Lexer.h revision 9a6119437672f42be5f50c3fe89fe843b1bfa5b5
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 "llvm/ADT/SmallVector.h" 21#include <string> 22#include <vector> 23#include <cassert> 24 25namespace clang { 26class Diagnostic; 27class SourceManager; 28class Preprocessor; 29 30/// Lexer - This provides a simple interface that turns a text buffer into a 31/// stream of tokens. This provides no support for file reading or buffering, 32/// or buffering/seeking of tokens, only forward lexing is supported. It relies 33/// on the specified Preprocessor object to handle preprocessor directives, etc. 34class Lexer { 35 //===--------------------------------------------------------------------===// 36 // Constant configuration values for this lexer. 37 const char *BufferStart; // Start of the buffer. 38 const char *BufferEnd; // End of the buffer. 39 SourceLocation FileLoc; // Location for start of file. 40 Preprocessor *PP; // Preprocessor object controlling lexing. 41 LangOptions Features; // Features enabled by this language (cache). 42 bool Is_PragmaLexer; // True if lexer for _Pragma handling. 43 44 //===--------------------------------------------------------------------===// 45 // Context-specific lexing flags set by the preprocessor. 46 // 47 48 /// ParsingPreprocessorDirective - This is true when parsing #XXX. This turns 49 /// '\n' into a tok::eom token. 50 bool ParsingPreprocessorDirective; 51 52 /// ParsingFilename - True after #include: this turns <xx> into a 53 /// tok::angle_string_literal token. 54 bool ParsingFilename; 55 56 /// LexingRawMode - True if in raw mode: This flag disables interpretation of 57 /// tokens and is a far faster mode to lex in than non-raw-mode. This flag: 58 /// 1. If EOF of the current lexer is found, the include stack isn't popped. 59 /// 2. Identifier information is not looked up for identifier tokens. As an 60 /// effect of this, implicit macro expansion is naturally disabled. 61 /// 3. "#" tokens at the start of a line are treated as normal tokens, not 62 /// implicitly transformed by the lexer. 63 /// 4. All diagnostic messages are disabled, except for unterminated /*. 64 /// 5. The only callback made into the preprocessor is to report a hard error 65 /// on an unterminated '/*' comment. 66 /// 67 /// Note that in raw mode that the PP pointer may be null. 68 bool LexingRawMode; 69 70 /// KeepCommentMode - The lexer can optionally keep C & BCPL-style comments, 71 /// and return them as tokens. This is used for -C and -CC modes. 72 bool KeepCommentMode; 73 74 //===--------------------------------------------------------------------===// 75 // Context that changes as the file is lexed. 76 // NOTE: any state that mutates when in raw mode must have save/restore code 77 // in Lexer::isNextPPTokenLParen. 78 79 // BufferPtr - Current pointer into the buffer. This is the next character 80 // to be lexed. 81 const char *BufferPtr; 82 83 // IsAtStartOfLine - True if the next lexed token should get the "start of 84 // line" flag set on it. 85 bool IsAtStartOfLine; 86 87 /// MIOpt - This is a state machine that detects the #ifndef-wrapping a file 88 /// idiom for the multiple-include optimization. 89 MultipleIncludeOpt MIOpt; 90 91 /// ConditionalStack - Information about the set of #if/#ifdef/#ifndef blocks 92 /// we are currently in. 93 std::vector<PPConditionalInfo> ConditionalStack; 94 95 Lexer(const Lexer&); // DO NOT IMPLEMENT 96 void operator=(const Lexer&); // DO NOT IMPLEMENT 97 friend class Preprocessor; 98public: 99 100 /// Lexer constructor - Create a new lexer object for the specified buffer 101 /// with the specified preprocessor managing the lexing process. This lexer 102 /// assumes that the associated file buffer and Preprocessor objects will 103 /// outlive it, so it doesn't take ownership of either of them. 104 Lexer(SourceLocation FileLoc, Preprocessor &PP, 105 const char *BufStart = 0, const char *BufEnd = 0); 106 107 /// Lexer constructor - Create a new raw lexer object. This object is only 108 /// suitable for calls to 'LexRawToken'. This lexer assumes that the 109 /// associated file buffer will outlive it, so it doesn't take ownership of 110 /// either of them. 111 Lexer(SourceLocation FileLoc, const LangOptions &Features, 112 const char *BufStart, const char *BufEnd); 113 114 /// getFeatures - Return the language features currently enabled. NOTE: this 115 /// lexer modifies features as a file is parsed! 116 const LangOptions &getFeatures() const { return Features; } 117 118 /// getFileLoc - Return the File Location for the file we are lexing out of. 119 /// The physical location encodes the location where the characters come from, 120 /// the virtual location encodes where we should *claim* the characters came 121 /// from. Currently this is only used by _Pragma handling. 122 SourceLocation getFileLoc() const { return FileLoc; } 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 = PP == 0; 152 // Note that lexing to the end of the buffer doesn't implicitly delete the 153 // lexer when in raw mode. 154 return BufferPtr == BufferEnd; 155 } 156 157 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 158 /// uninterpreted string. This switches the lexer out of directive mode. 159 std::string ReadToEndOfLine(); 160 161 162 /// Diag - Forwarding function for diagnostics. This translate a source 163 /// position in the current buffer into a SourceLocation object for rendering. 164 void Diag(const char *Loc, unsigned DiagID, 165 const std::string &Msg = std::string()) const; 166 void Diag(SourceLocation Loc, unsigned DiagID, 167 const std::string &Msg = std::string()) const; 168 169 /// getSourceLocation - Return a source location identifier for the specified 170 /// offset in the current file. 171 SourceLocation getSourceLocation(const char *Loc) const; 172 173 /// Stringify - Convert the specified string into a C string by escaping '\' 174 /// and " characters. This does not add surrounding ""'s to the string. 175 /// If Charify is true, this escapes the ' character instead of ". 176 static std::string Stringify(const std::string &Str, bool Charify = false); 177 178 /// Stringify - Convert the specified string into a C string by escaping '\' 179 /// and " characters. This does not add surrounding ""'s to the string. 180 static void Stringify(llvm::SmallVectorImpl<char> &Str); 181 182 /// MeasureTokenLength - Relex the token at the specified location and return 183 /// its length in bytes in the input file. If the token needs cleaning (e.g. 184 /// includes a trigraph or an escaped newline) then this count includes bytes 185 /// that are part of that. 186 static unsigned MeasureTokenLength(SourceLocation Loc, 187 const SourceManager &SM); 188 189 //===--------------------------------------------------------------------===// 190 // Internal implementation interfaces. 191private: 192 193 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 194 /// by Lex. 195 /// 196 void LexTokenInternal(Token &Result); 197 198 /// FormTokenWithChars - When we lex a token, we have identified a span 199 /// starting at BufferPtr, going to TokEnd that forms the token. This method 200 /// takes that range and assigns it to the token as its location and size. In 201 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 202 /// TokEnd. 203 void FormTokenWithChars(Token &Result, const char *TokEnd) { 204 Result.setLocation(getSourceLocation(BufferPtr)); 205 Result.setLength(TokEnd-BufferPtr); 206 BufferPtr = TokEnd; 207 } 208 209 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 210 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 211 /// tokens in the buffer controlled by this lexer. 212 unsigned isNextPPTokenLParen(); 213 214 //===--------------------------------------------------------------------===// 215 // Lexer character reading interfaces. 216public: 217 218 // This lexer is built on two interfaces for reading characters, both of which 219 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 220 // when we know that we will be reading a character from the input buffer and 221 // that this character will be part of the result token. This occurs in (f.e.) 222 // string processing, because we know we need to read until we find the 223 // closing '"' character. 224 // 225 // The second interface is the combination of PeekCharAndSize with 226 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 227 // returning it and its size. If the lexer decides that this character is 228 // part of the current token, it calls ConsumeChar on it. This two stage 229 // approach allows us to emit diagnostics for characters (e.g. warnings about 230 // trigraphs), knowing that they only are emitted if the character is 231 // consumed. 232 233 /// isObviouslySimpleCharacter - Return true if the specified character is 234 /// obviously the same in translation phase 1 and translation phase 3. This 235 /// can return false for characters that end up being the same, but it will 236 /// never return true for something that needs to be mapped. 237 static bool isObviouslySimpleCharacter(char C) { 238 return C != '?' && C != '\\'; 239 } 240 241 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 242 /// advance over it, and return it. This is tricky in several cases. Here we 243 /// just handle the trivial case and fall-back to the non-inlined 244 /// getCharAndSizeSlow method to handle the hard case. 245 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 246 // If this is not a trigraph and not a UCN or escaped newline, return 247 // quickly. 248 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 249 250 unsigned Size = 0; 251 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 252 Ptr += Size; 253 return C; 254 } 255 256private: 257 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 258 /// and added to a given token, check to see if there are diagnostics that 259 /// need to be emitted or flags that need to be set on the token. If so, do 260 /// it. 261 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 262 // Normal case, we consumed exactly one token. Just return it. 263 if (Size == 1) 264 return Ptr+Size; 265 266 // Otherwise, re-lex the character with a current token, allowing 267 // diagnostics to be emitted and flags to be set. 268 Size = 0; 269 getCharAndSizeSlow(Ptr, Size, &Tok); 270 return Ptr+Size; 271 } 272 273 /// getCharAndSize - Peek a single 'character' from the specified buffer, 274 /// get its size, and return it. This is tricky in several cases. Here we 275 /// just handle the trivial case and fall-back to the non-inlined 276 /// getCharAndSizeSlow method to handle the hard case. 277 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 278 // If this is not a trigraph and not a UCN or escaped newline, return 279 // quickly. 280 if (isObviouslySimpleCharacter(Ptr[0])) { 281 Size = 1; 282 return *Ptr; 283 } 284 285 Size = 0; 286 return getCharAndSizeSlow(Ptr, Size); 287 } 288 289 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 290 /// method. 291 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 292 293 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 294 /// emit a warning. 295 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 296 const LangOptions &Features) { 297 // If this is not a trigraph and not a UCN or escaped newline, return 298 // quickly. 299 if (isObviouslySimpleCharacter(Ptr[0])) { 300 Size = 1; 301 return *Ptr; 302 } 303 304 Size = 0; 305 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 306 } 307 308 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 309 /// diagnostic. 310 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 311 const LangOptions &Features); 312 313 //===--------------------------------------------------------------------===// 314 // #if directive handling. 315 316 /// pushConditionalLevel - When we enter a #if directive, this keeps track of 317 /// what we are currently in for diagnostic emission (e.g. #if with missing 318 /// #endif). 319 void pushConditionalLevel(SourceLocation DirectiveStart, bool WasSkipping, 320 bool FoundNonSkip, bool FoundElse) { 321 PPConditionalInfo CI; 322 CI.IfLoc = DirectiveStart; 323 CI.WasSkipping = WasSkipping; 324 CI.FoundNonSkip = FoundNonSkip; 325 CI.FoundElse = FoundElse; 326 ConditionalStack.push_back(CI); 327 } 328 void pushConditionalLevel(const PPConditionalInfo &CI) { 329 ConditionalStack.push_back(CI); 330 } 331 332 /// popConditionalLevel - Remove an entry off the top of the conditional 333 /// stack, returning information about it. If the conditional stack is empty, 334 /// this returns true and does not fill in the arguments. 335 bool popConditionalLevel(PPConditionalInfo &CI) { 336 if (ConditionalStack.empty()) return true; 337 CI = ConditionalStack.back(); 338 ConditionalStack.pop_back(); 339 return false; 340 } 341 342 /// peekConditionalLevel - Return the top of the conditional stack. This 343 /// requires that there be a conditional active. 344 PPConditionalInfo &peekConditionalLevel() { 345 assert(!ConditionalStack.empty() && "No conditionals active!"); 346 return ConditionalStack.back(); 347 } 348 349 unsigned getConditionalStackDepth() const { return ConditionalStack.size(); } 350 351 //===--------------------------------------------------------------------===// 352 // Other lexer functions. 353 354 // Helper functions to lex the remainder of a token of the specific type. 355 void LexIdentifier (Token &Result, const char *CurPtr); 356 void LexNumericConstant (Token &Result, const char *CurPtr); 357 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 358 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 359 void LexCharConstant (Token &Result, const char *CurPtr); 360 bool LexEndOfFile (Token &Result, const char *CurPtr); 361 362 void SkipWhitespace (Token &Result, const char *CurPtr); 363 bool SkipBCPLComment (Token &Result, const char *CurPtr); 364 bool SkipBlockComment (Token &Result, const char *CurPtr); 365 bool SaveBCPLComment (Token &Result, const char *CurPtr); 366 367 /// LexIncludeFilename - After the preprocessor has parsed a #include, lex and 368 /// (potentially) macro expand the filename. If the sequence parsed is not 369 /// lexically legal, emit a diagnostic and return a result EOM token. 370 void LexIncludeFilename(Token &Result); 371}; 372 373 374} // end namespace clang 375 376#endif 377