Home | History | Annotate | Download | only in Lex 
      1 //===--- LiteralSupport.h ---------------------------------------*- 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 NumericLiteralParser, CharLiteralParser, and
     11 // StringLiteralParser interfaces.
     12 //
     13 //===----------------------------------------------------------------------===//
     14 
     15 #ifndef CLANG_LITERALSUPPORT_H
     16 #define CLANG_LITERALSUPPORT_H
     17 
     18 #include "llvm/ADT/APFloat.h"
     19 #include "llvm/ADT/SmallString.h"
     20 #include "llvm/Support/DataTypes.h"
     21 #include <cctype>
     22 
     23 namespace clang {
     24 
     25 class Diagnostic;
     26 class Preprocessor;
     27 class Token;
     28 class SourceLocation;
     29 class TargetInfo;
     30 class SourceManager;
     31 class LangOptions;
     32 
     33 /// NumericLiteralParser - This performs strict semantic analysis of the content
     34 /// of a ppnumber, classifying it as either integer, floating, or erroneous,
     35 /// determines the radix of the value and can convert it to a useful value.
     36 class NumericLiteralParser {
     37   Preprocessor &PP; // needed for diagnostics
     38 
     39   const char *const ThisTokBegin;
     40   const char *const ThisTokEnd;
     41   const char *DigitsBegin, *SuffixBegin; // markers
     42   const char *s; // cursor
     43 
     44   unsigned radix;
     45 
     46   bool saw_exponent, saw_period;
     47 
     48 public:
     49   NumericLiteralParser(const char *begin, const char *end,
     50                        SourceLocation Loc, Preprocessor &PP);
     51   bool hadError;
     52   bool isUnsigned;
     53   bool isLong;        // This is *not* set for long long.
     54   bool isLongLong;
     55   bool isFloat;       // 1.0f
     56   bool isImaginary;   // 1.0i
     57   bool isMicrosoftInteger;  // Microsoft suffix extension i8, i16, i32, or i64.
     58 
     59   bool isIntegerLiteral() const {
     60     return !saw_period && !saw_exponent;
     61   }
     62   bool isFloatingLiteral() const {
     63     return saw_period || saw_exponent;
     64   }
     65   bool hasSuffix() const {
     66     return SuffixBegin != ThisTokEnd;
     67   }
     68 
     69   unsigned getRadix() const { return radix; }
     70 
     71   /// GetIntegerValue - Convert this numeric literal value to an APInt that
     72   /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
     73   /// value read is larger than the APInt's bits will hold), set Val to the low
     74   /// bits of the result and return true.  Otherwise, return false.
     75   bool GetIntegerValue(llvm::APInt &Val);
     76 
     77   /// GetFloatValue - Convert this numeric literal to a floating value, using
     78   /// the specified APFloat fltSemantics (specifying float, double, etc).
     79   /// The optional bool isExact (passed-by-reference) has its value
     80   /// set to true if the returned APFloat can represent the number in the
     81   /// literal exactly, and false otherwise.
     82   llvm::APFloat::opStatus GetFloatValue(llvm::APFloat &Result);
     83 
     84 private:
     85 
     86   void ParseNumberStartingWithZero(SourceLocation TokLoc);
     87 
     88   /// SkipHexDigits - Read and skip over any hex digits, up to End.
     89   /// Return a pointer to the first non-hex digit or End.
     90   const char *SkipHexDigits(const char *ptr) {
     91     while (ptr != ThisTokEnd && isxdigit(*ptr))
     92       ptr++;
     93     return ptr;
     94   }
     95 
     96   /// SkipOctalDigits - Read and skip over any octal digits, up to End.
     97   /// Return a pointer to the first non-hex digit or End.
     98   const char *SkipOctalDigits(const char *ptr) {
     99     while (ptr != ThisTokEnd && ((*ptr >= '0') && (*ptr <= '7')))
    100       ptr++;
    101     return ptr;
    102   }
    103 
    104   /// SkipDigits - Read and skip over any digits, up to End.
    105   /// Return a pointer to the first non-hex digit or End.
    106   const char *SkipDigits(const char *ptr) {
    107     while (ptr != ThisTokEnd && isdigit(*ptr))
    108       ptr++;
    109     return ptr;
    110   }
    111 
    112   /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
    113   /// Return a pointer to the first non-binary digit or End.
    114   const char *SkipBinaryDigits(const char *ptr) {
    115     while (ptr != ThisTokEnd && (*ptr == '0' || *ptr == '1'))
    116       ptr++;
    117     return ptr;
    118   }
    119 
    120 };
    121 
    122 /// CharLiteralParser - Perform interpretation and semantic analysis of a
    123 /// character literal.
    124 class CharLiteralParser {
    125   uint64_t Value;
    126   bool IsWide;
    127   bool IsMultiChar;
    128   bool HadError;
    129 public:
    130   CharLiteralParser(const char *begin, const char *end,
    131                     SourceLocation Loc, Preprocessor &PP);
    132 
    133   bool hadError() const { return HadError; }
    134   bool isWide() const { return IsWide; }
    135   bool isMultiChar() const { return IsMultiChar; }
    136   uint64_t getValue() const { return Value; }
    137 };
    138 
    139 /// StringLiteralParser - This decodes string escape characters and performs
    140 /// wide string analysis and Translation Phase #6 (concatenation of string
    141 /// literals) (C99 5.1.1.2p1).
    142 class StringLiteralParser {
    143   const SourceManager &SM;
    144   const LangOptions &Features;
    145   const TargetInfo &Target;
    146   Diagnostic *Diags;
    147 
    148   unsigned MaxTokenLength;
    149   unsigned SizeBound;
    150   unsigned wchar_tByteWidth;
    151   llvm::SmallString<512> ResultBuf;
    152   char *ResultPtr; // cursor
    153 public:
    154   StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
    155                       Preprocessor &PP, bool Complain = true);
    156   StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
    157                       const SourceManager &sm, const LangOptions &features,
    158                       const TargetInfo &target, Diagnostic *diags = 0)
    159     : SM(sm), Features(features), Target(target), Diags(diags),
    160       MaxTokenLength(0), SizeBound(0), wchar_tByteWidth(0),
    161       ResultPtr(ResultBuf.data()), hadError(false), AnyWide(false), Pascal(false) {
    162     init(StringToks, NumStringToks);
    163   }
    164 
    165 
    166   bool hadError;
    167   bool AnyWide;
    168   bool Pascal;
    169 
    170   llvm::StringRef GetString() const {
    171     return llvm::StringRef(ResultBuf.data(), GetStringLength());
    172   }
    173   unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
    174 
    175   unsigned GetNumStringChars() const {
    176     if (AnyWide)
    177       return GetStringLength() / wchar_tByteWidth;
    178     return GetStringLength();
    179   }
    180   /// getOffsetOfStringByte - This function returns the offset of the
    181   /// specified byte of the string data represented by Token.  This handles
    182   /// advancing over escape sequences in the string.
    183   ///
    184   /// If the Diagnostics pointer is non-null, then this will do semantic
    185   /// checking of the string literal and emit errors and warnings.
    186   unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
    187 
    188 private:
    189   void init(const Token *StringToks, unsigned NumStringToks);
    190 };
    191 
    192 }  // end namespace clang
    193 
    194 #endif
    195