Lexer.h revision 22d91ca8d7c134eac5cc6a4869e6a84c461ad624
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(SourceLocation FileLoc, Preprocessor &PP, 81 const char *BufStart = 0, const char *BufEnd = 0); 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 *BufEnd, 88 const llvm::MemoryBuffer *FromFile = 0); 89 90 /// getFeatures - Return the language features currently enabled. NOTE: this 91 /// lexer modifies features as a file is parsed! 92 const LangOptions &getFeatures() const { return Features; } 93 94 /// getFileLoc - Return the File Location for the file we are lexing out of. 95 /// The physical location encodes the location where the characters come from, 96 /// the virtual location encodes where we should *claim* the characters came 97 /// from. Currently this is only used by _Pragma handling. 98 SourceLocation getFileLoc() const { return FileLoc; } 99 100 /// Lex - Return the next token in the file. If this is the end of file, it 101 /// return the tok::eof token. Return true if an error occurred and 102 /// compilation should terminate, false if normal. This implicitly involves 103 /// the preprocessor. 104 void Lex(Token &Result) { 105 // Start a new token. 106 Result.startToken(); 107 108 // NOTE, any changes here should also change code after calls to 109 // Preprocessor::HandleDirective 110 if (IsAtStartOfLine) { 111 Result.setFlag(Token::StartOfLine); 112 IsAtStartOfLine = false; 113 } 114 115 // Get a token. Note that this may delete the current lexer if the end of 116 // file is reached. 117 LexTokenInternal(Result); 118 } 119 120 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 121 bool isPragmaLexer() const { return Is_PragmaLexer; } 122 123 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 124 /// the PreprocessorLexer interface. 125 void IndirectLex(Token &Result) { Lex(Result); } 126 127 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 128 /// associated preprocessor object. Return true if the 'next character to 129 /// read' pointer points and the end of the lexer buffer, false otherwise. 130 bool LexFromRawLexer(Token &Result) { 131 assert(LexingRawMode && "Not already in raw mode!"); 132 Lex(Result); 133 // Note that lexing to the end of the buffer doesn't implicitly delete the 134 // lexer when in raw mode. 135 return BufferPtr == BufferEnd; 136 } 137 138 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 139 /// every character in the file, including whitespace and comments. This 140 /// should only be used in raw mode, as the preprocessor is not prepared to 141 /// deal with the excess tokens. 142 bool isKeepWhitespaceMode() const { 143 return ExtendedTokenMode > 1; 144 } 145 146 /// SetKeepWhitespaceMode - This method lets clients enable or disable 147 /// whitespace retention mode. 148 void SetKeepWhitespaceMode(bool Val) { 149 assert((!Val || LexingRawMode) && 150 "Can only enable whitespace retention in raw mode"); 151 ExtendedTokenMode = Val ? 2 : 0; 152 } 153 154 /// inKeepCommentMode - Return true if the lexer should return comments as 155 /// tokens. 156 bool inKeepCommentMode() const { 157 return ExtendedTokenMode > 0; 158 } 159 160 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 161 /// to the specified mode. This is really only useful when lexing in raw 162 /// mode, because otherwise the lexer needs to manage this. 163 void SetCommentRetentionState(bool Mode) { 164 assert(!isKeepWhitespaceMode() && 165 "Can't play with comment retention state when retaining whitespace"); 166 ExtendedTokenMode = Mode ? 1 : 0; 167 } 168 169 170 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 171 /// uninterpreted string. This switches the lexer out of directive mode. 172 std::string ReadToEndOfLine(); 173 174 175 /// Diag - Forwarding function for diagnostics. This translate a source 176 /// position in the current buffer into a SourceLocation object for rendering. 177 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const; 178 179 /// getSourceLocation - Return a source location identifier for the specified 180 /// offset in the current file. 181 SourceLocation getSourceLocation(const char *Loc) const; 182 183 /// getSourceLocation - Return a source location for the next character in 184 /// the current file. 185 SourceLocation getSourceLocation() { return getSourceLocation(BufferPtr); } 186 187 /// Stringify - Convert the specified string into a C string by escaping '\' 188 /// and " characters. This does not add surrounding ""'s to the string. 189 /// If Charify is true, this escapes the ' character instead of ". 190 static std::string Stringify(const std::string &Str, bool Charify = false); 191 192 /// Stringify - Convert the specified string into a C string by escaping '\' 193 /// and " characters. This does not add surrounding ""'s to the string. 194 static void Stringify(llvm::SmallVectorImpl<char> &Str); 195 196 /// MeasureTokenLength - Relex the token at the specified location and return 197 /// its length in bytes in the input file. If the token needs cleaning (e.g. 198 /// includes a trigraph or an escaped newline) then this count includes bytes 199 /// that are part of that. 200 static unsigned MeasureTokenLength(SourceLocation Loc, 201 const SourceManager &SM); 202 203 //===--------------------------------------------------------------------===// 204 // Internal implementation interfaces. 205private: 206 207 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 208 /// by Lex. 209 /// 210 void LexTokenInternal(Token &Result); 211 212 /// FormTokenWithChars - When we lex a token, we have identified a span 213 /// starting at BufferPtr, going to TokEnd that forms the token. This method 214 /// takes that range and assigns it to the token as its location and size. In 215 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 216 /// TokEnd. 217 void FormTokenWithChars(Token &Result, const char *TokEnd, 218 tok::TokenKind Kind) { 219 Result.setLocation(getSourceLocation(BufferPtr)); 220 Result.setLength(TokEnd-BufferPtr); 221 Result.setKind(Kind); 222 BufferPtr = TokEnd; 223 } 224 225 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 226 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 227 /// tokens in the buffer controlled by this lexer. 228 unsigned isNextPPTokenLParen(); 229 230 //===--------------------------------------------------------------------===// 231 // Lexer character reading interfaces. 232public: 233 234 // This lexer is built on two interfaces for reading characters, both of which 235 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 236 // when we know that we will be reading a character from the input buffer and 237 // that this character will be part of the result token. This occurs in (f.e.) 238 // string processing, because we know we need to read until we find the 239 // closing '"' character. 240 // 241 // The second interface is the combination of PeekCharAndSize with 242 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 243 // returning it and its size. If the lexer decides that this character is 244 // part of the current token, it calls ConsumeChar on it. This two stage 245 // approach allows us to emit diagnostics for characters (e.g. warnings about 246 // trigraphs), knowing that they only are emitted if the character is 247 // consumed. 248 249 /// isObviouslySimpleCharacter - Return true if the specified character is 250 /// obviously the same in translation phase 1 and translation phase 3. This 251 /// can return false for characters that end up being the same, but it will 252 /// never return true for something that needs to be mapped. 253 static bool isObviouslySimpleCharacter(char C) { 254 return C != '?' && C != '\\'; 255 } 256 257 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 258 /// advance over it, and return it. This is tricky in several cases. Here we 259 /// just handle the trivial case and fall-back to the non-inlined 260 /// getCharAndSizeSlow method to handle the hard case. 261 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 262 // If this is not a trigraph and not a UCN or escaped newline, return 263 // quickly. 264 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 265 266 unsigned Size = 0; 267 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 268 Ptr += Size; 269 return C; 270 } 271 272private: 273 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 274 /// and added to a given token, check to see if there are diagnostics that 275 /// need to be emitted or flags that need to be set on the token. If so, do 276 /// it. 277 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 278 // Normal case, we consumed exactly one token. Just return it. 279 if (Size == 1) 280 return Ptr+Size; 281 282 // Otherwise, re-lex the character with a current token, allowing 283 // diagnostics to be emitted and flags to be set. 284 Size = 0; 285 getCharAndSizeSlow(Ptr, Size, &Tok); 286 return Ptr+Size; 287 } 288 289 /// getCharAndSize - Peek a single 'character' from the specified buffer, 290 /// get its size, and return it. This is tricky in several cases. Here we 291 /// just handle the trivial case and fall-back to the non-inlined 292 /// getCharAndSizeSlow method to handle the hard case. 293 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 294 // If this is not a trigraph and not a UCN or escaped newline, return 295 // quickly. 296 if (isObviouslySimpleCharacter(Ptr[0])) { 297 Size = 1; 298 return *Ptr; 299 } 300 301 Size = 0; 302 return getCharAndSizeSlow(Ptr, Size); 303 } 304 305 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 306 /// method. 307 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 308 309 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 310 /// emit a warning. 311 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 312 const LangOptions &Features) { 313 // If this is not a trigraph and not a UCN or escaped newline, return 314 // quickly. 315 if (isObviouslySimpleCharacter(Ptr[0])) { 316 Size = 1; 317 return *Ptr; 318 } 319 320 Size = 0; 321 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 322 } 323 324 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 325 /// diagnostic. 326 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 327 const LangOptions &Features); 328 329 //===--------------------------------------------------------------------===// 330 // Other lexer functions. 331 332 // Helper functions to lex the remainder of a token of the specific type. 333 void LexIdentifier (Token &Result, const char *CurPtr); 334 void LexNumericConstant (Token &Result, const char *CurPtr); 335 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 336 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 337 void LexCharConstant (Token &Result, const char *CurPtr); 338 bool LexEndOfFile (Token &Result, const char *CurPtr); 339 340 bool SkipWhitespace (Token &Result, const char *CurPtr); 341 bool SkipBCPLComment (Token &Result, const char *CurPtr); 342 bool SkipBlockComment (Token &Result, const char *CurPtr); 343 bool SaveBCPLComment (Token &Result, const char *CurPtr); 344}; 345 346 347} // end namespace clang 348 349#endif 350