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 <cassert> 22 23 namespace clang { 24 class DiagnosticsEngine; 25 class SourceManager; 26 class Preprocessor; 27 class DiagnosticBuilder; 28 29 /// ConflictMarkerKind - Kinds of conflict marker which the lexer might be 30 /// recovering from. 31 enum ConflictMarkerKind { 32 /// Not within a conflict marker. 33 CMK_None, 34 /// A normal or diff3 conflict marker, initiated by at least 7 <s, 35 /// separated by at least 7 =s or |s, and terminated by at least 7 >s. 36 CMK_Normal, 37 /// A Perforce-style conflict marker, initiated by 4 >s, separated by 4 =s, 38 /// and terminated by 4 <s. 39 CMK_Perforce 40 }; 41 42 /// Lexer - This provides a simple interface that turns a text buffer into a 43 /// stream of tokens. This provides no support for file reading or buffering, 44 /// or buffering/seeking of tokens, only forward lexing is supported. It relies 45 /// on the specified Preprocessor object to handle preprocessor directives, etc. 46 class Lexer : public PreprocessorLexer { 47 //===--------------------------------------------------------------------===// 48 // Constant configuration values for this lexer. 49 const char *BufferStart; // Start of the buffer. 50 const char *BufferEnd; // End of the buffer. 51 SourceLocation FileLoc; // Location for start of file. 52 LangOptions Features; // Features enabled by this language (cache). 53 bool Is_PragmaLexer; // True if lexer for _Pragma handling. 54 55 //===--------------------------------------------------------------------===// 56 // Context-specific lexing flags set by the preprocessor. 57 // 58 59 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace 60 /// and return them as tokens. This is used for -C and -CC modes, and 61 /// whitespace preservation can be useful for some clients that want to lex 62 /// the file in raw mode and get every character from the file. 63 /// 64 /// When this is set to 2 it returns comments and whitespace. When set to 1 65 /// it returns comments, when it is set to 0 it returns normal tokens only. 66 unsigned char ExtendedTokenMode; 67 68 //===--------------------------------------------------------------------===// 69 // Context that changes as the file is lexed. 70 // NOTE: any state that mutates when in raw mode must have save/restore code 71 // in Lexer::isNextPPTokenLParen. 72 73 // BufferPtr - Current pointer into the buffer. This is the next character 74 // to be lexed. 75 const char *BufferPtr; 76 77 // IsAtStartOfLine - True if the next lexed token should get the "start of 78 // line" flag set on it. 79 bool IsAtStartOfLine; 80 81 // CurrentConflictMarkerState - The kind of conflict marker we are handling. 82 ConflictMarkerKind CurrentConflictMarkerState; 83 84 Lexer(const Lexer&); // DO NOT IMPLEMENT 85 void operator=(const Lexer&); // DO NOT IMPLEMENT 86 friend class Preprocessor; 87 88 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd); 89 public: 90 91 /// Lexer constructor - Create a new lexer object for the specified buffer 92 /// with the specified preprocessor managing the lexing process. This lexer 93 /// assumes that the associated file buffer and Preprocessor objects will 94 /// outlive it, so it doesn't take ownership of either of them. 95 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, Preprocessor &PP); 96 97 /// Lexer constructor - Create a new raw lexer object. This object is only 98 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 99 /// range will outlive it, so it doesn't take ownership of it. 100 Lexer(SourceLocation FileLoc, const LangOptions &Features, 101 const char *BufStart, const char *BufPtr, const char *BufEnd); 102 103 /// Lexer constructor - Create a new raw lexer object. This object is only 104 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 105 /// range will outlive it, so it doesn't take ownership of it. 106 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, 107 const SourceManager &SM, const LangOptions &Features); 108 109 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for 110 /// _Pragma expansion. This has a variety of magic semantics that this method 111 /// sets up. It returns a new'd Lexer that must be delete'd when done. 112 static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc, 113 SourceLocation ExpansionLocStart, 114 SourceLocation ExpansionLocEnd, 115 unsigned TokLen, Preprocessor &PP); 116 117 118 /// getFeatures - Return the language features currently enabled. NOTE: this 119 /// lexer modifies features as a file is parsed! 120 const LangOptions &getFeatures() const { return Features; } 121 122 /// getFileLoc - Return the File Location for the file we are lexing out of. 123 /// The physical location encodes the location where the characters come from, 124 /// the virtual location encodes where we should *claim* the characters came 125 /// from. Currently this is only used by _Pragma handling. 126 SourceLocation getFileLoc() const { return FileLoc; } 127 128 /// Lex - Return the next token in the file. If this is the end of file, it 129 /// return the tok::eof token. Return true if an error occurred and 130 /// compilation should terminate, false if normal. This implicitly involves 131 /// the preprocessor. 132 void Lex(Token &Result) { 133 // Start a new token. 134 Result.startToken(); 135 136 // NOTE, any changes here should also change code after calls to 137 // Preprocessor::HandleDirective 138 if (IsAtStartOfLine) { 139 Result.setFlag(Token::StartOfLine); 140 IsAtStartOfLine = false; 141 } 142 143 // Get a token. Note that this may delete the current lexer if the end of 144 // file is reached. 145 LexTokenInternal(Result); 146 } 147 148 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 149 bool isPragmaLexer() const { return Is_PragmaLexer; } 150 151 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 152 /// the PreprocessorLexer interface. 153 void IndirectLex(Token &Result) { Lex(Result); } 154 155 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 156 /// associated preprocessor object. Return true if the 'next character to 157 /// read' pointer points at the end of the lexer buffer, false otherwise. 158 bool LexFromRawLexer(Token &Result) { 159 assert(LexingRawMode && "Not already in raw mode!"); 160 Lex(Result); 161 // Note that lexing to the end of the buffer doesn't implicitly delete the 162 // lexer when in raw mode. 163 return BufferPtr == BufferEnd; 164 } 165 166 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 167 /// every character in the file, including whitespace and comments. This 168 /// should only be used in raw mode, as the preprocessor is not prepared to 169 /// deal with the excess tokens. 170 bool isKeepWhitespaceMode() const { 171 return ExtendedTokenMode > 1; 172 } 173 174 /// SetKeepWhitespaceMode - This method lets clients enable or disable 175 /// whitespace retention mode. 176 void SetKeepWhitespaceMode(bool Val) { 177 assert((!Val || LexingRawMode) && 178 "Can only enable whitespace retention in raw mode"); 179 ExtendedTokenMode = Val ? 2 : 0; 180 } 181 182 /// inKeepCommentMode - Return true if the lexer should return comments as 183 /// tokens. 184 bool inKeepCommentMode() const { 185 return ExtendedTokenMode > 0; 186 } 187 188 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 189 /// to the specified mode. This is really only useful when lexing in raw 190 /// mode, because otherwise the lexer needs to manage this. 191 void SetCommentRetentionState(bool Mode) { 192 assert(!isKeepWhitespaceMode() && 193 "Can't play with comment retention state when retaining whitespace"); 194 ExtendedTokenMode = Mode ? 1 : 0; 195 } 196 197 const char *getBufferStart() const { return BufferStart; } 198 199 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 200 /// uninterpreted string. This switches the lexer out of directive mode. 201 std::string ReadToEndOfLine(); 202 203 204 /// Diag - Forwarding function for diagnostics. This translate a source 205 /// position in the current buffer into a SourceLocation object for rendering. 206 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const; 207 208 /// getSourceLocation - Return a source location identifier for the specified 209 /// offset in the current file. 210 SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const; 211 212 /// getSourceLocation - Return a source location for the next character in 213 /// the current file. 214 SourceLocation getSourceLocation() { return getSourceLocation(BufferPtr); } 215 216 /// \brief Return the current location in the buffer. 217 const char *getBufferLocation() const { return BufferPtr; } 218 219 /// Stringify - Convert the specified string into a C string by escaping '\' 220 /// and " characters. This does not add surrounding ""'s to the string. 221 /// If Charify is true, this escapes the ' character instead of ". 222 static std::string Stringify(const std::string &Str, bool Charify = false); 223 224 /// Stringify - Convert the specified string into a C string by escaping '\' 225 /// and " characters. This does not add surrounding ""'s to the string. 226 static void Stringify(SmallVectorImpl<char> &Str); 227 228 229 /// getSpelling - This method is used to get the spelling of a token into a 230 /// preallocated buffer, instead of as an std::string. The caller is required 231 /// to allocate enough space for the token, which is guaranteed to be at least 232 /// Tok.getLength() bytes long. The length of the actual result is returned. 233 /// 234 /// Note that this method may do two possible things: it may either fill in 235 /// the buffer specified with characters, or it may *change the input pointer* 236 /// to point to a constant buffer with the data already in it (avoiding a 237 /// copy). The caller is not allowed to modify the returned buffer pointer 238 /// if an internal buffer is returned. 239 static unsigned getSpelling(const Token &Tok, const char *&Buffer, 240 const SourceManager &SourceMgr, 241 const LangOptions &Features, 242 bool *Invalid = 0); 243 244 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 245 /// token is the characters used to represent the token in the source file 246 /// after trigraph expansion and escaped-newline folding. In particular, this 247 /// wants to get the true, uncanonicalized, spelling of things like digraphs 248 /// UCNs, etc. 249 static std::string getSpelling(const Token &Tok, 250 const SourceManager &SourceMgr, 251 const LangOptions &Features, 252 bool *Invalid = 0); 253 254 /// getSpelling - This method is used to get the spelling of the 255 /// token at the given source location. If, as is usually true, it 256 /// is not necessary to copy any data, then the returned string may 257 /// not point into the provided buffer. 258 /// 259 /// This method lexes at the expansion depth of the given 260 /// location and does not jump to the expansion or spelling 261 /// location. 262 static StringRef getSpelling(SourceLocation loc, 263 SmallVectorImpl<char> &buffer, 264 const SourceManager &SourceMgr, 265 const LangOptions &Features, 266 bool *invalid = 0); 267 268 /// MeasureTokenLength - Relex the token at the specified location and return 269 /// its length in bytes in the input file. If the token needs cleaning (e.g. 270 /// includes a trigraph or an escaped newline) then this count includes bytes 271 /// that are part of that. 272 static unsigned MeasureTokenLength(SourceLocation Loc, 273 const SourceManager &SM, 274 const LangOptions &LangOpts); 275 276 /// \brief Given a location any where in a source buffer, find the location 277 /// that corresponds to the beginning of the token in which the original 278 /// source location lands. 279 /// 280 /// \param Loc 281 static SourceLocation GetBeginningOfToken(SourceLocation Loc, 282 const SourceManager &SM, 283 const LangOptions &LangOpts); 284 285 /// AdvanceToTokenCharacter - If the current SourceLocation specifies a 286 /// location at the start of a token, return a new location that specifies a 287 /// character within the token. This handles trigraphs and escaped newlines. 288 static SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 289 unsigned Character, 290 const SourceManager &SM, 291 const LangOptions &Features); 292 293 /// \brief Computes the source location just past the end of the 294 /// token at this source location. 295 /// 296 /// This routine can be used to produce a source location that 297 /// points just past the end of the token referenced by \p Loc, and 298 /// is generally used when a diagnostic needs to point just after a 299 /// token where it expected something different that it received. If 300 /// the returned source location would not be meaningful (e.g., if 301 /// it points into a macro), this routine returns an invalid 302 /// source location. 303 /// 304 /// \param Offset an offset from the end of the token, where the source 305 /// location should refer to. The default offset (0) produces a source 306 /// location pointing just past the end of the token; an offset of 1 produces 307 /// a source location pointing to the last character in the token, etc. 308 static SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset, 309 const SourceManager &SM, 310 const LangOptions &Features); 311 312 /// \brief Returns true if the given MacroID location points at the first 313 /// token of the macro expansion. 314 static bool isAtStartOfMacroExpansion(SourceLocation loc, 315 const SourceManager &SM, 316 const LangOptions &LangOpts); 317 318 /// \brief Returns true if the given MacroID location points at the last 319 /// token of the macro expansion. 320 static bool isAtEndOfMacroExpansion(SourceLocation loc, 321 const SourceManager &SM, 322 const LangOptions &LangOpts); 323 324 /// \brief Compute the preamble of the given file. 325 /// 326 /// The preamble of a file contains the initial comments, include directives, 327 /// and other preprocessor directives that occur before the code in this 328 /// particular file actually begins. The preamble of the main source file is 329 /// a potential prefix header. 330 /// 331 /// \param Buffer The memory buffer containing the file's contents. 332 /// 333 /// \param MaxLines If non-zero, restrict the length of the preamble 334 /// to fewer than this number of lines. 335 /// 336 /// \returns The offset into the file where the preamble ends and the rest 337 /// of the file begins along with a boolean value indicating whether 338 /// the preamble ends at the beginning of a new line. 339 static std::pair<unsigned, bool> 340 ComputePreamble(const llvm::MemoryBuffer *Buffer, const LangOptions &Features, 341 unsigned MaxLines = 0); 342 343 //===--------------------------------------------------------------------===// 344 // Internal implementation interfaces. 345 private: 346 347 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 348 /// by Lex. 349 /// 350 void LexTokenInternal(Token &Result); 351 352 /// FormTokenWithChars - When we lex a token, we have identified a span 353 /// starting at BufferPtr, going to TokEnd that forms the token. This method 354 /// takes that range and assigns it to the token as its location and size. In 355 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 356 /// TokEnd. 357 void FormTokenWithChars(Token &Result, const char *TokEnd, 358 tok::TokenKind Kind) { 359 unsigned TokLen = TokEnd-BufferPtr; 360 Result.setLength(TokLen); 361 Result.setLocation(getSourceLocation(BufferPtr, TokLen)); 362 Result.setKind(Kind); 363 BufferPtr = TokEnd; 364 } 365 366 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 367 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 368 /// tokens in the buffer controlled by this lexer. 369 unsigned isNextPPTokenLParen(); 370 371 //===--------------------------------------------------------------------===// 372 // Lexer character reading interfaces. 373 public: 374 375 // This lexer is built on two interfaces for reading characters, both of which 376 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 377 // when we know that we will be reading a character from the input buffer and 378 // that this character will be part of the result token. This occurs in (f.e.) 379 // string processing, because we know we need to read until we find the 380 // closing '"' character. 381 // 382 // The second interface is the combination of getCharAndSize with 383 // ConsumeChar. getCharAndSize reads a phase 1/2 translated character, 384 // returning it and its size. If the lexer decides that this character is 385 // part of the current token, it calls ConsumeChar on it. This two stage 386 // approach allows us to emit diagnostics for characters (e.g. warnings about 387 // trigraphs), knowing that they only are emitted if the character is 388 // consumed. 389 390 /// isObviouslySimpleCharacter - Return true if the specified character is 391 /// obviously the same in translation phase 1 and translation phase 3. This 392 /// can return false for characters that end up being the same, but it will 393 /// never return true for something that needs to be mapped. 394 static bool isObviouslySimpleCharacter(char C) { 395 return C != '?' && C != '\\'; 396 } 397 398 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 399 /// advance over it, and return it. This is tricky in several cases. Here we 400 /// just handle the trivial case and fall-back to the non-inlined 401 /// getCharAndSizeSlow method to handle the hard case. 402 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 403 // If this is not a trigraph and not a UCN or escaped newline, return 404 // quickly. 405 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 406 407 unsigned Size = 0; 408 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 409 Ptr += Size; 410 return C; 411 } 412 413 private: 414 /// ConsumeChar - When a character (identified by getCharAndSize) is consumed 415 /// and added to a given token, check to see if there are diagnostics that 416 /// need to be emitted or flags that need to be set on the token. If so, do 417 /// it. 418 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 419 // Normal case, we consumed exactly one token. Just return it. 420 if (Size == 1) 421 return Ptr+Size; 422 423 // Otherwise, re-lex the character with a current token, allowing 424 // diagnostics to be emitted and flags to be set. 425 Size = 0; 426 getCharAndSizeSlow(Ptr, Size, &Tok); 427 return Ptr+Size; 428 } 429 430 /// getCharAndSize - Peek a single 'character' from the specified buffer, 431 /// get its size, and return it. This is tricky in several cases. Here we 432 /// just handle the trivial case and fall-back to the non-inlined 433 /// getCharAndSizeSlow method to handle the hard case. 434 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 435 // If this is not a trigraph and not a UCN or escaped newline, return 436 // quickly. 437 if (isObviouslySimpleCharacter(Ptr[0])) { 438 Size = 1; 439 return *Ptr; 440 } 441 442 Size = 0; 443 return getCharAndSizeSlow(Ptr, Size); 444 } 445 446 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 447 /// method. 448 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 449 public: 450 451 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 452 /// emit a warning. 453 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 454 const LangOptions &Features) { 455 // If this is not a trigraph and not a UCN or escaped newline, return 456 // quickly. 457 if (isObviouslySimpleCharacter(Ptr[0])) { 458 Size = 1; 459 return *Ptr; 460 } 461 462 Size = 0; 463 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 464 } 465 466 /// getEscapedNewLineSize - Return the size of the specified escaped newline, 467 /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry 468 /// to this function. 469 static unsigned getEscapedNewLineSize(const char *P); 470 471 /// SkipEscapedNewLines - If P points to an escaped newline (or a series of 472 /// them), skip over them and return the first non-escaped-newline found, 473 /// otherwise return P. 474 static const char *SkipEscapedNewLines(const char *P); 475 476 /// \brief Checks that the given token is the first token that occurs after 477 /// the given location (this excludes comments and whitespace). Returns the 478 /// location immediately after the specified token. If the token is not found 479 /// or the location is inside a macro, the returned source location will be 480 /// invalid. 481 static SourceLocation findLocationAfterToken(SourceLocation loc, 482 tok::TokenKind TKind, 483 const SourceManager &SM, 484 const LangOptions &LangOpts, 485 bool SkipTrailingWhitespaceAndNewLine); 486 487 private: 488 489 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 490 /// diagnostic. 491 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 492 const LangOptions &Features); 493 494 //===--------------------------------------------------------------------===// 495 // Other lexer functions. 496 497 void SkipBytes(unsigned Bytes, bool StartOfLine); 498 499 // Helper functions to lex the remainder of a token of the specific type. 500 void LexIdentifier (Token &Result, const char *CurPtr); 501 void LexNumericConstant (Token &Result, const char *CurPtr); 502 void LexStringLiteral (Token &Result, const char *CurPtr, 503 tok::TokenKind Kind); 504 void LexRawStringLiteral (Token &Result, const char *CurPtr, 505 tok::TokenKind Kind); 506 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 507 void LexCharConstant (Token &Result, const char *CurPtr, 508 tok::TokenKind Kind); 509 bool LexEndOfFile (Token &Result, const char *CurPtr); 510 511 bool SkipWhitespace (Token &Result, const char *CurPtr); 512 bool SkipBCPLComment (Token &Result, const char *CurPtr); 513 bool SkipBlockComment (Token &Result, const char *CurPtr); 514 bool SaveBCPLComment (Token &Result, const char *CurPtr); 515 516 bool IsStartOfConflictMarker(const char *CurPtr); 517 bool HandleEndOfConflictMarker(const char *CurPtr); 518 519 bool isCodeCompletionPoint(const char *CurPtr) const; 520 void cutOffLexing() { BufferPtr = BufferEnd; } 521 }; 522 523 524 } // end namespace clang 525 526 #endif 527