1 //===--- Lexer.h - C Language Family Lexer ----------------------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file defines the Lexer interface.
12 //===----------------------------------------------------------------------===//
14 #ifndef LLVM_CLANG_LEXER_H
15 #define LLVM_CLANG_LEXER_H
17 #include "clang/Basic/LangOptions.h"
18 #include "clang/Lex/PreprocessorLexer.h"
19 #include "llvm/ADT/SmallVector.h"
24 class DiagnosticsEngine;
27 class DiagnosticBuilder;
29 /// ConflictMarkerKind - Kinds of conflict marker which the lexer might be
31 enum ConflictMarkerKind {
32 /// Not within a conflict marker.
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.
37 /// A Perforce-style conflict marker, initiated by 4 ">"s,
38 /// separated by 4 "="s, and terminated by 4 "<"s.
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 virtual void anchor();
49 //===--------------------------------------------------------------------===//
50 // Constant configuration values for this lexer.
51 const char *BufferStart; // Start of the buffer.
52 const char *BufferEnd; // End of the buffer.
53 SourceLocation FileLoc; // Location for start of file.
54 LangOptions LangOpts; // LangOpts enabled by this language (cache).
55 bool Is_PragmaLexer; // True if lexer for _Pragma handling.
57 //===--------------------------------------------------------------------===//
58 // Context-specific lexing flags set by the preprocessor.
61 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace
62 /// and return them as tokens. This is used for -C and -CC modes, and
63 /// whitespace preservation can be useful for some clients that want to lex
64 /// the file in raw mode and get every character from the file.
66 /// When this is set to 2 it returns comments and whitespace. When set to 1
67 /// it returns comments, when it is set to 0 it returns normal tokens only.
68 unsigned char ExtendedTokenMode;
70 //===--------------------------------------------------------------------===//
71 // Context that changes as the file is lexed.
72 // NOTE: any state that mutates when in raw mode must have save/restore code
73 // in Lexer::isNextPPTokenLParen.
75 // BufferPtr - Current pointer into the buffer. This is the next character
77 const char *BufferPtr;
79 // IsAtStartOfLine - True if the next lexed token should get the "start of
80 // line" flag set on it.
83 // CurrentConflictMarkerState - The kind of conflict marker we are handling.
84 ConflictMarkerKind CurrentConflictMarkerState;
86 Lexer(const Lexer &) LLVM_DELETED_FUNCTION;
87 void operator=(const Lexer &) LLVM_DELETED_FUNCTION;
88 friend class Preprocessor;
90 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd);
93 /// Lexer constructor - Create a new lexer object for the specified buffer
94 /// with the specified preprocessor managing the lexing process. This lexer
95 /// assumes that the associated file buffer and Preprocessor objects will
96 /// outlive it, so it doesn't take ownership of either of them.
97 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, Preprocessor &PP);
99 /// Lexer constructor - Create a new raw lexer object. This object is only
100 /// suitable for calls to 'LexFromRawLexer'. This lexer assumes that the
101 /// text range will outlive it, so it doesn't take ownership of it.
102 Lexer(SourceLocation FileLoc, const LangOptions &LangOpts,
103 const char *BufStart, const char *BufPtr, const char *BufEnd);
105 /// Lexer constructor - Create a new raw lexer object. This object is only
106 /// suitable for calls to 'LexFromRawLexer'. This lexer assumes that the
107 /// text range will outlive it, so it doesn't take ownership of it.
108 Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer,
109 const SourceManager &SM, const LangOptions &LangOpts);
111 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for
112 /// _Pragma expansion. This has a variety of magic semantics that this method
113 /// sets up. It returns a new'd Lexer that must be delete'd when done.
114 static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc,
115 SourceLocation ExpansionLocStart,
116 SourceLocation ExpansionLocEnd,
117 unsigned TokLen, Preprocessor &PP);
120 /// getLangOpts - Return the language features currently enabled.
121 /// NOTE: this lexer modifies features as a file is parsed!
122 const LangOptions &getLangOpts() const { return LangOpts; }
124 /// getFileLoc - Return the File Location for the file we are lexing out of.
125 /// The physical location encodes the location where the characters come from,
126 /// the virtual location encodes where we should *claim* the characters came
127 /// from. Currently this is only used by _Pragma handling.
128 SourceLocation getFileLoc() const { return FileLoc; }
130 /// Lex - Return the next token in the file. If this is the end of file, it
131 /// return the tok::eof token. This implicitly involves the preprocessor.
132 void Lex(Token &Result) {
133 // Start a new token.
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;
143 // Get a token. Note that this may delete the current lexer if the end of
145 LexTokenInternal(Result);
148 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma.
149 bool isPragmaLexer() const { return Is_PragmaLexer; }
151 /// IndirectLex - An indirect call to 'Lex' that can be invoked via
152 /// the PreprocessorLexer interface.
153 void IndirectLex(Token &Result) { Lex(Result); }
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!");
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;
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;
174 /// SetKeepWhitespaceMode - This method lets clients enable or disable
175 /// whitespace retention mode.
176 void SetKeepWhitespaceMode(bool Val) {
177 assert((!Val || LexingRawMode || LangOpts.TraditionalCPP) &&
178 "Can only retain whitespace in raw mode or -traditional-cpp");
179 ExtendedTokenMode = Val ? 2 : 0;
182 /// inKeepCommentMode - Return true if the lexer should return comments as
184 bool inKeepCommentMode() const {
185 return ExtendedTokenMode > 0;
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;
197 /// Sets the extended token mode back to its initial value, according to the
198 /// language options and preprocessor. This controls whether the lexer
199 /// produces comment and whitespace tokens.
201 /// This requires the lexer to have an associated preprocessor. A standalone
202 /// lexer has nothing to reset to.
203 void resetExtendedTokenMode();
205 const char *getBufferStart() const { return BufferStart; }
207 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an
208 /// uninterpreted string. This switches the lexer out of directive mode.
209 void ReadToEndOfLine(SmallVectorImpl<char> *Result = 0);
212 /// Diag - Forwarding function for diagnostics. This translate a source
213 /// position in the current buffer into a SourceLocation object for rendering.
214 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const;
216 /// getSourceLocation - Return a source location identifier for the specified
217 /// offset in the current file.
218 SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const;
220 /// getSourceLocation - Return a source location for the next character in
221 /// the current file.
222 SourceLocation getSourceLocation() { return getSourceLocation(BufferPtr); }
224 /// \brief Return the current location in the buffer.
225 const char *getBufferLocation() const { return BufferPtr; }
227 /// Stringify - Convert the specified string into a C string by escaping '\'
228 /// and " characters. This does not add surrounding ""'s to the string.
229 /// If Charify is true, this escapes the ' character instead of ".
230 static std::string Stringify(const std::string &Str, bool Charify = false);
232 /// Stringify - Convert the specified string into a C string by escaping '\'
233 /// and " characters. This does not add surrounding ""'s to the string.
234 static void Stringify(SmallVectorImpl<char> &Str);
237 /// getSpelling - This method is used to get the spelling of a token into a
238 /// preallocated buffer, instead of as an std::string. The caller is required
239 /// to allocate enough space for the token, which is guaranteed to be at least
240 /// Tok.getLength() bytes long. The length of the actual result is returned.
242 /// Note that this method may do two possible things: it may either fill in
243 /// the buffer specified with characters, or it may *change the input pointer*
244 /// to point to a constant buffer with the data already in it (avoiding a
245 /// copy). The caller is not allowed to modify the returned buffer pointer
246 /// if an internal buffer is returned.
247 static unsigned getSpelling(const Token &Tok, const char *&Buffer,
248 const SourceManager &SourceMgr,
249 const LangOptions &LangOpts,
252 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a
253 /// token is the characters used to represent the token in the source file
254 /// after trigraph expansion and escaped-newline folding. In particular, this
255 /// wants to get the true, uncanonicalized, spelling of things like digraphs
257 static std::string getSpelling(const Token &Tok,
258 const SourceManager &SourceMgr,
259 const LangOptions &LangOpts,
262 /// getSpelling - This method is used to get the spelling of the
263 /// token at the given source location. If, as is usually true, it
264 /// is not necessary to copy any data, then the returned string may
265 /// not point into the provided buffer.
267 /// This method lexes at the expansion depth of the given
268 /// location and does not jump to the expansion or spelling
270 static StringRef getSpelling(SourceLocation loc,
271 SmallVectorImpl<char> &buffer,
272 const SourceManager &SourceMgr,
273 const LangOptions &LangOpts,
276 /// MeasureTokenLength - Relex the token at the specified location and return
277 /// its length in bytes in the input file. If the token needs cleaning (e.g.
278 /// includes a trigraph or an escaped newline) then this count includes bytes
279 /// that are part of that.
280 static unsigned MeasureTokenLength(SourceLocation Loc,
281 const SourceManager &SM,
282 const LangOptions &LangOpts);
284 /// \brief Relex the token at the specified location.
285 /// \returns true if there was a failure, false on success.
286 static bool getRawToken(SourceLocation Loc, Token &Result,
287 const SourceManager &SM,
288 const LangOptions &LangOpts);
290 /// \brief Given a location any where in a source buffer, find the location
291 /// that corresponds to the beginning of the token in which the original
292 /// source location lands.
293 static SourceLocation GetBeginningOfToken(SourceLocation Loc,
294 const SourceManager &SM,
295 const LangOptions &LangOpts);
297 /// AdvanceToTokenCharacter - If the current SourceLocation specifies a
298 /// location at the start of a token, return a new location that specifies a
299 /// character within the token. This handles trigraphs and escaped newlines.
300 static SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
302 const SourceManager &SM,
303 const LangOptions &LangOpts);
305 /// \brief Computes the source location just past the end of the
306 /// token at this source location.
308 /// This routine can be used to produce a source location that
309 /// points just past the end of the token referenced by \p Loc, and
310 /// is generally used when a diagnostic needs to point just after a
311 /// token where it expected something different that it received. If
312 /// the returned source location would not be meaningful (e.g., if
313 /// it points into a macro), this routine returns an invalid
316 /// \param Offset an offset from the end of the token, where the source
317 /// location should refer to. The default offset (0) produces a source
318 /// location pointing just past the end of the token; an offset of 1 produces
319 /// a source location pointing to the last character in the token, etc.
320 static SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset,
321 const SourceManager &SM,
322 const LangOptions &LangOpts);
324 /// \brief Returns true if the given MacroID location points at the first
325 /// token of the macro expansion.
327 /// \param MacroBegin If non-null and function returns true, it is set to
328 /// begin location of the macro.
329 static bool isAtStartOfMacroExpansion(SourceLocation loc,
330 const SourceManager &SM,
331 const LangOptions &LangOpts,
332 SourceLocation *MacroBegin = 0);
334 /// \brief Returns true if the given MacroID location points at the last
335 /// token of the macro expansion.
337 /// \param MacroEnd If non-null and function returns true, it is set to
338 /// end location of the macro.
339 static bool isAtEndOfMacroExpansion(SourceLocation loc,
340 const SourceManager &SM,
341 const LangOptions &LangOpts,
342 SourceLocation *MacroEnd = 0);
344 /// \brief Accepts a range and returns a character range with file locations.
346 /// Returns a null range if a part of the range resides inside a macro
347 /// expansion or the range does not reside on the same FileID.
349 /// This function is trying to deal with macros and return a range based on
350 /// file locations. The cases where it can successfully handle macros are:
352 /// -begin or end range lies at the start or end of a macro expansion, in
353 /// which case the location will be set to the expansion point, e.g:
356 /// If you have a range [a, 2] (where 2 came from the macro), the function
357 /// will return a range for "a M"
358 /// if you have range [a, 1], the function will fail because the range
359 /// overlaps with only a part of the macro
361 /// -The macro is a function macro and the range can be mapped to the macro
366 /// if you have range [b, 2], the function will return the file range "b M"
367 /// inside the macro arguments.
368 /// if you have range [a, 2], the function will return the file range
369 /// "FM(a b M)" since the range includes all of the macro expansion.
370 static CharSourceRange makeFileCharRange(CharSourceRange Range,
371 const SourceManager &SM,
372 const LangOptions &LangOpts);
374 /// \brief Returns a string for the source that the range encompasses.
375 static StringRef getSourceText(CharSourceRange Range,
376 const SourceManager &SM,
377 const LangOptions &LangOpts,
380 /// \brief Retrieve the name of the immediate macro expansion.
382 /// This routine starts from a source location, and finds the name of the macro
383 /// responsible for its immediate expansion. It looks through any intervening
384 /// macro argument expansions to compute this. It returns a StringRef which
385 /// refers to the SourceManager-owned buffer of the source where that macro
386 /// name is spelled. Thus, the result shouldn't out-live that SourceManager.
387 static StringRef getImmediateMacroName(SourceLocation Loc,
388 const SourceManager &SM,
389 const LangOptions &LangOpts);
391 /// \brief Compute the preamble of the given file.
393 /// The preamble of a file contains the initial comments, include directives,
394 /// and other preprocessor directives that occur before the code in this
395 /// particular file actually begins. The preamble of the main source file is
396 /// a potential prefix header.
398 /// \param Buffer The memory buffer containing the file's contents.
400 /// \param MaxLines If non-zero, restrict the length of the preamble
401 /// to fewer than this number of lines.
403 /// \returns The offset into the file where the preamble ends and the rest
404 /// of the file begins along with a boolean value indicating whether
405 /// the preamble ends at the beginning of a new line.
406 static std::pair<unsigned, bool>
407 ComputePreamble(const llvm::MemoryBuffer *Buffer, const LangOptions &LangOpts,
408 unsigned MaxLines = 0);
410 /// \brief Checks that the given token is the first token that occurs after
411 /// the given location (this excludes comments and whitespace). Returns the
412 /// location immediately after the specified token. If the token is not found
413 /// or the location is inside a macro, the returned source location will be
415 static SourceLocation findLocationAfterToken(SourceLocation loc,
416 tok::TokenKind TKind,
417 const SourceManager &SM,
418 const LangOptions &LangOpts,
419 bool SkipTrailingWhitespaceAndNewLine);
421 /// \brief Returns true if the given character could appear in an identifier.
422 static bool isIdentifierBodyChar(char c, const LangOptions &LangOpts);
424 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever
426 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size,
427 const LangOptions &LangOpts) {
428 // If this is not a trigraph and not a UCN or escaped newline, return
430 if (isObviouslySimpleCharacter(Ptr[0])) {
436 return getCharAndSizeSlowNoWarn(Ptr, Size, LangOpts);
439 //===--------------------------------------------------------------------===//
440 // Internal implementation interfaces.
443 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called
446 void LexTokenInternal(Token &Result);
448 /// Given that a token begins with the Unicode character \p C, figure out
449 /// what kind of token it is and dispatch to the appropriate lexing helper
451 void LexUnicode(Token &Result, uint32_t C, const char *CurPtr);
453 /// FormTokenWithChars - When we lex a token, we have identified a span
454 /// starting at BufferPtr, going to TokEnd that forms the token. This method
455 /// takes that range and assigns it to the token as its location and size. In
456 /// addition, since tokens cannot overlap, this also updates BufferPtr to be
458 void FormTokenWithChars(Token &Result, const char *TokEnd,
459 tok::TokenKind Kind) {
460 unsigned TokLen = TokEnd-BufferPtr;
461 Result.setLength(TokLen);
462 Result.setLocation(getSourceLocation(BufferPtr, TokLen));
463 Result.setKind(Kind);
467 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a
468 /// tok::l_paren token, 0 if it is something else and 2 if there are no more
469 /// tokens in the buffer controlled by this lexer.
470 unsigned isNextPPTokenLParen();
472 //===--------------------------------------------------------------------===//
473 // Lexer character reading interfaces.
475 // This lexer is built on two interfaces for reading characters, both of which
476 // automatically provide phase 1/2 translation. getAndAdvanceChar is used
477 // when we know that we will be reading a character from the input buffer and
478 // that this character will be part of the result token. This occurs in (f.e.)
479 // string processing, because we know we need to read until we find the
480 // closing '"' character.
482 // The second interface is the combination of getCharAndSize with
483 // ConsumeChar. getCharAndSize reads a phase 1/2 translated character,
484 // returning it and its size. If the lexer decides that this character is
485 // part of the current token, it calls ConsumeChar on it. This two stage
486 // approach allows us to emit diagnostics for characters (e.g. warnings about
487 // trigraphs), knowing that they only are emitted if the character is
490 /// isObviouslySimpleCharacter - Return true if the specified character is
491 /// obviously the same in translation phase 1 and translation phase 3. This
492 /// can return false for characters that end up being the same, but it will
493 /// never return true for something that needs to be mapped.
494 static bool isObviouslySimpleCharacter(char C) {
495 return C != '?' && C != '\\';
498 /// getAndAdvanceChar - Read a single 'character' from the specified buffer,
499 /// advance over it, and return it. This is tricky in several cases. Here we
500 /// just handle the trivial case and fall-back to the non-inlined
501 /// getCharAndSizeSlow method to handle the hard case.
502 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) {
503 // If this is not a trigraph and not a UCN or escaped newline, return
505 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++;
508 char C = getCharAndSizeSlow(Ptr, Size, &Tok);
513 /// ConsumeChar - When a character (identified by getCharAndSize) is consumed
514 /// and added to a given token, check to see if there are diagnostics that
515 /// need to be emitted or flags that need to be set on the token. If so, do
517 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) {
518 // Normal case, we consumed exactly one token. Just return it.
522 // Otherwise, re-lex the character with a current token, allowing
523 // diagnostics to be emitted and flags to be set.
525 getCharAndSizeSlow(Ptr, Size, &Tok);
529 /// getCharAndSize - Peek a single 'character' from the specified buffer,
530 /// get its size, and return it. This is tricky in several cases. Here we
531 /// just handle the trivial case and fall-back to the non-inlined
532 /// getCharAndSizeSlow method to handle the hard case.
533 inline char getCharAndSize(const char *Ptr, unsigned &Size) {
534 // If this is not a trigraph and not a UCN or escaped newline, return
536 if (isObviouslySimpleCharacter(Ptr[0])) {
542 return getCharAndSizeSlow(Ptr, Size);
545 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize
547 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0);
549 /// getEscapedNewLineSize - Return the size of the specified escaped newline,
550 /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry
551 /// to this function.
552 static unsigned getEscapedNewLineSize(const char *P);
554 /// SkipEscapedNewLines - If P points to an escaped newline (or a series of
555 /// them), skip over them and return the first non-escaped-newline found,
556 /// otherwise return P.
557 static const char *SkipEscapedNewLines(const char *P);
559 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a
561 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size,
562 const LangOptions &LangOpts);
564 //===--------------------------------------------------------------------===//
565 // Other lexer functions.
567 void SkipBytes(unsigned Bytes, bool StartOfLine);
569 const char *LexUDSuffix(Token &Result, const char *CurPtr);
571 // Helper functions to lex the remainder of a token of the specific type.
572 void LexIdentifier (Token &Result, const char *CurPtr);
573 void LexNumericConstant (Token &Result, const char *CurPtr);
574 void LexStringLiteral (Token &Result, const char *CurPtr,
575 tok::TokenKind Kind);
576 void LexRawStringLiteral (Token &Result, const char *CurPtr,
577 tok::TokenKind Kind);
578 void LexAngledStringLiteral(Token &Result, const char *CurPtr);
579 void LexCharConstant (Token &Result, const char *CurPtr,
580 tok::TokenKind Kind);
581 bool LexEndOfFile (Token &Result, const char *CurPtr);
583 bool SkipWhitespace (Token &Result, const char *CurPtr);
584 bool SkipLineComment (Token &Result, const char *CurPtr);
585 bool SkipBlockComment (Token &Result, const char *CurPtr);
586 bool SaveLineComment (Token &Result, const char *CurPtr);
588 bool IsStartOfConflictMarker(const char *CurPtr);
589 bool HandleEndOfConflictMarker(const char *CurPtr);
591 bool isCodeCompletionPoint(const char *CurPtr) const;
592 void cutOffLexing() { BufferPtr = BufferEnd; }
594 bool isHexaLiteral(const char *Start, const LangOptions &LangOpts);
597 /// Read a universal character name.
599 /// \param CurPtr The position in the source buffer after the initial '\'.
600 /// If the UCN is syntactically well-formed (but not necessarily
601 /// valid), this parameter will be updated to point to the
602 /// character after the UCN.
603 /// \param SlashLoc The position in the source buffer of the '\'.
604 /// \param Tok The token being formed. Pass \c NULL to suppress diagnostics
605 /// and handle token formation in the caller.
607 /// \return The Unicode codepoint specified by the UCN, or 0 if the UCN is
609 uint32_t tryReadUCN(const char *&CurPtr, const char *SlashLoc, Token *Tok);
613 } // end namespace clang