Lexer.h revision 0016d519b831859526b79405cdae4c64c73731c8
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 "llvm/Support/Allocator.h" 21#include <string> 22#include <vector> 23#include <cassert> 24 25namespace clang { 26class Diagnostic; 27class SourceManager; 28class Preprocessor; 29class DiagnosticBuilder; 30 31/// Lexer - This provides a simple interface that turns a text buffer into a 32/// stream of tokens. This provides no support for file reading or buffering, 33/// or buffering/seeking of tokens, only forward lexing is supported. It relies 34/// on the specified Preprocessor object to handle preprocessor directives, etc. 35class Lexer : public PreprocessorLexer { 36 //===--------------------------------------------------------------------===// 37 // Constant configuration values for this lexer. 38 const char *BufferStart; // Start of the buffer. 39 const char *BufferEnd; // End of the buffer. 40 SourceLocation FileLoc; // Location for start of file. 41 LangOptions Features; // Features enabled by this language (cache). 42 bool Is_PragmaLexer : 1; // True if lexer for _Pragma handling. 43 bool IsInConflictMarker : 1; // True if in a VCS conflict marker '<<<<<<<' 44 45 //===--------------------------------------------------------------------===// 46 // Context-specific lexing flags set by the preprocessor. 47 // 48 49 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace 50 /// and return them as tokens. This is used for -C and -CC modes, and 51 /// whitespace preservation can be useful for some clients that want to lex 52 /// the file in raw mode and get every character from the file. 53 /// 54 /// When this is set to 2 it returns comments and whitespace. When set to 1 55 /// it returns comments, when it is set to 0 it returns normal tokens only. 56 unsigned char ExtendedTokenMode; 57 58 //===--------------------------------------------------------------------===// 59 // Context that changes as the file is lexed. 60 // NOTE: any state that mutates when in raw mode must have save/restore code 61 // in Lexer::isNextPPTokenLParen. 62 63 // BufferPtr - Current pointer into the buffer. This is the next character 64 // to be lexed. 65 const char *BufferPtr; 66 67 // IsAtStartOfLine - True if the next lexed token should get the "start of 68 // line" flag set on it. 69 bool IsAtStartOfLine; 70 71 // ExtraDataAllocator - An allocator for extra data on a token. 72 llvm::BumpPtrAllocator ExtraDataAllocator; 73 74 Lexer(const Lexer&); // DO NOT IMPLEMENT 75 void operator=(const Lexer&); // DO NOT IMPLEMENT 76 friend class Preprocessor; 77 78 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd); 79public: 80 81 /// Lexer constructor - Create a new lexer object for the specified buffer 82 /// with the specified preprocessor managing the lexing process. This lexer 83 /// assumes that the associated file buffer and Preprocessor objects will 84 /// outlive it, so it doesn't take ownership of either of them. 85 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, Preprocessor &PP); 86 87 /// Lexer constructor - Create a new raw lexer object. This object is only 88 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 89 /// range will outlive it, so it doesn't take ownership of it. 90 Lexer(SourceLocation FileLoc, const LangOptions &Features, 91 const char *BufStart, const char *BufPtr, const char *BufEnd); 92 93 /// Lexer constructor - Create a new raw lexer object. This object is only 94 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 95 /// range will outlive it, so it doesn't take ownership of it. 96 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, 97 const SourceManager &SM, const LangOptions &Features); 98 99 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for 100 /// _Pragma expansion. This has a variety of magic semantics that this method 101 /// sets up. It returns a new'd Lexer that must be delete'd when done. 102 static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc, 103 SourceLocation InstantiationLocStart, 104 SourceLocation InstantiationLocEnd, 105 unsigned TokLen, Preprocessor &PP); 106 107 108 /// getFeatures - Return the language features currently enabled. NOTE: this 109 /// lexer modifies features as a file is parsed! 110 const LangOptions &getFeatures() const { return Features; } 111 112 /// getFileLoc - Return the File Location for the file we are lexing out of. 113 /// The physical location encodes the location where the characters come from, 114 /// the virtual location encodes where we should *claim* the characters came 115 /// from. Currently this is only used by _Pragma handling. 116 SourceLocation getFileLoc() const { return FileLoc; } 117 118 /// Lex - Return the next token in the file. If this is the end of file, it 119 /// return the tok::eof token. Return true if an error occurred and 120 /// compilation should terminate, false if normal. This implicitly involves 121 /// the preprocessor. 122 void Lex(Token &Result) { 123 // Start a new token. 124 Result.startToken(); 125 126 // NOTE, any changes here should also change code after calls to 127 // Preprocessor::HandleDirective 128 if (IsAtStartOfLine) { 129 Result.setFlag(Token::StartOfLine); 130 IsAtStartOfLine = false; 131 } 132 133 // Get a token. Note that this may delete the current lexer if the end of 134 // file is reached. 135 LexTokenInternal(Result); 136 } 137 138 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 139 bool isPragmaLexer() const { return Is_PragmaLexer; } 140 141 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 142 /// the PreprocessorLexer interface. 143 void IndirectLex(Token &Result) { Lex(Result); } 144 145 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 146 /// associated preprocessor object. Return true if the 'next character to 147 /// read' pointer points at the end of the lexer buffer, false otherwise. 148 bool LexFromRawLexer(Token &Result) { 149 assert(LexingRawMode && "Not already in raw mode!"); 150 Lex(Result); 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 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 157 /// every character in the file, including whitespace and comments. This 158 /// should only be used in raw mode, as the preprocessor is not prepared to 159 /// deal with the excess tokens. 160 bool isKeepWhitespaceMode() const { 161 return ExtendedTokenMode > 1; 162 } 163 164 /// SetKeepWhitespaceMode - This method lets clients enable or disable 165 /// whitespace retention mode. 166 void SetKeepWhitespaceMode(bool Val) { 167 assert((!Val || LexingRawMode) && 168 "Can only enable whitespace retention in raw mode"); 169 ExtendedTokenMode = Val ? 2 : 0; 170 } 171 172 /// inKeepCommentMode - Return true if the lexer should return comments as 173 /// tokens. 174 bool inKeepCommentMode() const { 175 return ExtendedTokenMode > 0; 176 } 177 178 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 179 /// to the specified mode. This is really only useful when lexing in raw 180 /// mode, because otherwise the lexer needs to manage this. 181 void SetCommentRetentionState(bool Mode) { 182 assert(!isKeepWhitespaceMode() && 183 "Can't play with comment retention state when retaining whitespace"); 184 ExtendedTokenMode = Mode ? 1 : 0; 185 } 186 187 const char *getBufferStart() const { return BufferStart; } 188 189 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 190 /// uninterpreted string. This switches the lexer out of directive mode. 191 std::string ReadToEndOfLine(); 192 193 194 /// Diag - Forwarding function for diagnostics. This translate a source 195 /// position in the current buffer into a SourceLocation object for rendering. 196 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const; 197 198 /// getSourceLocation - Return a source location identifier for the specified 199 /// offset in the current file. 200 SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const; 201 202 /// getSourceLocation - Return a source location for the next character in 203 /// the current file. 204 SourceLocation getSourceLocation() { return getSourceLocation(BufferPtr); } 205 206 /// \brief Return the current location in the buffer. 207 const char *getBufferLocation() const { return BufferPtr; } 208 209 /// Stringify - Convert the specified string into a C string by escaping '\' 210 /// and " characters. This does not add surrounding ""'s to the string. 211 /// If Charify is true, this escapes the ' character instead of ". 212 static std::string Stringify(const std::string &Str, bool Charify = false); 213 214 /// Stringify - Convert the specified string into a C string by escaping '\' 215 /// and " characters. This does not add surrounding ""'s to the string. 216 static void Stringify(llvm::SmallVectorImpl<char> &Str); 217 218 /// MeasureTokenLength - Relex the token at the specified location and return 219 /// its length in bytes in the input file. If the token needs cleaning (e.g. 220 /// includes a trigraph or an escaped newline) then this count includes bytes 221 /// that are part of that. 222 static unsigned MeasureTokenLength(SourceLocation Loc, 223 const SourceManager &SM, 224 const LangOptions &LangOpts); 225 226 /// \brief Given a location any where in a source buffer, find the location 227 /// that corresponds to the beginning of the token in which the original 228 /// source location lands. 229 /// 230 /// \param Loc 231 static SourceLocation GetBeginningOfToken(SourceLocation Loc, 232 const SourceManager &SM, 233 const LangOptions &LangOpts); 234 235 /// \brief Compute the preamble of the given file. 236 /// 237 /// The preamble of a file contains the initial comments, include directives, 238 /// and other preprocessor directives that occur before the code in this 239 /// particular file actually begins. The preamble of the main source file is 240 /// a potential prefix header. 241 /// 242 /// \param Buffer The memory buffer containing the file's contents. 243 /// 244 /// \param MaxLines If non-zero, restrict the length of the preamble 245 /// to fewer than this number of lines. 246 /// 247 /// \returns The offset into the file where the preamble ends and the rest 248 /// of the file begins along with a boolean value indicating whether 249 /// the preamble ends at the beginning of a new line. 250 static std::pair<unsigned, bool> 251 ComputePreamble(const llvm::MemoryBuffer *Buffer, unsigned MaxLines = 0); 252 253 //===--------------------------------------------------------------------===// 254 // Internal implementation interfaces. 255private: 256 257 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 258 /// by Lex. 259 /// 260 void LexTokenInternal(Token &Result); 261 262 /// FormTokenWithChars - When we lex a token, we have identified a span 263 /// starting at BufferPtr, going to TokEnd that forms the token. This method 264 /// takes that range and assigns it to the token as its location and size. In 265 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 266 /// TokEnd. 267 void FormTokenWithChars(Token &Result, const char *TokEnd, 268 tok::TokenKind Kind) { 269 unsigned TokLen = TokEnd-BufferPtr; 270 Result.setLength(TokLen); 271 Result.setLocation(getSourceLocation(BufferPtr, TokLen)); 272 Result.setKind(Kind); 273 BufferPtr = TokEnd; 274 } 275 276 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 277 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 278 /// tokens in the buffer controlled by this lexer. 279 unsigned isNextPPTokenLParen(); 280 281 //===--------------------------------------------------------------------===// 282 // Lexer character reading interfaces. 283public: 284 285 // This lexer is built on two interfaces for reading characters, both of which 286 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 287 // when we know that we will be reading a character from the input buffer and 288 // that this character will be part of the result token. This occurs in (f.e.) 289 // string processing, because we know we need to read until we find the 290 // closing '"' character. 291 // 292 // The second interface is the combination of getCharAndSize with 293 // ConsumeChar. getCharAndSize reads a phase 1/2 translated character, 294 // returning it and its size. If the lexer decides that this character is 295 // part of the current token, it calls ConsumeChar on it. This two stage 296 // approach allows us to emit diagnostics for characters (e.g. warnings about 297 // trigraphs), knowing that they only are emitted if the character is 298 // consumed. 299 300 /// isObviouslySimpleCharacter - Return true if the specified character is 301 /// obviously the same in translation phase 1 and translation phase 3. This 302 /// can return false for characters that end up being the same, but it will 303 /// never return true for something that needs to be mapped. 304 static bool isObviouslySimpleCharacter(char C) { 305 return C != '?' && C != '\\'; 306 } 307 308 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 309 /// advance over it, and return it. This is tricky in several cases. Here we 310 /// just handle the trivial case and fall-back to the non-inlined 311 /// getCharAndSizeSlow method to handle the hard case. 312 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 313 // If this is not a trigraph and not a UCN or escaped newline, return 314 // quickly. 315 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 316 317 unsigned Size = 0; 318 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 319 Ptr += Size; 320 return C; 321 } 322 323private: 324 /// ConsumeChar - When a character (identified by getCharAndSize) is consumed 325 /// and added to a given token, check to see if there are diagnostics that 326 /// need to be emitted or flags that need to be set on the token. If so, do 327 /// it. 328 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 329 // Normal case, we consumed exactly one token. Just return it. 330 if (Size == 1) 331 return Ptr+Size; 332 333 // Otherwise, re-lex the character with a current token, allowing 334 // diagnostics to be emitted and flags to be set. 335 Size = 0; 336 getCharAndSizeSlow(Ptr, Size, &Tok); 337 return Ptr+Size; 338 } 339 340 /// getCharAndSize - Peek a single 'character' from the specified buffer, 341 /// get its size, and return it. This is tricky in several cases. Here we 342 /// just handle the trivial case and fall-back to the non-inlined 343 /// getCharAndSizeSlow method to handle the hard case. 344 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 345 // If this is not a trigraph and not a UCN or escaped newline, return 346 // quickly. 347 if (isObviouslySimpleCharacter(Ptr[0])) { 348 Size = 1; 349 return *Ptr; 350 } 351 352 Size = 0; 353 return getCharAndSizeSlow(Ptr, Size); 354 } 355 356 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 357 /// method. 358 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 359public: 360 361 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 362 /// emit a warning. 363 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 364 const LangOptions &Features) { 365 // If this is not a trigraph and not a UCN or escaped newline, return 366 // quickly. 367 if (isObviouslySimpleCharacter(Ptr[0])) { 368 Size = 1; 369 return *Ptr; 370 } 371 372 Size = 0; 373 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 374 } 375 376 /// getEscapedNewLineSize - Return the size of the specified escaped newline, 377 /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry 378 /// to this function. 379 static unsigned getEscapedNewLineSize(const char *P); 380 381 /// SkipEscapedNewLines - If P points to an escaped newline (or a series of 382 /// them), skip over them and return the first non-escaped-newline found, 383 /// otherwise return P. 384 static const char *SkipEscapedNewLines(const char *P); 385private: 386 387 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 388 /// diagnostic. 389 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 390 const LangOptions &Features); 391 392 //===--------------------------------------------------------------------===// 393 // Other lexer functions. 394 395 void SkipBytes(unsigned Bytes, bool StartOfLine); 396 397 // Helper functions to lex the remainder of a token of the specific type. 398 void LexIdentifier (Token &Result, const char *CurPtr); 399 void LexNumericConstant (Token &Result, const char *CurPtr); 400 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 401 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 402 void LexCharConstant (Token &Result, const char *CurPtr); 403 bool LexEndOfFile (Token &Result, const char *CurPtr); 404 405 bool SkipWhitespace (Token &Result, const char *CurPtr); 406 bool SkipBCPLComment (Token &Result, const char *CurPtr); 407 bool SkipBlockComment (Token &Result, const char *CurPtr); 408 bool SaveBCPLComment (Token &Result, const char *CurPtr); 409 410 bool IsStartOfConflictMarker(const char *CurPtr); 411 bool HandleEndOfConflictMarker(const char *CurPtr); 412}; 413 414 415} // end namespace clang 416 417#endif 418