Lexer.h revision 7ef5c27eb6e8ebe58b52013246c06753c3613263
1//===--- Lexer.h - C Language Family Lexer ----------------------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file is distributed under the University of Illinois Open Source 6// 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/PreprocessorLexer.h" 18#include "clang/Basic/LangOptions.h" 19#include "llvm/ADT/SmallVector.h" 20#include <string> 21#include <vector> 22#include <cassert> 23 24namespace clang { 25class Diagnostic; 26class SourceManager; 27class Preprocessor; 28class DiagnosticBuilder; 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 : public PreprocessorLexer { 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 LangOptions Features; // Features enabled by this language (cache). 41 bool Is_PragmaLexer : 1; // True if lexer for _Pragma handling. 42 bool IsInConflictMarker : 1; // True if in a VCS conflict marker '<<<<<<<' 43 44 //===--------------------------------------------------------------------===// 45 // Context-specific lexing flags set by the preprocessor. 46 // 47 48 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace 49 /// and return them as tokens. This is used for -C and -CC modes, and 50 /// whitespace preservation can be useful for some clients that want to lex 51 /// the file in raw mode and get every character from the file. 52 /// 53 /// When this is set to 2 it returns comments and whitespace. When set to 1 54 /// it returns comments, when it is set to 0 it returns normal tokens only. 55 unsigned char ExtendedTokenMode; 56 57 //===--------------------------------------------------------------------===// 58 // Context that changes as the file is lexed. 59 // NOTE: any state that mutates when in raw mode must have save/restore code 60 // in Lexer::isNextPPTokenLParen. 61 62 // BufferPtr - Current pointer into the buffer. This is the next character 63 // to be lexed. 64 const char *BufferPtr; 65 66 // IsAtStartOfLine - True if the next lexed token should get the "start of 67 // line" flag set on it. 68 bool IsAtStartOfLine; 69 70 Lexer(const Lexer&); // DO NOT IMPLEMENT 71 void operator=(const Lexer&); // DO NOT IMPLEMENT 72 friend class Preprocessor; 73 74 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd); 75public: 76 77 /// Lexer constructor - Create a new lexer object for the specified buffer 78 /// with the specified preprocessor managing the lexing process. This lexer 79 /// assumes that the associated file buffer and Preprocessor objects will 80 /// outlive it, so it doesn't take ownership of either of them. 81 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, Preprocessor &PP); 82 83 /// Lexer constructor - Create a new raw lexer object. This object is only 84 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 85 /// range will outlive it, so it doesn't take ownership of it. 86 Lexer(SourceLocation FileLoc, const LangOptions &Features, 87 const char *BufStart, const char *BufPtr, const char *BufEnd); 88 89 /// Lexer constructor - Create a new raw lexer object. This object is only 90 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 91 /// range will outlive it, so it doesn't take ownership of it. 92 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, 93 const SourceManager &SM, const LangOptions &Features); 94 95 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for 96 /// _Pragma expansion. This has a variety of magic semantics that this method 97 /// sets up. It returns a new'd Lexer that must be delete'd when done. 98 static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc, 99 SourceLocation InstantiationLocStart, 100 SourceLocation InstantiationLocEnd, 101 unsigned TokLen, Preprocessor &PP); 102 103 104 /// 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 /// Lex - Return the next token in the file. If this is the end of file, it 115 /// return the tok::eof token. Return true if an error occurred and 116 /// compilation should terminate, false if normal. This implicitly involves 117 /// the preprocessor. 118 void Lex(Token &Result) { 119 // Start a new token. 120 Result.startToken(); 121 122 // NOTE, any changes here should also change code after calls to 123 // Preprocessor::HandleDirective 124 if (IsAtStartOfLine) { 125 Result.setFlag(Token::StartOfLine); 126 IsAtStartOfLine = false; 127 } 128 129 // Get a token. Note that this may delete the current lexer if the end of 130 // file is reached. 131 LexTokenInternal(Result); 132 } 133 134 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 135 bool isPragmaLexer() const { return Is_PragmaLexer; } 136 137 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 138 /// the PreprocessorLexer interface. 139 void IndirectLex(Token &Result) { Lex(Result); } 140 141 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 142 /// associated preprocessor object. Return true if the 'next character to 143 /// read' pointer points at the end of the lexer buffer, false otherwise. 144 bool LexFromRawLexer(Token &Result) { 145 assert(LexingRawMode && "Not already in raw mode!"); 146 Lex(Result); 147 // Note that lexing to the end of the buffer doesn't implicitly delete the 148 // lexer when in raw mode. 149 return BufferPtr == BufferEnd; 150 } 151 152 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 153 /// every character in the file, including whitespace and comments. This 154 /// should only be used in raw mode, as the preprocessor is not prepared to 155 /// deal with the excess tokens. 156 bool isKeepWhitespaceMode() const { 157 return ExtendedTokenMode > 1; 158 } 159 160 /// SetKeepWhitespaceMode - This method lets clients enable or disable 161 /// whitespace retention mode. 162 void SetKeepWhitespaceMode(bool Val) { 163 assert((!Val || LexingRawMode) && 164 "Can only enable whitespace retention in raw mode"); 165 ExtendedTokenMode = Val ? 2 : 0; 166 } 167 168 /// inKeepCommentMode - Return true if the lexer should return comments as 169 /// tokens. 170 bool inKeepCommentMode() const { 171 return ExtendedTokenMode > 0; 172 } 173 174 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 175 /// to the specified mode. This is really only useful when lexing in raw 176 /// mode, because otherwise the lexer needs to manage this. 177 void SetCommentRetentionState(bool Mode) { 178 assert(!isKeepWhitespaceMode() && 179 "Can't play with comment retention state when retaining whitespace"); 180 ExtendedTokenMode = Mode ? 1 : 0; 181 } 182 183 const char *getBufferStart() const { return BufferStart; } 184 185 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 186 /// uninterpreted string. This switches the lexer out of directive mode. 187 std::string ReadToEndOfLine(); 188 189 190 /// Diag - Forwarding function for diagnostics. This translate a source 191 /// position in the current buffer into a SourceLocation object for rendering. 192 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const; 193 194 /// getSourceLocation - Return a source location identifier for the specified 195 /// offset in the current file. 196 SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const; 197 198 /// getSourceLocation - Return a source location for the next character in 199 /// the current file. 200 SourceLocation getSourceLocation() { return getSourceLocation(BufferPtr); } 201 202 /// \brief Return the current location in the buffer. 203 const char *getBufferLocation() const { return BufferPtr; } 204 205 /// Stringify - Convert the specified string into a C string by escaping '\' 206 /// and " characters. This does not add surrounding ""'s to the string. 207 /// If Charify is true, this escapes the ' character instead of ". 208 static std::string Stringify(const std::string &Str, bool Charify = false); 209 210 /// Stringify - Convert the specified string into a C string by escaping '\' 211 /// and " characters. This does not add surrounding ""'s to the string. 212 static void Stringify(llvm::SmallVectorImpl<char> &Str); 213 214 /// MeasureTokenLength - Relex the token at the specified location and return 215 /// its length in bytes in the input file. If the token needs cleaning (e.g. 216 /// includes a trigraph or an escaped newline) then this count includes bytes 217 /// that are part of that. 218 static unsigned MeasureTokenLength(SourceLocation Loc, 219 const SourceManager &SM, 220 const LangOptions &LangOpts); 221 222 /// \brief Given a location any where in a source buffer, find the location 223 /// that corresponds to the beginning of the token in which the original 224 /// source location lands. 225 /// 226 /// \param Loc 227 static SourceLocation GetBeginningOfToken(SourceLocation Loc, 228 const SourceManager &SM, 229 const LangOptions &LangOpts); 230 231 /// AdvanceToTokenCharacter - If the current SourceLocation specifies a 232 /// location at the start of a token, return a new location that specifies a 233 /// character within the token. This handles trigraphs and escaped newlines. 234 static SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 235 unsigned Character, 236 const SourceManager &SM, 237 const LangOptions &Features); 238 239 /// \brief Computes the source location just past the end of the 240 /// token at this source location. 241 /// 242 /// This routine can be used to produce a source location that 243 /// points just past the end of the token referenced by \p Loc, and 244 /// is generally used when a diagnostic needs to point just after a 245 /// token where it expected something different that it received. If 246 /// the returned source location would not be meaningful (e.g., if 247 /// it points into a macro), this routine returns an invalid 248 /// source location. 249 /// 250 /// \param Offset an offset from the end of the token, where the source 251 /// location should refer to. The default offset (0) produces a source 252 /// location pointing just past the end of the token; an offset of 1 produces 253 /// a source location pointing to the last character in the token, etc. 254 static SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset, 255 const SourceManager &SM, 256 const LangOptions &Features); 257 258 /// \brief Compute the preamble of the given file. 259 /// 260 /// The preamble of a file contains the initial comments, include directives, 261 /// and other preprocessor directives that occur before the code in this 262 /// particular file actually begins. The preamble of the main source file is 263 /// a potential prefix header. 264 /// 265 /// \param Buffer The memory buffer containing the file's contents. 266 /// 267 /// \param MaxLines If non-zero, restrict the length of the preamble 268 /// to fewer than this number of lines. 269 /// 270 /// \returns The offset into the file where the preamble ends and the rest 271 /// of the file begins along with a boolean value indicating whether 272 /// the preamble ends at the beginning of a new line. 273 static std::pair<unsigned, bool> 274 ComputePreamble(const llvm::MemoryBuffer *Buffer, unsigned MaxLines = 0); 275 276 //===--------------------------------------------------------------------===// 277 // Internal implementation interfaces. 278private: 279 280 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 281 /// by Lex. 282 /// 283 void LexTokenInternal(Token &Result); 284 285 /// FormTokenWithChars - When we lex a token, we have identified a span 286 /// starting at BufferPtr, going to TokEnd that forms the token. This method 287 /// takes that range and assigns it to the token as its location and size. In 288 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 289 /// TokEnd. 290 void FormTokenWithChars(Token &Result, const char *TokEnd, 291 tok::TokenKind Kind) { 292 unsigned TokLen = TokEnd-BufferPtr; 293 Result.setLength(TokLen); 294 Result.setLocation(getSourceLocation(BufferPtr, TokLen)); 295 Result.setKind(Kind); 296 BufferPtr = TokEnd; 297 } 298 299 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 300 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 301 /// tokens in the buffer controlled by this lexer. 302 unsigned isNextPPTokenLParen(); 303 304 //===--------------------------------------------------------------------===// 305 // Lexer character reading interfaces. 306public: 307 308 // This lexer is built on two interfaces for reading characters, both of which 309 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 310 // when we know that we will be reading a character from the input buffer and 311 // that this character will be part of the result token. This occurs in (f.e.) 312 // string processing, because we know we need to read until we find the 313 // closing '"' character. 314 // 315 // The second interface is the combination of getCharAndSize with 316 // ConsumeChar. getCharAndSize reads a phase 1/2 translated character, 317 // returning it and its size. If the lexer decides that this character is 318 // part of the current token, it calls ConsumeChar on it. This two stage 319 // approach allows us to emit diagnostics for characters (e.g. warnings about 320 // trigraphs), knowing that they only are emitted if the character is 321 // consumed. 322 323 /// isObviouslySimpleCharacter - Return true if the specified character is 324 /// obviously the same in translation phase 1 and translation phase 3. This 325 /// can return false for characters that end up being the same, but it will 326 /// never return true for something that needs to be mapped. 327 static bool isObviouslySimpleCharacter(char C) { 328 return C != '?' && C != '\\'; 329 } 330 331 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 332 /// advance over it, and return it. This is tricky in several cases. Here we 333 /// just handle the trivial case and fall-back to the non-inlined 334 /// getCharAndSizeSlow method to handle the hard case. 335 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 336 // If this is not a trigraph and not a UCN or escaped newline, return 337 // quickly. 338 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 339 340 unsigned Size = 0; 341 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 342 Ptr += Size; 343 return C; 344 } 345 346private: 347 /// ConsumeChar - When a character (identified by getCharAndSize) is consumed 348 /// and added to a given token, check to see if there are diagnostics that 349 /// need to be emitted or flags that need to be set on the token. If so, do 350 /// it. 351 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 352 // Normal case, we consumed exactly one token. Just return it. 353 if (Size == 1) 354 return Ptr+Size; 355 356 // Otherwise, re-lex the character with a current token, allowing 357 // diagnostics to be emitted and flags to be set. 358 Size = 0; 359 getCharAndSizeSlow(Ptr, Size, &Tok); 360 return Ptr+Size; 361 } 362 363 /// getCharAndSize - Peek a single 'character' from the specified buffer, 364 /// get its size, and return it. This is tricky in several cases. Here we 365 /// just handle the trivial case and fall-back to the non-inlined 366 /// getCharAndSizeSlow method to handle the hard case. 367 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 368 // If this is not a trigraph and not a UCN or escaped newline, return 369 // quickly. 370 if (isObviouslySimpleCharacter(Ptr[0])) { 371 Size = 1; 372 return *Ptr; 373 } 374 375 Size = 0; 376 return getCharAndSizeSlow(Ptr, Size); 377 } 378 379 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 380 /// method. 381 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 382public: 383 384 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 385 /// emit a warning. 386 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 387 const LangOptions &Features) { 388 // If this is not a trigraph and not a UCN or escaped newline, return 389 // quickly. 390 if (isObviouslySimpleCharacter(Ptr[0])) { 391 Size = 1; 392 return *Ptr; 393 } 394 395 Size = 0; 396 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 397 } 398 399 /// getEscapedNewLineSize - Return the size of the specified escaped newline, 400 /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry 401 /// to this function. 402 static unsigned getEscapedNewLineSize(const char *P); 403 404 /// SkipEscapedNewLines - If P points to an escaped newline (or a series of 405 /// them), skip over them and return the first non-escaped-newline found, 406 /// otherwise return P. 407 static const char *SkipEscapedNewLines(const char *P); 408private: 409 410 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 411 /// diagnostic. 412 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 413 const LangOptions &Features); 414 415 //===--------------------------------------------------------------------===// 416 // Other lexer functions. 417 418 void SkipBytes(unsigned Bytes, bool StartOfLine); 419 420 // Helper functions to lex the remainder of a token of the specific type. 421 void LexIdentifier (Token &Result, const char *CurPtr); 422 void LexNumericConstant (Token &Result, const char *CurPtr); 423 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 424 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 425 void LexCharConstant (Token &Result, const char *CurPtr); 426 bool LexEndOfFile (Token &Result, const char *CurPtr); 427 428 bool SkipWhitespace (Token &Result, const char *CurPtr); 429 bool SkipBCPLComment (Token &Result, const char *CurPtr); 430 bool SkipBlockComment (Token &Result, const char *CurPtr); 431 bool SaveBCPLComment (Token &Result, const char *CurPtr); 432 433 bool IsStartOfConflictMarker(const char *CurPtr); 434 bool HandleEndOfConflictMarker(const char *CurPtr); 435}; 436 437 438} // end namespace clang 439 440#endif 441