Home | History | Annotate | Download | only in Lex
      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_LEX_LEXER_H
     15 #define LLVM_CLANG_LEX_LEXER_H
     16 
     17 #include "clang/Basic/LangOptions.h"
     18 #include "clang/Lex/PreprocessorLexer.h"
     19 #include "llvm/ADT/SmallVector.h"
     20 #include <cassert>
     21 #include <string>
     22 
     23 namespace clang {
     24 class DiagnosticsEngine;
     25 class SourceManager;
     26 class Preprocessor;
     27 class DiagnosticBuilder;
     28 
     29 /// ConflictMarkerKind - Kinds of conflict marker which the lexer might be
     30 /// recovering from.
     31 enum ConflictMarkerKind {
     32   /// Not within a conflict marker.
     33   CMK_None,
     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.
     36   CMK_Normal,
     37   /// A Perforce-style conflict marker, initiated by 4 ">"s,
     38   /// separated by 4 "="s, and terminated by 4 "<"s.
     39   CMK_Perforce
     40 };
     41 
     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   void anchor() override;
     48 
     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.
     56 
     57   //===--------------------------------------------------------------------===//
     58   // Context-specific lexing flags set by the preprocessor.
     59   //
     60 
     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.
     65   ///
     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;
     69 
     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.
     74 
     75   // BufferPtr - Current pointer into the buffer.  This is the next character
     76   // to be lexed.
     77   const char *BufferPtr;
     78 
     79   // IsAtStartOfLine - True if the next lexed token should get the "start of
     80   // line" flag set on it.
     81   bool IsAtStartOfLine;
     82 
     83   bool IsAtPhysicalStartOfLine;
     84 
     85   bool HasLeadingSpace;
     86 
     87   bool HasLeadingEmptyMacro;
     88 
     89   // CurrentConflictMarkerState - The kind of conflict marker we are handling.
     90   ConflictMarkerKind CurrentConflictMarkerState;
     91 
     92   Lexer(const Lexer &) = delete;
     93   void operator=(const Lexer &) = delete;
     94   friend class Preprocessor;
     95 
     96   void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd);
     97 public:
     98 
     99   /// Lexer constructor - Create a new lexer object for the specified buffer
    100   /// with the specified preprocessor managing the lexing process.  This lexer
    101   /// assumes that the associated file buffer and Preprocessor objects will
    102   /// outlive it, so it doesn't take ownership of either of them.
    103   Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer, Preprocessor &PP);
    104 
    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(SourceLocation FileLoc, const LangOptions &LangOpts,
    109         const char *BufStart, const char *BufPtr, const char *BufEnd);
    110 
    111   /// Lexer constructor - Create a new raw lexer object.  This object is only
    112   /// suitable for calls to 'LexFromRawLexer'.  This lexer assumes that the
    113   /// text range will outlive it, so it doesn't take ownership of it.
    114   Lexer(FileID FID, const llvm::MemoryBuffer *InputBuffer,
    115         const SourceManager &SM, const LangOptions &LangOpts);
    116 
    117   /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for
    118   /// _Pragma expansion.  This has a variety of magic semantics that this method
    119   /// sets up.  It returns a new'd Lexer that must be delete'd when done.
    120   static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc,
    121                                    SourceLocation ExpansionLocStart,
    122                                    SourceLocation ExpansionLocEnd,
    123                                    unsigned TokLen, Preprocessor &PP);
    124 
    125 
    126   /// getLangOpts - Return the language features currently enabled.
    127   /// NOTE: this lexer modifies features as a file is parsed!
    128   const LangOptions &getLangOpts() const { return LangOpts; }
    129 
    130   /// getFileLoc - Return the File Location for the file we are lexing out of.
    131   /// The physical location encodes the location where the characters come from,
    132   /// the virtual location encodes where we should *claim* the characters came
    133   /// from.  Currently this is only used by _Pragma handling.
    134   SourceLocation getFileLoc() const { return FileLoc; }
    135 
    136 private:
    137   /// Lex - Return the next token in the file.  If this is the end of file, it
    138   /// return the tok::eof token.  This implicitly involves the preprocessor.
    139   bool Lex(Token &Result);
    140 
    141 public:
    142   /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma.
    143   bool isPragmaLexer() const { return Is_PragmaLexer; }
    144 
    145 private:
    146   /// IndirectLex - An indirect call to 'Lex' that can be invoked via
    147   ///  the PreprocessorLexer interface.
    148   void IndirectLex(Token &Result) override { Lex(Result); }
    149 
    150 public:
    151   /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no
    152   /// associated preprocessor object.  Return true if the 'next character to
    153   /// read' pointer points at the end of the lexer buffer, false otherwise.
    154   bool LexFromRawLexer(Token &Result) {
    155     assert(LexingRawMode && "Not already in raw mode!");
    156     Lex(Result);
    157     // Note that lexing to the end of the buffer doesn't implicitly delete the
    158     // lexer when in raw mode.
    159     return BufferPtr == BufferEnd;
    160   }
    161 
    162   /// isKeepWhitespaceMode - Return true if the lexer should return tokens for
    163   /// every character in the file, including whitespace and comments.  This
    164   /// should only be used in raw mode, as the preprocessor is not prepared to
    165   /// deal with the excess tokens.
    166   bool isKeepWhitespaceMode() const {
    167     return ExtendedTokenMode > 1;
    168   }
    169 
    170   /// SetKeepWhitespaceMode - This method lets clients enable or disable
    171   /// whitespace retention mode.
    172   void SetKeepWhitespaceMode(bool Val) {
    173     assert((!Val || LexingRawMode || LangOpts.TraditionalCPP) &&
    174            "Can only retain whitespace in raw mode or -traditional-cpp");
    175     ExtendedTokenMode = Val ? 2 : 0;
    176   }
    177 
    178   /// inKeepCommentMode - Return true if the lexer should return comments as
    179   /// tokens.
    180   bool inKeepCommentMode() const {
    181     return ExtendedTokenMode > 0;
    182   }
    183 
    184   /// SetCommentRetentionMode - Change the comment retention mode of the lexer
    185   /// to the specified mode.  This is really only useful when lexing in raw
    186   /// mode, because otherwise the lexer needs to manage this.
    187   void SetCommentRetentionState(bool Mode) {
    188     assert(!isKeepWhitespaceMode() &&
    189            "Can't play with comment retention state when retaining whitespace");
    190     ExtendedTokenMode = Mode ? 1 : 0;
    191   }
    192 
    193   /// Sets the extended token mode back to its initial value, according to the
    194   /// language options and preprocessor. This controls whether the lexer
    195   /// produces comment and whitespace tokens.
    196   ///
    197   /// This requires the lexer to have an associated preprocessor. A standalone
    198   /// lexer has nothing to reset to.
    199   void resetExtendedTokenMode();
    200 
    201   /// Gets source code buffer.
    202   StringRef getBuffer() const {
    203     return StringRef(BufferStart, BufferEnd - BufferStart);
    204   }
    205 
    206   /// ReadToEndOfLine - Read the rest of the current preprocessor line as an
    207   /// uninterpreted string.  This switches the lexer out of directive mode.
    208   void ReadToEndOfLine(SmallVectorImpl<char> *Result = nullptr);
    209 
    210 
    211   /// Diag - Forwarding function for diagnostics.  This translate a source
    212   /// position in the current buffer into a SourceLocation object for rendering.
    213   DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const;
    214 
    215   /// getSourceLocation - Return a source location identifier for the specified
    216   /// offset in the current file.
    217   SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const;
    218 
    219   /// getSourceLocation - Return a source location for the next character in
    220   /// the current file.
    221   SourceLocation getSourceLocation() override {
    222     return getSourceLocation(BufferPtr);
    223   }
    224 
    225   /// \brief Return the current location in the buffer.
    226   const char *getBufferLocation() const { return BufferPtr; }
    227 
    228   /// Stringify - Convert the specified string into a C string by escaping '\'
    229   /// and " characters.  This does not add surrounding ""'s to the string.
    230   /// If Charify is true, this escapes the ' character instead of ".
    231   static std::string Stringify(StringRef Str, bool Charify = false);
    232 
    233   /// Stringify - Convert the specified string into a C string by escaping '\'
    234   /// and " characters.  This does not add surrounding ""'s to the string.
    235   static void Stringify(SmallVectorImpl<char> &Str);
    236 
    237 
    238   /// getSpelling - This method is used to get the spelling of a token into a
    239   /// preallocated buffer, instead of as an std::string.  The caller is required
    240   /// to allocate enough space for the token, which is guaranteed to be at least
    241   /// Tok.getLength() bytes long.  The length of the actual result is returned.
    242   ///
    243   /// Note that this method may do two possible things: it may either fill in
    244   /// the buffer specified with characters, or it may *change the input pointer*
    245   /// to point to a constant buffer with the data already in it (avoiding a
    246   /// copy).  The caller is not allowed to modify the returned buffer pointer
    247   /// if an internal buffer is returned.
    248   static unsigned getSpelling(const Token &Tok, const char *&Buffer,
    249                               const SourceManager &SourceMgr,
    250                               const LangOptions &LangOpts,
    251                               bool *Invalid = nullptr);
    252 
    253   /// getSpelling() - Return the 'spelling' of the Tok token.  The spelling of a
    254   /// token is the characters used to represent the token in the source file
    255   /// after trigraph expansion and escaped-newline folding.  In particular, this
    256   /// wants to get the true, uncanonicalized, spelling of things like digraphs
    257   /// UCNs, etc.
    258   static std::string getSpelling(const Token &Tok,
    259                                  const SourceManager &SourceMgr,
    260                                  const LangOptions &LangOpts,
    261                                  bool *Invalid = nullptr);
    262 
    263   /// getSpelling - This method is used to get the spelling of the
    264   /// token at the given source location.  If, as is usually true, it
    265   /// is not necessary to copy any data, then the returned string may
    266   /// not point into the provided buffer.
    267   ///
    268   /// This method lexes at the expansion depth of the given
    269   /// location and does not jump to the expansion or spelling
    270   /// location.
    271   static StringRef getSpelling(SourceLocation loc,
    272                                SmallVectorImpl<char> &buffer,
    273                                const SourceManager &SourceMgr,
    274                                const LangOptions &LangOpts,
    275                                bool *invalid = nullptr);
    276 
    277   /// MeasureTokenLength - Relex the token at the specified location and return
    278   /// its length in bytes in the input file.  If the token needs cleaning (e.g.
    279   /// includes a trigraph or an escaped newline) then this count includes bytes
    280   /// that are part of that.
    281   static unsigned MeasureTokenLength(SourceLocation Loc,
    282                                      const SourceManager &SM,
    283                                      const LangOptions &LangOpts);
    284 
    285   /// \brief Relex the token at the specified location.
    286   /// \returns true if there was a failure, false on success.
    287   static bool getRawToken(SourceLocation Loc, Token &Result,
    288                           const SourceManager &SM,
    289                           const LangOptions &LangOpts,
    290                           bool IgnoreWhiteSpace = false);
    291 
    292   /// \brief Given a location any where in a source buffer, find the location
    293   /// that corresponds to the beginning of the token in which the original
    294   /// source location lands.
    295   static SourceLocation GetBeginningOfToken(SourceLocation Loc,
    296                                             const SourceManager &SM,
    297                                             const LangOptions &LangOpts);
    298 
    299   /// AdvanceToTokenCharacter - If the current SourceLocation specifies a
    300   /// location at the start of a token, return a new location that specifies a
    301   /// character within the token.  This handles trigraphs and escaped newlines.
    302   static SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
    303                                                 unsigned Character,
    304                                                 const SourceManager &SM,
    305                                                 const LangOptions &LangOpts);
    306 
    307   /// \brief Computes the source location just past the end of the
    308   /// token at this source location.
    309   ///
    310   /// This routine can be used to produce a source location that
    311   /// points just past the end of the token referenced by \p Loc, and
    312   /// is generally used when a diagnostic needs to point just after a
    313   /// token where it expected something different that it received. If
    314   /// the returned source location would not be meaningful (e.g., if
    315   /// it points into a macro), this routine returns an invalid
    316   /// source location.
    317   ///
    318   /// \param Offset an offset from the end of the token, where the source
    319   /// location should refer to. The default offset (0) produces a source
    320   /// location pointing just past the end of the token; an offset of 1 produces
    321   /// a source location pointing to the last character in the token, etc.
    322   static SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset,
    323                                             const SourceManager &SM,
    324                                             const LangOptions &LangOpts);
    325 
    326   /// \brief Given a token range, produce a corresponding CharSourceRange that
    327   /// is not a token range. This allows the source range to be used by
    328   /// components that don't have access to the lexer and thus can't find the
    329   /// end of the range for themselves.
    330   static CharSourceRange getAsCharRange(SourceRange Range,
    331                                         const SourceManager &SM,
    332                                         const LangOptions &LangOpts) {
    333     SourceLocation End = getLocForEndOfToken(Range.getEnd(), 0, SM, LangOpts);
    334     return End.isInvalid() ? CharSourceRange()
    335                            : CharSourceRange::getCharRange(
    336                                  Range.getBegin(), End.getLocWithOffset(-1));
    337   }
    338   static CharSourceRange getAsCharRange(CharSourceRange Range,
    339                                         const SourceManager &SM,
    340                                         const LangOptions &LangOpts) {
    341     return Range.isTokenRange()
    342                ? getAsCharRange(Range.getAsRange(), SM, LangOpts)
    343                : Range;
    344   }
    345 
    346   /// \brief Returns true if the given MacroID location points at the first
    347   /// token of the macro expansion.
    348   ///
    349   /// \param MacroBegin If non-null and function returns true, it is set to
    350   /// begin location of the macro.
    351   static bool isAtStartOfMacroExpansion(SourceLocation loc,
    352                                         const SourceManager &SM,
    353                                         const LangOptions &LangOpts,
    354                                         SourceLocation *MacroBegin = nullptr);
    355 
    356   /// \brief Returns true if the given MacroID location points at the last
    357   /// token of the macro expansion.
    358   ///
    359   /// \param MacroEnd If non-null and function returns true, it is set to
    360   /// end location of the macro.
    361   static bool isAtEndOfMacroExpansion(SourceLocation loc,
    362                                       const SourceManager &SM,
    363                                       const LangOptions &LangOpts,
    364                                       SourceLocation *MacroEnd = nullptr);
    365 
    366   /// \brief Accepts a range and returns a character range with file locations.
    367   ///
    368   /// Returns a null range if a part of the range resides inside a macro
    369   /// expansion or the range does not reside on the same FileID.
    370   ///
    371   /// This function is trying to deal with macros and return a range based on
    372   /// file locations. The cases where it can successfully handle macros are:
    373   ///
    374   /// -begin or end range lies at the start or end of a macro expansion, in
    375   ///  which case the location will be set to the expansion point, e.g:
    376   ///    \#define M 1 2
    377   ///    a M
    378   /// If you have a range [a, 2] (where 2 came from the macro), the function
    379   /// will return a range for "a M"
    380   /// if you have range [a, 1], the function will fail because the range
    381   /// overlaps with only a part of the macro
    382   ///
    383   /// -The macro is a function macro and the range can be mapped to the macro
    384   ///  arguments, e.g:
    385   ///    \#define M 1 2
    386   ///    \#define FM(x) x
    387   ///    FM(a b M)
    388   /// if you have range [b, 2], the function will return the file range "b M"
    389   /// inside the macro arguments.
    390   /// if you have range [a, 2], the function will return the file range
    391   /// "FM(a b M)" since the range includes all of the macro expansion.
    392   static CharSourceRange makeFileCharRange(CharSourceRange Range,
    393                                            const SourceManager &SM,
    394                                            const LangOptions &LangOpts);
    395 
    396   /// \brief Returns a string for the source that the range encompasses.
    397   static StringRef getSourceText(CharSourceRange Range,
    398                                  const SourceManager &SM,
    399                                  const LangOptions &LangOpts,
    400                                  bool *Invalid = nullptr);
    401 
    402   /// \brief Retrieve the name of the immediate macro expansion.
    403   ///
    404   /// This routine starts from a source location, and finds the name of the macro
    405   /// responsible for its immediate expansion. It looks through any intervening
    406   /// macro argument expansions to compute this. It returns a StringRef which
    407   /// refers to the SourceManager-owned buffer of the source where that macro
    408   /// name is spelled. Thus, the result shouldn't out-live that SourceManager.
    409   static StringRef getImmediateMacroName(SourceLocation Loc,
    410                                          const SourceManager &SM,
    411                                          const LangOptions &LangOpts);
    412 
    413   /// \brief Retrieve the name of the immediate macro expansion.
    414   ///
    415   /// This routine starts from a source location, and finds the name of the
    416   /// macro responsible for its immediate expansion. It looks through any
    417   /// intervening macro argument expansions to compute this. It returns a
    418   /// StringRef which refers to the SourceManager-owned buffer of the source
    419   /// where that macro name is spelled. Thus, the result shouldn't out-live
    420   /// that SourceManager.
    421   ///
    422   /// This differs from Lexer::getImmediateMacroName in that any macro argument
    423   /// location will result in the topmost function macro that accepted it.
    424   /// e.g.
    425   /// \code
    426   ///   MAC1( MAC2(foo) )
    427   /// \endcode
    428   /// for location of 'foo' token, this function will return "MAC1" while
    429   /// Lexer::getImmediateMacroName will return "MAC2".
    430   static StringRef getImmediateMacroNameForDiagnostics(
    431       SourceLocation Loc, const SourceManager &SM, const LangOptions &LangOpts);
    432 
    433   /// \brief Compute the preamble of the given file.
    434   ///
    435   /// The preamble of a file contains the initial comments, include directives,
    436   /// and other preprocessor directives that occur before the code in this
    437   /// particular file actually begins. The preamble of the main source file is
    438   /// a potential prefix header.
    439   ///
    440   /// \param Buffer The memory buffer containing the file's contents.
    441   ///
    442   /// \param MaxLines If non-zero, restrict the length of the preamble
    443   /// to fewer than this number of lines.
    444   ///
    445   /// \returns The offset into the file where the preamble ends and the rest
    446   /// of the file begins along with a boolean value indicating whether
    447   /// the preamble ends at the beginning of a new line.
    448   static std::pair<unsigned, bool> ComputePreamble(StringRef Buffer,
    449                                                    const LangOptions &LangOpts,
    450                                                    unsigned MaxLines = 0);
    451 
    452   /// \brief Checks that the given token is the first token that occurs after
    453   /// the given location (this excludes comments and whitespace). Returns the
    454   /// location immediately after the specified token. If the token is not found
    455   /// or the location is inside a macro, the returned source location will be
    456   /// invalid.
    457   static SourceLocation findLocationAfterToken(SourceLocation loc,
    458                                          tok::TokenKind TKind,
    459                                          const SourceManager &SM,
    460                                          const LangOptions &LangOpts,
    461                                          bool SkipTrailingWhitespaceAndNewLine);
    462 
    463   /// \brief Returns true if the given character could appear in an identifier.
    464   static bool isIdentifierBodyChar(char c, const LangOptions &LangOpts);
    465 
    466   /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever
    467   /// emit a warning.
    468   static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size,
    469                                           const LangOptions &LangOpts) {
    470     // If this is not a trigraph and not a UCN or escaped newline, return
    471     // quickly.
    472     if (isObviouslySimpleCharacter(Ptr[0])) {
    473       Size = 1;
    474       return *Ptr;
    475     }
    476 
    477     Size = 0;
    478     return getCharAndSizeSlowNoWarn(Ptr, Size, LangOpts);
    479   }
    480 
    481   //===--------------------------------------------------------------------===//
    482   // Internal implementation interfaces.
    483 private:
    484 
    485   /// LexTokenInternal - Internal interface to lex a preprocessing token. Called
    486   /// by Lex.
    487   ///
    488   bool LexTokenInternal(Token &Result, bool TokAtPhysicalStartOfLine);
    489 
    490   bool CheckUnicodeWhitespace(Token &Result, uint32_t C, const char *CurPtr);
    491 
    492   /// Given that a token begins with the Unicode character \p C, figure out
    493   /// what kind of token it is and dispatch to the appropriate lexing helper
    494   /// function.
    495   bool LexUnicode(Token &Result, uint32_t C, const char *CurPtr);
    496 
    497   /// FormTokenWithChars - When we lex a token, we have identified a span
    498   /// starting at BufferPtr, going to TokEnd that forms the token.  This method
    499   /// takes that range and assigns it to the token as its location and size.  In
    500   /// addition, since tokens cannot overlap, this also updates BufferPtr to be
    501   /// TokEnd.
    502   void FormTokenWithChars(Token &Result, const char *TokEnd,
    503                           tok::TokenKind Kind) {
    504     unsigned TokLen = TokEnd-BufferPtr;
    505     Result.setLength(TokLen);
    506     Result.setLocation(getSourceLocation(BufferPtr, TokLen));
    507     Result.setKind(Kind);
    508     BufferPtr = TokEnd;
    509   }
    510 
    511   /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a
    512   /// tok::l_paren token, 0 if it is something else and 2 if there are no more
    513   /// tokens in the buffer controlled by this lexer.
    514   unsigned isNextPPTokenLParen();
    515 
    516   //===--------------------------------------------------------------------===//
    517   // Lexer character reading interfaces.
    518 
    519   // This lexer is built on two interfaces for reading characters, both of which
    520   // automatically provide phase 1/2 translation.  getAndAdvanceChar is used
    521   // when we know that we will be reading a character from the input buffer and
    522   // that this character will be part of the result token. This occurs in (f.e.)
    523   // string processing, because we know we need to read until we find the
    524   // closing '"' character.
    525   //
    526   // The second interface is the combination of getCharAndSize with
    527   // ConsumeChar.  getCharAndSize reads a phase 1/2 translated character,
    528   // returning it and its size.  If the lexer decides that this character is
    529   // part of the current token, it calls ConsumeChar on it.  This two stage
    530   // approach allows us to emit diagnostics for characters (e.g. warnings about
    531   // trigraphs), knowing that they only are emitted if the character is
    532   // consumed.
    533 
    534   /// isObviouslySimpleCharacter - Return true if the specified character is
    535   /// obviously the same in translation phase 1 and translation phase 3.  This
    536   /// can return false for characters that end up being the same, but it will
    537   /// never return true for something that needs to be mapped.
    538   static bool isObviouslySimpleCharacter(char C) {
    539     return C != '?' && C != '\\';
    540   }
    541 
    542   /// getAndAdvanceChar - Read a single 'character' from the specified buffer,
    543   /// advance over it, and return it.  This is tricky in several cases.  Here we
    544   /// just handle the trivial case and fall-back to the non-inlined
    545   /// getCharAndSizeSlow method to handle the hard case.
    546   inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) {
    547     // If this is not a trigraph and not a UCN or escaped newline, return
    548     // quickly.
    549     if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++;
    550 
    551     unsigned Size = 0;
    552     char C = getCharAndSizeSlow(Ptr, Size, &Tok);
    553     Ptr += Size;
    554     return C;
    555   }
    556 
    557   /// ConsumeChar - When a character (identified by getCharAndSize) is consumed
    558   /// and added to a given token, check to see if there are diagnostics that
    559   /// need to be emitted or flags that need to be set on the token.  If so, do
    560   /// it.
    561   const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) {
    562     // Normal case, we consumed exactly one token.  Just return it.
    563     if (Size == 1)
    564       return Ptr+Size;
    565 
    566     // Otherwise, re-lex the character with a current token, allowing
    567     // diagnostics to be emitted and flags to be set.
    568     Size = 0;
    569     getCharAndSizeSlow(Ptr, Size, &Tok);
    570     return Ptr+Size;
    571   }
    572 
    573   /// getCharAndSize - Peek a single 'character' from the specified buffer,
    574   /// get its size, and return it.  This is tricky in several cases.  Here we
    575   /// just handle the trivial case and fall-back to the non-inlined
    576   /// getCharAndSizeSlow method to handle the hard case.
    577   inline char getCharAndSize(const char *Ptr, unsigned &Size) {
    578     // If this is not a trigraph and not a UCN or escaped newline, return
    579     // quickly.
    580     if (isObviouslySimpleCharacter(Ptr[0])) {
    581       Size = 1;
    582       return *Ptr;
    583     }
    584 
    585     Size = 0;
    586     return getCharAndSizeSlow(Ptr, Size);
    587   }
    588 
    589   /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize
    590   /// method.
    591   char getCharAndSizeSlow(const char *Ptr, unsigned &Size,
    592                           Token *Tok = nullptr);
    593 
    594   /// getEscapedNewLineSize - Return the size of the specified escaped newline,
    595   /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry
    596   /// to this function.
    597   static unsigned getEscapedNewLineSize(const char *P);
    598 
    599   /// SkipEscapedNewLines - If P points to an escaped newline (or a series of
    600   /// them), skip over them and return the first non-escaped-newline found,
    601   /// otherwise return P.
    602   static const char *SkipEscapedNewLines(const char *P);
    603 
    604   /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a
    605   /// diagnostic.
    606   static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size,
    607                                        const LangOptions &LangOpts);
    608 
    609   //===--------------------------------------------------------------------===//
    610   // Other lexer functions.
    611 
    612   void SkipBytes(unsigned Bytes, bool StartOfLine);
    613 
    614   void PropagateLineStartLeadingSpaceInfo(Token &Result);
    615 
    616   const char *LexUDSuffix(Token &Result, const char *CurPtr,
    617                           bool IsStringLiteral);
    618 
    619   // Helper functions to lex the remainder of a token of the specific type.
    620   bool LexIdentifier         (Token &Result, const char *CurPtr);
    621   bool LexNumericConstant    (Token &Result, const char *CurPtr);
    622   bool LexStringLiteral      (Token &Result, const char *CurPtr,
    623                               tok::TokenKind Kind);
    624   bool LexRawStringLiteral   (Token &Result, const char *CurPtr,
    625                               tok::TokenKind Kind);
    626   bool LexAngledStringLiteral(Token &Result, const char *CurPtr);
    627   bool LexCharConstant       (Token &Result, const char *CurPtr,
    628                               tok::TokenKind Kind);
    629   bool LexEndOfFile          (Token &Result, const char *CurPtr);
    630   bool SkipWhitespace        (Token &Result, const char *CurPtr,
    631                               bool &TokAtPhysicalStartOfLine);
    632   bool SkipLineComment       (Token &Result, const char *CurPtr,
    633                               bool &TokAtPhysicalStartOfLine);
    634   bool SkipBlockComment      (Token &Result, const char *CurPtr,
    635                               bool &TokAtPhysicalStartOfLine);
    636   bool SaveLineComment       (Token &Result, const char *CurPtr);
    637 
    638   bool IsStartOfConflictMarker(const char *CurPtr);
    639   bool HandleEndOfConflictMarker(const char *CurPtr);
    640 
    641   bool isCodeCompletionPoint(const char *CurPtr) const;
    642   void cutOffLexing() { BufferPtr = BufferEnd; }
    643 
    644   bool isHexaLiteral(const char *Start, const LangOptions &LangOpts);
    645 
    646 
    647   /// Read a universal character name.
    648   ///
    649   /// \param CurPtr The position in the source buffer after the initial '\'.
    650   ///               If the UCN is syntactically well-formed (but not necessarily
    651   ///               valid), this parameter will be updated to point to the
    652   ///               character after the UCN.
    653   /// \param SlashLoc The position in the source buffer of the '\'.
    654   /// \param Tok The token being formed. Pass \c NULL to suppress diagnostics
    655   ///            and handle token formation in the caller.
    656   ///
    657   /// \return The Unicode codepoint specified by the UCN, or 0 if the UCN is
    658   ///         invalid.
    659   uint32_t tryReadUCN(const char *&CurPtr, const char *SlashLoc, Token *Tok);
    660 
    661   /// \brief Try to consume a UCN as part of an identifier at the current
    662   /// location.
    663   /// \param CurPtr Initially points to the range of characters in the source
    664   ///               buffer containing the '\'. Updated to point past the end of
    665   ///               the UCN on success.
    666   /// \param Size The number of characters occupied by the '\' (including
    667   ///             trigraphs and escaped newlines).
    668   /// \param Result The token being produced. Marked as containing a UCN on
    669   ///               success.
    670   /// \return \c true if a UCN was lexed and it produced an acceptable
    671   ///         identifier character, \c false otherwise.
    672   bool tryConsumeIdentifierUCN(const char *&CurPtr, unsigned Size,
    673                                Token &Result);
    674 
    675   /// \brief Try to consume an identifier character encoded in UTF-8.
    676   /// \param CurPtr Points to the start of the (potential) UTF-8 code unit
    677   ///        sequence. On success, updated to point past the end of it.
    678   /// \return \c true if a UTF-8 sequence mapping to an acceptable identifier
    679   ///         character was lexed, \c false otherwise.
    680   bool tryConsumeIdentifierUTF8Char(const char *&CurPtr);
    681 };
    682 
    683 }  // end namespace clang
    684 
    685 #endif
    686