Lexer.h revision 168ae2d44a443da75ea85db5f3b5081eb0bce113
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 Preprocessor; 28 29/// Lexer - This provides a simple interface that turns a text buffer into a 30/// stream of tokens. This provides no support for file reading or buffering, 31/// or buffering/seeking of tokens, only forward lexing is supported. It relies 32/// on the specified Preprocessor object to handle preprocessor directives, etc. 33class Lexer { 34 //===--------------------------------------------------------------------===// 35 // Constant configuration values for this lexer. 36 const char *BufferStart; // Start of the buffer. 37 const char *BufferEnd; // End of the buffer. 38 SourceLocation FileLoc; // Location for start of file. 39 Preprocessor *PP; // Preprocessor object controlling lexing. 40 LangOptions Features; // Features enabled by this language (cache). 41 bool Is_PragmaLexer; // True if lexer for _Pragma handling. 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 /// 66 /// Note that in raw mode that the PP pointer may be null. 67 bool LexingRawMode; 68 69 /// KeepCommentMode - The lexer can optionally keep C & BCPL-style comments, 70 /// and return them as tokens. This is used for -C and -CC modes. 71 bool KeepCommentMode; 72 73 //===--------------------------------------------------------------------===// 74 // Context that changes as the file is lexed. 75 // NOTE: any state that mutates when in raw mode must have save/restore code 76 // in Lexer::isNextPPTokenLParen. 77 78 // BufferPtr - Current pointer into the buffer. This is the next character 79 // to be lexed. 80 const char *BufferPtr; 81 82 // IsAtStartOfLine - True if the next lexed token should get the "start of 83 // line" flag set on it. 84 bool IsAtStartOfLine; 85 86 /// MIOpt - This is a state machine that detects the #ifndef-wrapping a file 87 /// idiom for the multiple-include optimization. 88 MultipleIncludeOpt MIOpt; 89 90 /// ConditionalStack - Information about the set of #if/#ifdef/#ifndef blocks 91 /// we are currently in. 92 std::vector<PPConditionalInfo> ConditionalStack; 93 94 Lexer(const Lexer&); // DO NOT IMPLEMENT 95 void operator=(const Lexer&); // DO NOT IMPLEMENT 96 friend class Preprocessor; 97public: 98 99 /// Lexer constructor - Create a new lexer object for the specified buffer 100 /// with the specified preprocessor managing the lexing process. This lexer 101 /// assumes that the associated file buffer and Preprocessor objects will 102 /// outlive it, so it doesn't take ownership of either of them. 103 Lexer(SourceLocation FileLoc, Preprocessor &PP, 104 const char *BufStart = 0, const char *BufEnd = 0); 105 106 /// Lexer constructor - Create a new raw lexer object. This object is only 107 /// suitable for calls to 'LexRawToken'. This lexer assumes that the 108 /// associated file buffer will outlive it, so it doesn't take ownership of 109 /// either of them. 110 Lexer(SourceLocation FileLoc, const LangOptions &Features, 111 const char *BufStart, const char *BufEnd); 112 113 /// getFeatures - Return the language features currently enabled. NOTE: this 114 /// lexer modifies features as a file is parsed! 115 const LangOptions &getFeatures() const { return Features; } 116 117 /// getFileLoc - Return the File Location for the file we are lexing out of. 118 /// The physical location encodes the location where the characters come from, 119 /// the virtual location encodes where we should *claim* the characters came 120 /// from. Currently this is only used by _Pragma handling. 121 SourceLocation getFileLoc() const { return FileLoc; } 122 123 /// Lex - Return the next token in the file. If this is the end of file, it 124 /// return the tok::eof token. Return true if an error occurred and 125 /// compilation should terminate, false if normal. This implicitly involves 126 /// the preprocessor. 127 void Lex(Token &Result) { 128 // Start a new token. 129 Result.startToken(); 130 131 // NOTE, any changes here should also change code after calls to 132 // Preprocessor::HandleDirective 133 if (IsAtStartOfLine) { 134 Result.setFlag(Token::StartOfLine); 135 IsAtStartOfLine = false; 136 } 137 138 // Get a token. Note that this may delete the current lexer if the end of 139 // file is reached. 140 LexTokenInternal(Result); 141 } 142 143 /// LexRawToken - Switch the lexer to raw mode, lex a token into Result and 144 /// switch it back. Return true if the 'next character to read' pointer 145 /// points and the end of the lexer buffer, false otherwise. 146 bool LexRawToken(Token &Result) { 147 assert(!LexingRawMode && "Already in raw mode!"); 148 LexingRawMode = true; 149 Lex(Result); 150 LexingRawMode = PP == 0; 151 // Note that lexing to the end of the buffer doesn't implicitly delete the 152 // lexer when in raw mode. 153 return BufferPtr == BufferEnd; 154 } 155 156 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 157 /// uninterpreted string. This switches the lexer out of directive mode. 158 std::string ReadToEndOfLine(); 159 160 161 /// Diag - Forwarding function for diagnostics. This translate a source 162 /// position in the current buffer into a SourceLocation object for rendering. 163 void Diag(const char *Loc, unsigned DiagID, 164 const std::string &Msg = std::string()) const; 165 void Diag(SourceLocation Loc, unsigned DiagID, 166 const std::string &Msg = std::string()) const; 167 168 /// getSourceLocation - Return a source location identifier for the specified 169 /// offset in the current file. 170 SourceLocation getSourceLocation(const char *Loc) const; 171 172 /// Stringify - Convert the specified string into a C string by escaping '\' 173 /// and " characters. This does not add surrounding ""'s to the string. 174 /// If Charify is true, this escapes the ' character instead of ". 175 static std::string Stringify(const std::string &Str, bool Charify = false); 176 177 /// Stringify - Convert the specified string into a C string by escaping '\' 178 /// and " characters. This does not add surrounding ""'s to the string. 179 static void Stringify(llvm::SmallVectorImpl<char> &Str); 180 181 //===--------------------------------------------------------------------===// 182 // Internal implementation interfaces. 183private: 184 185 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 186 /// by Lex. 187 /// 188 void LexTokenInternal(Token &Result); 189 190 /// FormTokenWithChars - When we lex a token, we have identified a span 191 /// starting at BufferPtr, going to TokEnd that forms the token. This method 192 /// takes that range and assigns it to the token as its location and size. In 193 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 194 /// TokEnd. 195 void FormTokenWithChars(Token &Result, const char *TokEnd) { 196 Result.setLocation(getSourceLocation(BufferPtr)); 197 Result.setLength(TokEnd-BufferPtr); 198 BufferPtr = TokEnd; 199 } 200 201 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 202 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 203 /// tokens in the buffer controlled by this lexer. 204 unsigned isNextPPTokenLParen(); 205 206 //===--------------------------------------------------------------------===// 207 // Lexer character reading interfaces. 208public: 209 210 // This lexer is built on two interfaces for reading characters, both of which 211 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 212 // when we know that we will be reading a character from the input buffer and 213 // that this character will be part of the result token. This occurs in (f.e.) 214 // string processing, because we know we need to read until we find the 215 // closing '"' character. 216 // 217 // The second interface is the combination of PeekCharAndSize with 218 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 219 // returning it and its size. If the lexer decides that this character is 220 // part of the current token, it calls ConsumeChar on it. This two stage 221 // approach allows us to emit diagnostics for characters (e.g. warnings about 222 // trigraphs), knowing that they only are emitted if the character is 223 // consumed. 224 225 /// isObviouslySimpleCharacter - Return true if the specified character is 226 /// obviously the same in translation phase 1 and translation phase 3. This 227 /// can return false for characters that end up being the same, but it will 228 /// never return true for something that needs to be mapped. 229 static bool isObviouslySimpleCharacter(char C) { 230 return C != '?' && C != '\\'; 231 } 232 233 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 234 /// advance over it, and return it. This is tricky in several cases. Here we 235 /// just handle the trivial case and fall-back to the non-inlined 236 /// getCharAndSizeSlow method to handle the hard case. 237 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 238 // If this is not a trigraph and not a UCN or escaped newline, return 239 // quickly. 240 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 241 242 unsigned Size = 0; 243 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 244 Ptr += Size; 245 return C; 246 } 247 248private: 249 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 250 /// and added to a given token, check to see if there are diagnostics that 251 /// need to be emitted or flags that need to be set on the token. If so, do 252 /// it. 253 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 254 // Normal case, we consumed exactly one token. Just return it. 255 if (Size == 1) 256 return Ptr+Size; 257 258 // Otherwise, re-lex the character with a current token, allowing 259 // diagnostics to be emitted and flags to be set. 260 Size = 0; 261 getCharAndSizeSlow(Ptr, Size, &Tok); 262 return Ptr+Size; 263 } 264 265 /// getCharAndSize - Peek a single 'character' from the specified buffer, 266 /// get its size, and return it. This is tricky in several cases. Here we 267 /// just handle the trivial case and fall-back to the non-inlined 268 /// getCharAndSizeSlow method to handle the hard case. 269 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 270 // If this is not a trigraph and not a UCN or escaped newline, return 271 // quickly. 272 if (isObviouslySimpleCharacter(Ptr[0])) { 273 Size = 1; 274 return *Ptr; 275 } 276 277 Size = 0; 278 return getCharAndSizeSlow(Ptr, Size); 279 } 280 281 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 282 /// method. 283 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 284 285 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 286 /// emit a warning. 287 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 288 const LangOptions &Features) { 289 // If this is not a trigraph and not a UCN or escaped newline, return 290 // quickly. 291 if (isObviouslySimpleCharacter(Ptr[0])) { 292 Size = 1; 293 return *Ptr; 294 } 295 296 Size = 0; 297 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 298 } 299 300 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 301 /// diagnostic. 302 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 303 const LangOptions &Features); 304 305 //===--------------------------------------------------------------------===// 306 // #if directive handling. 307 308 /// pushConditionalLevel - When we enter a #if directive, this keeps track of 309 /// what we are currently in for diagnostic emission (e.g. #if with missing 310 /// #endif). 311 void pushConditionalLevel(SourceLocation DirectiveStart, bool WasSkipping, 312 bool FoundNonSkip, bool FoundElse) { 313 PPConditionalInfo CI; 314 CI.IfLoc = DirectiveStart; 315 CI.WasSkipping = WasSkipping; 316 CI.FoundNonSkip = FoundNonSkip; 317 CI.FoundElse = FoundElse; 318 ConditionalStack.push_back(CI); 319 } 320 void pushConditionalLevel(const PPConditionalInfo &CI) { 321 ConditionalStack.push_back(CI); 322 } 323 324 /// popConditionalLevel - Remove an entry off the top of the conditional 325 /// stack, returning information about it. If the conditional stack is empty, 326 /// this returns true and does not fill in the arguments. 327 bool popConditionalLevel(PPConditionalInfo &CI) { 328 if (ConditionalStack.empty()) return true; 329 CI = ConditionalStack.back(); 330 ConditionalStack.pop_back(); 331 return false; 332 } 333 334 /// peekConditionalLevel - Return the top of the conditional stack. This 335 /// requires that there be a conditional active. 336 PPConditionalInfo &peekConditionalLevel() { 337 assert(!ConditionalStack.empty() && "No conditionals active!"); 338 return ConditionalStack.back(); 339 } 340 341 unsigned getConditionalStackDepth() const { return ConditionalStack.size(); } 342 343 //===--------------------------------------------------------------------===// 344 // Other lexer functions. 345 346 // Helper functions to lex the remainder of a token of the specific type. 347 void LexIdentifier (Token &Result, const char *CurPtr); 348 void LexNumericConstant (Token &Result, const char *CurPtr); 349 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 350 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 351 void LexCharConstant (Token &Result, const char *CurPtr); 352 bool LexEndOfFile (Token &Result, const char *CurPtr); 353 354 void SkipWhitespace (Token &Result, const char *CurPtr); 355 bool SkipBCPLComment (Token &Result, const char *CurPtr); 356 bool SkipBlockComment (Token &Result, const char *CurPtr); 357 bool SaveBCPLComment (Token &Result, const char *CurPtr); 358 359 /// LexIncludeFilename - After the preprocessor has parsed a #include, lex and 360 /// (potentially) macro expand the filename. If the sequence parsed is not 361 /// lexically legal, emit a diagnostic and return a result EOM token. 362 void LexIncludeFilename(Token &Result); 363}; 364 365 366} // end namespace clang 367 368#endif 369