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 "clang/Basic/CharInfo.h"
     19 #include "clang/Basic/LLVM.h"
     20 #include "clang/Basic/TokenKinds.h"
     21 #include "llvm/ADT/APFloat.h"
     22 #include "llvm/ADT/SmallString.h"
     23 #include "llvm/ADT/StringRef.h"
     24 #include "llvm/Support/DataTypes.h"
     25 
     26 namespace clang {
     27 
     28 class DiagnosticsEngine;
     29 class Preprocessor;
     30 class Token;
     31 class SourceLocation;
     32 class TargetInfo;
     33 class SourceManager;
     34 class LangOptions;
     35 
     36 /// NumericLiteralParser - This performs strict semantic analysis of the content
     37 /// of a ppnumber, classifying it as either integer, floating, or erroneous,
     38 /// determines the radix of the value and can convert it to a useful value.
     39 class NumericLiteralParser {
     40   Preprocessor &PP; // needed for diagnostics
     41 
     42   const char *const ThisTokBegin;
     43   const char *const ThisTokEnd;
     44   const char *DigitsBegin, *SuffixBegin; // markers
     45   const char *s; // cursor
     46 
     47   unsigned radix;
     48 
     49   bool saw_exponent, saw_period, saw_ud_suffix;
     50 
     51 public:
     52   NumericLiteralParser(StringRef TokSpelling,
     53                        SourceLocation TokLoc,
     54                        Preprocessor &PP);
     55   bool hadError;
     56   bool isUnsigned;
     57   bool isLong;        // This is *not* set for long long.
     58   bool isLongLong;
     59   bool isFloat;       // 1.0f
     60   bool isImaginary;   // 1.0i
     61   bool isMicrosoftInteger;  // Microsoft suffix extension i8, i16, i32, or i64.
     62 
     63   bool isIntegerLiteral() const {
     64     return !saw_period && !saw_exponent;
     65   }
     66   bool isFloatingLiteral() const {
     67     return saw_period || saw_exponent;
     68   }
     69 
     70   bool hasUDSuffix() const {
     71     return saw_ud_suffix;
     72   }
     73   StringRef getUDSuffix() const {
     74     assert(saw_ud_suffix);
     75     return StringRef(SuffixBegin, ThisTokEnd - SuffixBegin);
     76   }
     77   unsigned getUDSuffixOffset() const {
     78     assert(saw_ud_suffix);
     79     return SuffixBegin - ThisTokBegin;
     80   }
     81 
     82   unsigned getRadix() const { return radix; }
     83 
     84   /// GetIntegerValue - Convert this numeric literal value to an APInt that
     85   /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
     86   /// value read is larger than the APInt's bits will hold), set Val to the low
     87   /// bits of the result and return true.  Otherwise, return false.
     88   bool GetIntegerValue(llvm::APInt &Val);
     89 
     90   /// GetFloatValue - Convert this numeric literal to a floating value, using
     91   /// the specified APFloat fltSemantics (specifying float, double, etc).
     92   /// The optional bool isExact (passed-by-reference) has its value
     93   /// set to true if the returned APFloat can represent the number in the
     94   /// literal exactly, and false otherwise.
     95   llvm::APFloat::opStatus GetFloatValue(llvm::APFloat &Result);
     96 
     97 private:
     98 
     99   void ParseNumberStartingWithZero(SourceLocation TokLoc);
    100 
    101   /// SkipHexDigits - Read and skip over any hex digits, up to End.
    102   /// Return a pointer to the first non-hex digit or End.
    103   const char *SkipHexDigits(const char *ptr) {
    104     while (ptr != ThisTokEnd && isHexDigit(*ptr))
    105       ptr++;
    106     return ptr;
    107   }
    108 
    109   /// SkipOctalDigits - Read and skip over any octal digits, up to End.
    110   /// Return a pointer to the first non-hex digit or End.
    111   const char *SkipOctalDigits(const char *ptr) {
    112     while (ptr != ThisTokEnd && ((*ptr >= '0') && (*ptr <= '7')))
    113       ptr++;
    114     return ptr;
    115   }
    116 
    117   /// SkipDigits - Read and skip over any digits, up to End.
    118   /// Return a pointer to the first non-hex digit or End.
    119   const char *SkipDigits(const char *ptr) {
    120     while (ptr != ThisTokEnd && isDigit(*ptr))
    121       ptr++;
    122     return ptr;
    123   }
    124 
    125   /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
    126   /// Return a pointer to the first non-binary digit or End.
    127   const char *SkipBinaryDigits(const char *ptr) {
    128     while (ptr != ThisTokEnd && (*ptr == '0' || *ptr == '1'))
    129       ptr++;
    130     return ptr;
    131   }
    132 
    133 };
    134 
    135 /// CharLiteralParser - Perform interpretation and semantic analysis of a
    136 /// character literal.
    137 class CharLiteralParser {
    138   uint64_t Value;
    139   tok::TokenKind Kind;
    140   bool IsMultiChar;
    141   bool HadError;
    142   SmallString<32> UDSuffixBuf;
    143   unsigned UDSuffixOffset;
    144 public:
    145   CharLiteralParser(const char *begin, const char *end,
    146                     SourceLocation Loc, Preprocessor &PP,
    147                     tok::TokenKind kind);
    148 
    149   bool hadError() const { return HadError; }
    150   bool isAscii() const { return Kind == tok::char_constant; }
    151   bool isWide() const { return Kind == tok::wide_char_constant; }
    152   bool isUTF16() const { return Kind == tok::utf16_char_constant; }
    153   bool isUTF32() const { return Kind == tok::utf32_char_constant; }
    154   bool isMultiChar() const { return IsMultiChar; }
    155   uint64_t getValue() const { return Value; }
    156   StringRef getUDSuffix() const { return UDSuffixBuf; }
    157   unsigned getUDSuffixOffset() const {
    158     assert(!UDSuffixBuf.empty() && "no ud-suffix");
    159     return UDSuffixOffset;
    160   }
    161 };
    162 
    163 /// StringLiteralParser - This decodes string escape characters and performs
    164 /// wide string analysis and Translation Phase #6 (concatenation of string
    165 /// literals) (C99 5.1.1.2p1).
    166 class StringLiteralParser {
    167   const SourceManager &SM;
    168   const LangOptions &Features;
    169   const TargetInfo &Target;
    170   DiagnosticsEngine *Diags;
    171 
    172   unsigned MaxTokenLength;
    173   unsigned SizeBound;
    174   unsigned CharByteWidth;
    175   tok::TokenKind Kind;
    176   SmallString<512> ResultBuf;
    177   char *ResultPtr; // cursor
    178   SmallString<32> UDSuffixBuf;
    179   unsigned UDSuffixToken;
    180   unsigned UDSuffixOffset;
    181 public:
    182   StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
    183                       Preprocessor &PP, bool Complain = true);
    184   StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
    185                       const SourceManager &sm, const LangOptions &features,
    186                       const TargetInfo &target, DiagnosticsEngine *diags = 0)
    187     : SM(sm), Features(features), Target(target), Diags(diags),
    188       MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
    189       ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
    190     init(StringToks, NumStringToks);
    191   }
    192 
    193 
    194   bool hadError;
    195   bool Pascal;
    196 
    197   StringRef GetString() const {
    198     return StringRef(ResultBuf.data(), GetStringLength());
    199   }
    200   unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
    201 
    202   unsigned GetNumStringChars() const {
    203     return GetStringLength() / CharByteWidth;
    204   }
    205   /// getOffsetOfStringByte - This function returns the offset of the
    206   /// specified byte of the string data represented by Token.  This handles
    207   /// advancing over escape sequences in the string.
    208   ///
    209   /// If the Diagnostics pointer is non-null, then this will do semantic
    210   /// checking of the string literal and emit errors and warnings.
    211   unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
    212 
    213   bool isAscii() const { return Kind == tok::string_literal; }
    214   bool isWide() const { return Kind == tok::wide_string_literal; }
    215   bool isUTF8() const { return Kind == tok::utf8_string_literal; }
    216   bool isUTF16() const { return Kind == tok::utf16_string_literal; }
    217   bool isUTF32() const { return Kind == tok::utf32_string_literal; }
    218   bool isPascal() const { return Pascal; }
    219 
    220   StringRef getUDSuffix() const { return UDSuffixBuf; }
    221 
    222   /// Get the index of a token containing a ud-suffix.
    223   unsigned getUDSuffixToken() const {
    224     assert(!UDSuffixBuf.empty() && "no ud-suffix");
    225     return UDSuffixToken;
    226   }
    227   /// Get the spelling offset of the first byte of the ud-suffix.
    228   unsigned getUDSuffixOffset() const {
    229     assert(!UDSuffixBuf.empty() && "no ud-suffix");
    230     return UDSuffixOffset;
    231   }
    232 
    233 private:
    234   void init(const Token *StringToks, unsigned NumStringToks);
    235   bool CopyStringFragment(const Token &Tok, const char *TokBegin,
    236                           StringRef Fragment);
    237   void DiagnoseLexingError(SourceLocation Loc);
    238 };
    239 
    240 }  // end namespace clang
    241 
    242 #endif
    243