LiteralSupport.h revision 2a1c363f38e59a5044fc349aa7e538a50954c244 
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 <string>
19#include "llvm/ADT/SmallString.h"
20
21namespace llvm {
22  class APInt;
23  class APFloat;
24  struct fltSemantics;
25}
26
27namespace clang {
28
29class Diagnostic;
30class Preprocessor;
31class Token;
32class SourceLocation;
33class TargetInfo;
34
35/// NumericLiteralParser - This performs strict semantic analysis of the content
36/// of a ppnumber, classifying it as either integer, floating, or erroneous,
37/// determines the radix of the value and can convert it to a useful value.
38class NumericLiteralParser {
39  Preprocessor &PP; // needed for diagnostics
40
41  const char *const ThisTokBegin;
42  const char *const ThisTokEnd;
43  const char *DigitsBegin, *SuffixBegin; // markers
44  const char *s; // cursor
45
46  unsigned radix;
47
48  bool saw_exponent, saw_period;
49
50public:
51  NumericLiteralParser(const char *begin, const char *end,
52                       SourceLocation Loc, Preprocessor &PP);
53  bool hadError;
54  bool isUnsigned;
55  bool isLong;        // This is *not* set for long long.
56  bool isLongLong;
57  bool isFloat;       // 1.0f
58  bool isImaginary;   // 1.0i
59
60  bool isIntegerLiteral() const {
61    return !saw_period && !saw_exponent;
62  }
63  bool isFloatingLiteral() const {
64    return saw_period || saw_exponent;
65  }
66  bool hasSuffix() const {
67    return SuffixBegin != ThisTokEnd;
68  }
69
70  unsigned getRadix() const { return radix; }
71
72  /// GetIntegerValue - Convert this numeric literal value to an APInt that
73  /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
74  /// value read is larger than the APInt's bits will hold), set Val to the low
75  /// bits of the result and return true.  Otherwise, return false.
76  bool GetIntegerValue(llvm::APInt &Val);
77
78  /// GetFloatValue - Convert this numeric literal to a floating value, using
79  /// the specified APFloat fltSemantics (specifying float, double, etc).
80  /// The optional bool isExact (passed-by-reference) has its value
81  /// set to true if the returned APFloat can represent the number in the
82  /// literal exactly, and false otherwise.
83  llvm::APFloat GetFloatValue(const llvm::fltSemantics &Format,
84                              bool* isExact = NULL);
85
86private:
87
88  void ParseNumberStartingWithZero(SourceLocation TokLoc);
89
90  /// SkipHexDigits - Read and skip over any hex digits, up to End.
91  /// Return a pointer to the first non-hex digit or End.
92  const char *SkipHexDigits(const char *ptr) {
93    while (ptr != ThisTokEnd && isxdigit(*ptr))
94      ptr++;
95    return ptr;
96  }
97
98  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
99  /// Return a pointer to the first non-hex digit or End.
100  const char *SkipOctalDigits(const char *ptr) {
101    while (ptr != ThisTokEnd && ((*ptr >= '0') && (*ptr <= '7')))
102      ptr++;
103    return ptr;
104  }
105
106  /// SkipDigits - Read and skip over any digits, up to End.
107  /// Return a pointer to the first non-hex digit or End.
108  const char *SkipDigits(const char *ptr) {
109    while (ptr != ThisTokEnd && isdigit(*ptr))
110      ptr++;
111    return ptr;
112  }
113
114  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
115  /// Return a pointer to the first non-binary digit or End.
116  const char *SkipBinaryDigits(const char *ptr) {
117    while (ptr != ThisTokEnd && (*ptr == '0' || *ptr == '1'))
118      ptr++;
119    return ptr;
120  }
121
122};
123
124/// CharLiteralParser - Perform interpretation and semantic analysis of a
125/// character literal.
126class CharLiteralParser {
127  uint64_t Value;
128  bool IsWide;
129  bool IsMultiChar;
130  bool HadError;
131public:
132  CharLiteralParser(const char *begin, const char *end,
133                    SourceLocation Loc, Preprocessor &PP);
134
135  bool hadError() const { return HadError; }
136  bool isWide() const { return IsWide; }
137  bool isMultiChar() const { return IsMultiChar; }
138  uint64_t getValue() const { return Value; }
139};
140
141/// StringLiteralParser - This decodes string escape characters and performs
142/// wide string analysis and Translation Phase #6 (concatenation of string
143/// literals) (C99 5.1.1.2p1).
144class StringLiteralParser {
145  Preprocessor &PP;
146
147  unsigned MaxTokenLength;
148  unsigned SizeBound;
149  unsigned wchar_tByteWidth;
150  llvm::SmallString<512> ResultBuf;
151  char *ResultPtr; // cursor
152public:
153  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
154                      Preprocessor &PP);
155  bool hadError;
156  bool AnyWide;
157  bool Pascal;
158
159  const char *GetString() { return &ResultBuf[0]; }
160  unsigned GetStringLength() const { return ResultPtr-&ResultBuf[0]; }
161
162  unsigned GetNumStringChars() const {
163    if (AnyWide)
164      return GetStringLength() / wchar_tByteWidth;
165    return GetStringLength();
166  }
167  /// getOffsetOfStringByte - This function returns the offset of the
168  /// specified byte of the string data represented by Token.  This handles
169  /// advancing over escape sequences in the string.
170  static unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo,
171                                        Preprocessor &PP);
172};
173
174}  // end namespace clang
175
176#endif
177