Lexer.h revision 306fda76b0818eb7c215094123f2e16729a44aed
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; 72public: 73 74 /// Lexer constructor - Create a new lexer object for the specified buffer 75 /// with the specified preprocessor managing the lexing process. This lexer 76 /// assumes that the associated file buffer and Preprocessor objects will 77 /// outlive it, so it doesn't take ownership of either of them. 78 Lexer(SourceLocation FileLoc, Preprocessor &PP, 79 const char *BufStart = 0, const char *BufEnd = 0); 80 81 /// Lexer constructor - Create a new raw lexer object. This object is only 82 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 83 /// range will outlive it, so it doesn't take ownership of it. 84 Lexer(SourceLocation FileLoc, const LangOptions &Features, 85 const char *BufStart, const char *BufEnd, 86 const llvm::MemoryBuffer *FromFile = 0); 87 88 /// getFeatures - Return the language features currently enabled. NOTE: this 89 /// lexer modifies features as a file is parsed! 90 const LangOptions &getFeatures() const { return Features; } 91 92 /// getFileLoc - Return the File Location for the file we are lexing out of. 93 /// The physical location encodes the location where the characters come from, 94 /// the virtual location encodes where we should *claim* the characters came 95 /// from. Currently this is only used by _Pragma handling. 96 SourceLocation getFileLoc() const { return FileLoc; } 97 98 /// Lex - Return the next token in the file. If this is the end of file, it 99 /// return the tok::eof token. Return true if an error occurred and 100 /// compilation should terminate, false if normal. This implicitly involves 101 /// the preprocessor. 102 void Lex(Token &Result) { 103 // Start a new token. 104 Result.startToken(); 105 106 // NOTE, any changes here should also change code after calls to 107 // Preprocessor::HandleDirective 108 if (IsAtStartOfLine) { 109 Result.setFlag(Token::StartOfLine); 110 IsAtStartOfLine = false; 111 } 112 113 // Get a token. Note that this may delete the current lexer if the end of 114 // file is reached. 115 LexTokenInternal(Result); 116 } 117 118 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 119 bool isPragmaLexer() const { return Is_PragmaLexer; } 120 121 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 122 /// the PreprocessorLexer interface. 123 void IndirectLex(Token &Result) { Lex(Result); } 124 125 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 126 /// associated preprocessor object. Return true if the 'next character to 127 /// read' pointer points and the end of the lexer buffer, false otherwise. 128 bool LexFromRawLexer(Token &Result) { 129 assert(LexingRawMode && "Not already in raw mode!"); 130 Lex(Result); 131 // Note that lexing to the end of the buffer doesn't implicitly delete the 132 // lexer when in raw mode. 133 return BufferPtr == BufferEnd; 134 } 135 136 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 137 /// every character in the file, including whitespace and comments. This 138 /// should only be used in raw mode, as the preprocessor is not prepared to 139 /// deal with the excess tokens. 140 bool isKeepWhitespaceMode() const { 141 return ExtendedTokenMode > 1; 142 } 143 144 /// SetKeepWhitespaceMode - This method lets clients enable or disable 145 /// whitespace retention mode. 146 void SetKeepWhitespaceMode(bool Val) { 147 assert((!Val || LexingRawMode) && 148 "Can only enable whitespace retention in raw mode"); 149 ExtendedTokenMode = Val ? 2 : 0; 150 } 151 152 /// inKeepCommentMode - Return true if the lexer should return comments as 153 /// tokens. 154 bool inKeepCommentMode() const { 155 return ExtendedTokenMode > 0; 156 } 157 158 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 159 /// to the specified mode. This is really only useful when lexing in raw 160 /// mode, because otherwise the lexer needs to manage this. 161 void SetCommentRetentionState(bool Mode) { 162 assert(!isKeepWhitespaceMode() && 163 "Can't play with comment retention state when retaining whitespace"); 164 ExtendedTokenMode = Mode ? 1 : 0; 165 } 166 167 168 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 169 /// uninterpreted string. This switches the lexer out of directive mode. 170 std::string ReadToEndOfLine(); 171 172 173 /// Diag - Forwarding function for diagnostics. This translate a source 174 /// position in the current buffer into a SourceLocation object for rendering. 175 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const; 176 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const; 177 178 /// getSourceLocation - Return a source location identifier for the specified 179 /// offset in the current file. 180 SourceLocation getSourceLocation(const char *Loc) const; 181 182 /// Stringify - Convert the specified string into a C string by escaping '\' 183 /// and " characters. This does not add surrounding ""'s to the string. 184 /// If Charify is true, this escapes the ' character instead of ". 185 static std::string Stringify(const std::string &Str, bool Charify = false); 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 static void Stringify(llvm::SmallVectorImpl<char> &Str); 190 191 /// MeasureTokenLength - Relex the token at the specified location and return 192 /// its length in bytes in the input file. If the token needs cleaning (e.g. 193 /// includes a trigraph or an escaped newline) then this count includes bytes 194 /// that are part of that. 195 static unsigned MeasureTokenLength(SourceLocation Loc, 196 const SourceManager &SM); 197 198 //===--------------------------------------------------------------------===// 199 // Internal implementation interfaces. 200private: 201 202 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 203 /// by Lex. 204 /// 205 void LexTokenInternal(Token &Result); 206 207 /// FormTokenWithChars - When we lex a token, we have identified a span 208 /// starting at BufferPtr, going to TokEnd that forms the token. This method 209 /// takes that range and assigns it to the token as its location and size. In 210 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 211 /// TokEnd. 212 void FormTokenWithChars(Token &Result, const char *TokEnd, 213 tok::TokenKind Kind) { 214 Result.setLocation(getSourceLocation(BufferPtr)); 215 Result.setLength(TokEnd-BufferPtr); 216 Result.setKind(Kind); 217 BufferPtr = TokEnd; 218 } 219 220 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 221 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 222 /// tokens in the buffer controlled by this lexer. 223 unsigned isNextPPTokenLParen(); 224 225 //===--------------------------------------------------------------------===// 226 // Lexer character reading interfaces. 227public: 228 229 // This lexer is built on two interfaces for reading characters, both of which 230 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 231 // when we know that we will be reading a character from the input buffer and 232 // that this character will be part of the result token. This occurs in (f.e.) 233 // string processing, because we know we need to read until we find the 234 // closing '"' character. 235 // 236 // The second interface is the combination of PeekCharAndSize with 237 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 238 // returning it and its size. If the lexer decides that this character is 239 // part of the current token, it calls ConsumeChar on it. This two stage 240 // approach allows us to emit diagnostics for characters (e.g. warnings about 241 // trigraphs), knowing that they only are emitted if the character is 242 // consumed. 243 244 /// isObviouslySimpleCharacter - Return true if the specified character is 245 /// obviously the same in translation phase 1 and translation phase 3. This 246 /// can return false for characters that end up being the same, but it will 247 /// never return true for something that needs to be mapped. 248 static bool isObviouslySimpleCharacter(char C) { 249 return C != '?' && C != '\\'; 250 } 251 252 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 253 /// advance over it, and return it. This is tricky in several cases. Here we 254 /// just handle the trivial case and fall-back to the non-inlined 255 /// getCharAndSizeSlow method to handle the hard case. 256 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 257 // If this is not a trigraph and not a UCN or escaped newline, return 258 // quickly. 259 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 260 261 unsigned Size = 0; 262 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 263 Ptr += Size; 264 return C; 265 } 266 267private: 268 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 269 /// and added to a given token, check to see if there are diagnostics that 270 /// need to be emitted or flags that need to be set on the token. If so, do 271 /// it. 272 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 273 // Normal case, we consumed exactly one token. Just return it. 274 if (Size == 1) 275 return Ptr+Size; 276 277 // Otherwise, re-lex the character with a current token, allowing 278 // diagnostics to be emitted and flags to be set. 279 Size = 0; 280 getCharAndSizeSlow(Ptr, Size, &Tok); 281 return Ptr+Size; 282 } 283 284 /// getCharAndSize - Peek a single 'character' from the specified buffer, 285 /// get its size, and return it. This is tricky in several cases. Here we 286 /// just handle the trivial case and fall-back to the non-inlined 287 /// getCharAndSizeSlow method to handle the hard case. 288 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 289 // If this is not a trigraph and not a UCN or escaped newline, return 290 // quickly. 291 if (isObviouslySimpleCharacter(Ptr[0])) { 292 Size = 1; 293 return *Ptr; 294 } 295 296 Size = 0; 297 return getCharAndSizeSlow(Ptr, Size); 298 } 299 300 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 301 /// method. 302 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 303 304 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 305 /// emit a warning. 306 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 307 const LangOptions &Features) { 308 // If this is not a trigraph and not a UCN or escaped newline, return 309 // quickly. 310 if (isObviouslySimpleCharacter(Ptr[0])) { 311 Size = 1; 312 return *Ptr; 313 } 314 315 Size = 0; 316 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 317 } 318 319 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 320 /// diagnostic. 321 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 322 const LangOptions &Features); 323 324 //===--------------------------------------------------------------------===// 325 // Other lexer functions. 326 327 // Helper functions to lex the remainder of a token of the specific type. 328 void LexIdentifier (Token &Result, const char *CurPtr); 329 void LexNumericConstant (Token &Result, const char *CurPtr); 330 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 331 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 332 void LexCharConstant (Token &Result, const char *CurPtr); 333 bool LexEndOfFile (Token &Result, const char *CurPtr); 334 335 bool SkipWhitespace (Token &Result, const char *CurPtr); 336 bool SkipBCPLComment (Token &Result, const char *CurPtr); 337 bool SkipBlockComment (Token &Result, const char *CurPtr); 338 bool SaveBCPLComment (Token &Result, const char *CurPtr); 339}; 340 341 342} // end namespace clang 343 344#endif 345