - This provides a simple interface that turns a text buffer into a 29/// stream of tokens. This provides no support for file reading or buffering, 30/// or buffering/seeking of tokens, only forward lexing is supported. It relies 31/// on the specified Preprocessor object to handle preprocessor directives, etc. 32class Lexer { 33 //===--------------------------------------------------------------------===// 34 // Constant configuration values for this lexer. 35 const char *BufferStart; // Start of the buffer. 36 const char *BufferEnd; // End of the buffer. 37 SourceLocation FileLoc; // Location for start of file. 38 Preprocessor &PP; // Preprocessor object controlling lexing. 39 LangOptions Features; // Features enabled by this language (cache). 40 bool Is_PragmaLexer; // True if lexer for _Pragma handling. 41 bool IsMainFile; // True if top-level file. 42 43 //===--------------------------------------------------------------------===// 44 // Context-specific lexing flags set by the preprocessor. 45 // 46 47 /// ParsingPreprocessorDirective - This is true when parsing #XXX. This turns 48 /// '\n' into a tok::eom token. 49 bool ParsingPreprocessorDirective; 50 51 /// ParsingFilename - True after #include: this turns <xx> into a 52 /// tok::angle_string_literal token. 53 bool ParsingFilename; 54 55 /// LexingRawMode - True if in raw mode: This flag disables interpretation of 56 /// tokens and is a far faster mode to lex in than non-raw-mode. This flag: 57 /// 1. If EOF of the current lexer is found, the include stack isn't popped. 58 /// 2. Identifier information is not looked up for identifier tokens. As an 59 /// effect of this, implicit macro expansion is naturally disabled. 60 /// 3. "#" tokens at the start of a line are treated as normal tokens, not 61 /// implicitly transformed by the lexer. 62 /// 4. All diagnostic messages are disabled, except for unterminated /*. 63 /// 5. The only callback made into the preprocessor is to report a hard error 64 /// on an unterminated '/*' comment. 65 bool LexingRawMode; 66 67 /// KeepCommentMode - The lexer can optionally keep C & BCPL-style comments, 68 /// and return them as tokens. This is used for -C and -CC modes. 69 bool KeepCommentMode; 70 71 //===--------------------------------------------------------------------===// 72 // Context that changes as the file is lexed. 73 // NOTE: any state that mutates when in raw mode must have save/restore code 74 // in Lexer::isNextPPTokenLParen. 75 76 // BufferPtr - Current pointer into the buffer. This is the next character 77 // to be lexed. 78 const char *BufferPtr; 79 80 // IsAtStartOfLine - True if the next lexed token should get the "start of 81 // line" flag set on it. 82 bool IsAtStartOfLine; 83 84 /// MIOpt - This is a state machine that detects the #ifndef-wrapping a file 85 /// idiom for the multiple-include optimization. 86 MultipleIncludeOpt MIOpt; 87 88 /// ConditionalStack - Information about the set of #if/#ifdef/#ifndef blocks 89 /// we are currently in. 90 std::vector<PPConditionalInfo> ConditionalStack; 91 92 Lexer(const Lexer&); // DO NOT IMPLEMENT 93 void operator=(const Lexer&); // DO NOT IMPLEMENT 94 friend class Preprocessor; 95public: 96 97 /// Lexer constructor - Create a new lexer object for the specified buffer 98 /// with the specified preprocessor managing the lexing process. This lexer 99 /// assumes that the associated MemoryBuffer and Preprocessor objects will 100 /// outlive it, so it doesn't take ownership of either of them. 101 Lexer(SourceLocation FileLoc, Preprocessor &PP, 102 const char *BufStart = 0, const char *BufEnd = 0); 103 104 /// getFeatures - Return the language features currently enabled. NOTE: this 105 /// lexer modifies features as a file is parsed! 106 const LangOptions &getFeatures() const { return Features; } 107 108 /// getFileLoc - Return the File Location for the file we are lexing out of. 109 /// The physical location encodes the location where the characters come from, 110 /// the virtual location encodes where we should *claim* the characters came 111 /// from. Currently this is only used by _Pragma handling. 112 SourceLocation getFileLoc() const { return FileLoc; } 113 114 /// setIsMainFile - Mark this lexer as being the lexer for the top-level 115 /// source file. 116 void setIsMainFile() { 117 IsMainFile = true; 118 } 119 120 /// isMainFile - Return true if this is the top-level file. 121 /// 122 bool isMainFile() const { return IsMainFile; } 123 124 /// Lex - Return the next token in the file. If this is the end of file, it 125 /// return the tok::eof token. Return true if an error occurred and 126 /// compilation should terminate, false if normal. This implicitly involves 127 /// the preprocessor. 128 void Lex(Token &Result) { 129 // Start a new token. 130 Result.startToken(); 131 132 // NOTE, any changes here should also change code after calls to 133 // Preprocessor::HandleDirective 134 if (IsAtStartOfLine) { 135 Result.setFlag(Token::StartOfLine); 136 IsAtStartOfLine = false; 137 } 138 139 // Get a token. Note that this may delete the current lexer if the end of 140 // file is reached. 141 LexTokenInternal(Result); 142 } 143 144 /// LexRawToken - Switch the lexer to raw mode, lex a token into Result and 145 /// switch it back. Return true if the 'next character to read' pointer 146 /// points and the end of the lexer buffer, false otherwise. 147 bool LexRawToken(Token &Result) { 148 assert(!LexingRawMode && "Already in raw mode!"); 149 LexingRawMode = true; 150 Lex(Result); 151 LexingRawMode = false; 152 return BufferPtr == BufferEnd; 153 } 154 155 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 156 /// uninterpreted string. This switches the lexer out of directive mode. 157 std::string ReadToEndOfLine(); 158 159 160 /// Diag - Forwarding function for diagnostics. This translate a source 161 /// position in the current buffer into a SourceLocation object for rendering. 162 void Diag(const char *Loc, unsigned DiagID, 163 const std::string &Msg = std::string()) const; 164 void Diag(SourceLocation Loc, unsigned DiagID, 165 const std::string &Msg = std::string()) const; 166 167 /// getSourceLocation - Return a source location identifier for the specified 168 /// offset in the current file. 169 SourceLocation getSourceLocation(const char *Loc) const; 170 171 /// Stringify - Convert the specified string into a C string by escaping '\' 172 /// and " characters. This does not add surrounding ""'s to the string. 173 /// If Charify is true, this escapes the ' character instead of ". 174 static std::string Stringify(const std::string &Str, bool Charify = false); 175 176 //===--------------------------------------------------------------------===// 177 // Internal implementation interfaces. 178private: 179 180 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 181 /// by Lex. 182 /// 183 void LexTokenInternal(Token &Result); 184 185 /// FormTokenWithChars - When we lex a token, we have identified a span 186 /// starting at BufferPtr, going to TokEnd that forms the token. This method 187 /// takes that range and assigns it to the token as its location and size. In 188 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 189 /// TokEnd. 190 void FormTokenWithChars(Token &Result, const char *TokEnd) { 191 Result.setLocation(getSourceLocation(BufferPtr)); 192 Result.setLength(TokEnd-BufferPtr); 193 BufferPtr = TokEnd; 194 } 195 196 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 197 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 198 /// tokens in the buffer controlled by this lexer. 199 unsigned isNextPPTokenLParen(); 200 201 //===--------------------------------------------------------------------===// 202 // Lexer character reading interfaces. 203public: 204 205 // This lexer is built on two interfaces for reading characters, both of which 206 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 207 // when we know that we will be reading a character from the input buffer and 208 // that this character will be part of the result token. This occurs in (f.e.) 209 // string processing, because we know we need to read until we find the 210 // closing '"' character. 211 // 212 // The second interface is the combination of PeekCharAndSize with 213 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 214 // returning it and its size. If the lexer decides that this character is 215 // part of the current token, it calls ConsumeChar on it. This two stage 216 // approach allows us to emit diagnostics for characters (e.g. warnings about 217 // trigraphs), knowing that they only are emitted if the character is 218 // consumed. 219 220 /// isObviouslySimpleCharacter - Return true if the specified character is 221 /// obviously the same in translation phase 1 and translation phase 3. This 222 /// can return false for characters that end up being the same, but it will 223 /// never return true for something that needs to be mapped. 224 static bool isObviouslySimpleCharacter(char C) { 225 return C != '?' && C != '\\'; 226 } 227 228 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 229 /// advance over it, and return it. This is tricky in several cases. Here we 230 /// just handle the trivial case and fall-back to the non-inlined 231 /// getCharAndSizeSlow method to handle the hard case. 232 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 233 // If this is not a trigraph and not a UCN or escaped newline, return 234 // quickly. 235 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 236 237 unsigned Size = 0; 238 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 239 Ptr += Size; 240 return C; 241 } 242 243private: 244 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 245 /// and added to a given token, check to see if there are diagnostics that 246 /// need to be emitted or flags that need to be set on the token. If so, do 247 /// it. 248 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 249 // Normal case, we consumed exactly one token. Just return it. 250 if (Size == 1) 251 return Ptr+Size; 252 253 // Otherwise, re-lex the character with a current token, allowing 254 // diagnostics to be emitted and flags to be set. 255 Size = 0; 256 getCharAndSizeSlow(Ptr, Size, &Tok); 257 return Ptr+Size; 258 } 259 260 /// getCharAndSize - Peek a single 'character' from the specified buffer, 261 /// get its size, and return it. This is tricky in several cases. Here we 262 /// just handle the trivial case and fall-back to the non-inlined 263 /// getCharAndSizeSlow method to handle the hard case. 264 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 265 // If this is not a trigraph and not a UCN or escaped newline, return 266 // quickly. 267 if (isObviouslySimpleCharacter(Ptr[0])) { 268 Size = 1; 269 return *Ptr; 270 } 271 272 Size = 0; 273 return getCharAndSizeSlow(Ptr, Size); 274 } 275 276 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 277 /// method. 278 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 279 280 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 281 /// emit a warning. 282 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 283 const LangOptions &Features) { 284 // If this is not a trigraph and not a UCN or escaped newline, return 285 // quickly. 286 if (isObviouslySimpleCharacter(Ptr[0])) { 287 Size = 1; 288 return *Ptr; 289 } 290 291 Size = 0; 292 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 293 } 294 295 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 296 /// diagnostic. 297 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 298 const LangOptions &Features); 299 300 //===--------------------------------------------------------------------===// 301 // #if directive handling. 302 303 /// pushConditionalLevel - When we enter a #if directive, this keeps track of 304 /// what we are currently in for diagnostic emission (e.g. #if with missing 305 /// #endif). 306 void pushConditionalLevel(SourceLocation DirectiveStart, bool WasSkipping, 307 bool FoundNonSkip, bool FoundElse) { 308 PPConditionalInfo CI; 309 CI.IfLoc = DirectiveStart; 310 CI.WasSkipping = WasSkipping; 311 CI.FoundNonSkip = FoundNonSkip; 312 CI.FoundElse = FoundElse; 313 ConditionalStack.push_back(CI); 314 } 315 void pushConditionalLevel(const PPConditionalInfo &CI) { 316 ConditionalStack.push_back(CI); 317 } 318 319 /// popConditionalLevel - Remove an entry off the top of the conditional 320 /// stack, returning information about it. If the conditional stack is empty, 321 /// this returns true and does not fill in the arguments. 322 bool popConditionalLevel(PPConditionalInfo &CI) { 323 if (ConditionalStack.empty()) return true; 324 CI = ConditionalStack.back(); 325 ConditionalStack.pop_back(); 326 return false; 327 } 328 329 /// peekConditionalLevel - Return the top of the conditional stack. This 330 /// requires that there be a conditional active. 331 PPConditionalInfo &peekConditionalLevel() { 332 assert(!ConditionalStack.empty() && "No conditionals active!"); 333 return ConditionalStack.back(); 334 } 335 336 unsigned getConditionalStackDepth() const { return ConditionalStack.size(); } 337 338 //===--------------------------------------------------------------------===// 339 // Other lexer functions. 340 341 // Helper functions to lex the remainder of a token of the specific type. 342 void LexIdentifier (Token &Result, const char *CurPtr); 343 void LexNumericConstant (Token &Result, const char *CurPtr); 344 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 345 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 346 void LexCharConstant (Token &Result, const char *CurPtr); 347 bool LexEndOfFile (Token &Result, const char *CurPtr); 348 349 void SkipWhitespace (Token &Result, const char *CurPtr); 350 bool SkipBCPLComment (Token &Result, const char *CurPtr); 351 bool SkipBlockComment (Token &Result, const char *CurPtr); 352 bool SaveBCPLComment (Token &Result, const char *CurPtr); 353 354 /// LexIncludeFilename - After the preprocessor has parsed a #include, lex and 355 /// (potentially) macro expand the filename. If the sequence parsed is not 356 /// lexically legal, emit a diagnostic and return a result EOM token. 357 void LexIncludeFilename(Token &Result); 358}; 359 360 361} // end namespace clang 362 363#endif 364