Lexer.h revision 1eb4433ac451dc16f4133a88af2d002ac26c58ef
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; // True if lexer for _Pragma handling. 42 43 //===--------------------------------------------------------------------===// 44 // Context-specific lexing flags set by the preprocessor. 45 // 46 47 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace 48 /// and return them as tokens. This is used for -C and -CC modes, and 49 /// whitespace preservation can be useful for some clients that want to lex 50 /// the file in raw mode and get every character from the file. 51 /// 52 /// When this is set to 2 it returns comments and whitespace. When set to 1 53 /// it returns comments, when it is set to 0 it returns normal tokens only. 54 unsigned char ExtendedTokenMode; 55 56 //===--------------------------------------------------------------------===// 57 // Context that changes as the file is lexed. 58 // NOTE: any state that mutates when in raw mode must have save/restore code 59 // in Lexer::isNextPPTokenLParen. 60 61 // BufferPtr - Current pointer into the buffer. This is the next character 62 // to be lexed. 63 const char *BufferPtr; 64 65 // IsAtStartOfLine - True if the next lexed token should get the "start of 66 // line" flag set on it. 67 bool IsAtStartOfLine; 68 69 Lexer(const Lexer&); // DO NOT IMPLEMENT 70 void operator=(const Lexer&); // DO NOT IMPLEMENT 71 friend class Preprocessor; 72 73 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd); 74public: 75 76 /// Lexer constructor - Create a new lexer object for the specified buffer 77 /// with the specified preprocessor managing the lexing process. This lexer 78 /// assumes that the associated file buffer and Preprocessor objects will 79 /// outlive it, so it doesn't take ownership of either of them. 80 Lexer(FileID FID, Preprocessor &PP); 81 82 /// Lexer constructor - Create a new raw lexer object. This object is only 83 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 84 /// range will outlive it, so it doesn't take ownership of it. 85 Lexer(SourceLocation FileLoc, const LangOptions &Features, 86 const char *BufStart, const char *BufPtr, const char *BufEnd); 87 88 /// Lexer constructor - Create a new raw lexer object. This object is only 89 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 90 /// range will outlive it, so it doesn't take ownership of it. 91 Lexer(FileID FID, const SourceManager &SM, const LangOptions &Features); 92 93 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for 94 /// _Pragma expansion. This has a variety of magic semantics that this method 95 /// sets up. It returns a new'd Lexer that must be delete'd when done. 96 static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc, 97 SourceLocation InstantiationLocStart, 98 SourceLocation InstantiationLocEnd, 99 unsigned TokLen, Preprocessor &PP); 100 101 102 /// getFeatures - Return the language features currently enabled. NOTE: this 103 /// lexer modifies features as a file is parsed! 104 const LangOptions &getFeatures() const { return Features; } 105 106 /// getFileLoc - Return the File Location for the file we are lexing out of. 107 /// The physical location encodes the location where the characters come from, 108 /// the virtual location encodes where we should *claim* the characters came 109 /// from. Currently this is only used by _Pragma handling. 110 SourceLocation getFileLoc() const { return FileLoc; } 111 112 /// Lex - Return the next token in the file. If this is the end of file, it 113 /// return the tok::eof token. Return true if an error occurred and 114 /// compilation should terminate, false if normal. This implicitly involves 115 /// the preprocessor. 116 void Lex(Token &Result) { 117 // Start a new token. 118 Result.startToken(); 119 120 // NOTE, any changes here should also change code after calls to 121 // Preprocessor::HandleDirective 122 if (IsAtStartOfLine) { 123 Result.setFlag(Token::StartOfLine); 124 IsAtStartOfLine = false; 125 } 126 127 // Get a token. Note that this may delete the current lexer if the end of 128 // file is reached. 129 LexTokenInternal(Result); 130 } 131 132 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 133 bool isPragmaLexer() const { return Is_PragmaLexer; } 134 135 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 136 /// the PreprocessorLexer interface. 137 void IndirectLex(Token &Result) { Lex(Result); } 138 139 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 140 /// associated preprocessor object. Return true if the 'next character to 141 /// read' pointer points at the end of the lexer buffer, false otherwise. 142 bool LexFromRawLexer(Token &Result) { 143 assert(LexingRawMode && "Not already in raw mode!"); 144 Lex(Result); 145 // Note that lexing to the end of the buffer doesn't implicitly delete the 146 // lexer when in raw mode. 147 return BufferPtr == BufferEnd; 148 } 149 150 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 151 /// every character in the file, including whitespace and comments. This 152 /// should only be used in raw mode, as the preprocessor is not prepared to 153 /// deal with the excess tokens. 154 bool isKeepWhitespaceMode() const { 155 return ExtendedTokenMode > 1; 156 } 157 158 /// SetKeepWhitespaceMode - This method lets clients enable or disable 159 /// whitespace retention mode. 160 void SetKeepWhitespaceMode(bool Val) { 161 assert((!Val || LexingRawMode) && 162 "Can only enable whitespace retention in raw mode"); 163 ExtendedTokenMode = Val ? 2 : 0; 164 } 165 166 /// inKeepCommentMode - Return true if the lexer should return comments as 167 /// tokens. 168 bool inKeepCommentMode() const { 169 return ExtendedTokenMode > 0; 170 } 171 172 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 173 /// to the specified mode. This is really only useful when lexing in raw 174 /// mode, because otherwise the lexer needs to manage this. 175 void SetCommentRetentionState(bool Mode) { 176 assert(!isKeepWhitespaceMode() && 177 "Can't play with comment retention state when retaining whitespace"); 178 ExtendedTokenMode = Mode ? 1 : 0; 179 } 180 181 const char *getBufferStart() const { return BufferStart; } 182 183 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 184 /// uninterpreted string. This switches the lexer out of directive mode. 185 std::string ReadToEndOfLine(); 186 187 188 /// Diag - Forwarding function for diagnostics. This translate a source 189 /// position in the current buffer into a SourceLocation object for rendering. 190 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const; 191 192 /// getSourceLocation - Return a source location identifier for the specified 193 /// offset in the current file. 194 SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const; 195 196 /// getSourceLocation - Return a source location for the next character in 197 /// the current file. 198 SourceLocation getSourceLocation() { return getSourceLocation(BufferPtr); } 199 200 /// Stringify - Convert the specified string into a C string by escaping '\' 201 /// and " characters. This does not add surrounding ""'s to the string. 202 /// If Charify is true, this escapes the ' character instead of ". 203 static std::string Stringify(const std::string &Str, bool Charify = false); 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 static void Stringify(llvm::SmallVectorImpl<char> &Str); 208 209 /// MeasureTokenLength - Relex the token at the specified location and return 210 /// its length in bytes in the input file. If the token needs cleaning (e.g. 211 /// includes a trigraph or an escaped newline) then this count includes bytes 212 /// that are part of that. 213 static unsigned MeasureTokenLength(SourceLocation Loc, 214 const SourceManager &SM, 215 const LangOptions &LangOpts); 216 217 //===--------------------------------------------------------------------===// 218 // Internal implementation interfaces. 219private: 220 221 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 222 /// by Lex. 223 /// 224 void LexTokenInternal(Token &Result); 225 226 /// FormTokenWithChars - When we lex a token, we have identified a span 227 /// starting at BufferPtr, going to TokEnd that forms the token. This method 228 /// takes that range and assigns it to the token as its location and size. In 229 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 230 /// TokEnd. 231 void FormTokenWithChars(Token &Result, const char *TokEnd, 232 tok::TokenKind Kind) { 233 unsigned TokLen = TokEnd-BufferPtr; 234 Result.setLength(TokLen); 235 Result.setLocation(getSourceLocation(BufferPtr, TokLen)); 236 Result.setKind(Kind); 237 BufferPtr = TokEnd; 238 } 239 240 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 241 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 242 /// tokens in the buffer controlled by this lexer. 243 unsigned isNextPPTokenLParen(); 244 245 //===--------------------------------------------------------------------===// 246 // Lexer character reading interfaces. 247public: 248 249 // This lexer is built on two interfaces for reading characters, both of which 250 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 251 // when we know that we will be reading a character from the input buffer and 252 // that this character will be part of the result token. This occurs in (f.e.) 253 // string processing, because we know we need to read until we find the 254 // closing '"' character. 255 // 256 // The second interface is the combination of PeekCharAndSize with 257 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 258 // returning it and its size. If the lexer decides that this character is 259 // part of the current token, it calls ConsumeChar on it. This two stage 260 // approach allows us to emit diagnostics for characters (e.g. warnings about 261 // trigraphs), knowing that they only are emitted if the character is 262 // consumed. 263 264 /// isObviouslySimpleCharacter - Return true if the specified character is 265 /// obviously the same in translation phase 1 and translation phase 3. This 266 /// can return false for characters that end up being the same, but it will 267 /// never return true for something that needs to be mapped. 268 static bool isObviouslySimpleCharacter(char C) { 269 return C != '?' && C != '\\'; 270 } 271 272 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 273 /// advance over it, and return it. This is tricky in several cases. Here we 274 /// just handle the trivial case and fall-back to the non-inlined 275 /// getCharAndSizeSlow method to handle the hard case. 276 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 277 // If this is not a trigraph and not a UCN or escaped newline, return 278 // quickly. 279 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 280 281 unsigned Size = 0; 282 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 283 Ptr += Size; 284 return C; 285 } 286 287private: 288 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 289 /// and added to a given token, check to see if there are diagnostics that 290 /// need to be emitted or flags that need to be set on the token. If so, do 291 /// it. 292 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 293 // Normal case, we consumed exactly one token. Just return it. 294 if (Size == 1) 295 return Ptr+Size; 296 297 // Otherwise, re-lex the character with a current token, allowing 298 // diagnostics to be emitted and flags to be set. 299 Size = 0; 300 getCharAndSizeSlow(Ptr, Size, &Tok); 301 return Ptr+Size; 302 } 303 304 /// getCharAndSize - Peek a single 'character' from the specified buffer, 305 /// get its size, and return it. This is tricky in several cases. Here we 306 /// just handle the trivial case and fall-back to the non-inlined 307 /// getCharAndSizeSlow method to handle the hard case. 308 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 309 // If this is not a trigraph and not a UCN or escaped newline, return 310 // quickly. 311 if (isObviouslySimpleCharacter(Ptr[0])) { 312 Size = 1; 313 return *Ptr; 314 } 315 316 Size = 0; 317 return getCharAndSizeSlow(Ptr, Size); 318 } 319 320 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 321 /// method. 322 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 323public: 324 325 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 326 /// emit a warning. 327 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 328 const LangOptions &Features) { 329 // If this is not a trigraph and not a UCN or escaped newline, return 330 // quickly. 331 if (isObviouslySimpleCharacter(Ptr[0])) { 332 Size = 1; 333 return *Ptr; 334 } 335 336 Size = 0; 337 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 338 } 339 340 /// getEscapedNewLineSize - Return the size of the specified escaped newline, 341 /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry 342 /// to this function. 343 static unsigned getEscapedNewLineSize(const char *P); 344 345 /// SkipEscapedNewLines - If P points to an escaped newline (or a series of 346 /// them), skip over them and return the first non-escaped-newline found, 347 /// otherwise return P. 348 static const char *SkipEscapedNewLines(const char *P); 349private: 350 351 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 352 /// diagnostic. 353 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 354 const LangOptions &Features); 355 356 //===--------------------------------------------------------------------===// 357 // Other lexer functions. 358 359 // Helper functions to lex the remainder of a token of the specific type. 360 void LexIdentifier (Token &Result, const char *CurPtr); 361 void LexNumericConstant (Token &Result, const char *CurPtr); 362 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 363 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 364 void LexCharConstant (Token &Result, const char *CurPtr); 365 bool LexEndOfFile (Token &Result, const char *CurPtr); 366 367 bool SkipWhitespace (Token &Result, const char *CurPtr); 368 bool SkipBCPLComment (Token &Result, const char *CurPtr); 369 bool SkipBlockComment (Token &Result, const char *CurPtr); 370 bool SaveBCPLComment (Token &Result, const char *CurPtr); 371}; 372 373 374} // end namespace clang 375 376#endif 377