LiteralSupport.h revision 686775deca8b8685eb90801495880e3abdd844c2 
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/LLVM.h"
19#include "llvm/ADT/APFloat.h"
20#include "llvm/ADT/SmallString.h"
21#include "llvm/Support/DataTypes.h"
22#include <cctype>
23
24namespace clang {
25
26class Diagnostic;
27class Preprocessor;
28class Token;
29class SourceLocation;
30class TargetInfo;
31class SourceManager;
32class LangOptions;
33
34/// NumericLiteralParser - This performs strict semantic analysis of the content
35/// of a ppnumber, classifying it as either integer, floating, or erroneous,
36/// determines the radix of the value and can convert it to a useful value.
37class NumericLiteralParser {
38  Preprocessor &PP; // needed for diagnostics
39
40  const char *const ThisTokBegin;
41  const char *const ThisTokEnd;
42  const char *DigitsBegin, *SuffixBegin; // markers
43  const char *s; // cursor
44
45  unsigned radix;
46
47  bool saw_exponent, saw_period;
48
49public:
50  NumericLiteralParser(const char *begin, const char *end,
51                       SourceLocation Loc, Preprocessor &PP);
52  bool hadError;
53  bool isUnsigned;
54  bool isLong;        // This is *not* set for long long.
55  bool isLongLong;
56  bool isFloat;       // 1.0f
57  bool isImaginary;   // 1.0i
58  bool isMicrosoftInteger;  // Microsoft suffix extension i8, i16, i32, or i64.
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::opStatus GetFloatValue(llvm::APFloat &Result);
84
85private:
86
87  void ParseNumberStartingWithZero(SourceLocation TokLoc);
88
89  /// SkipHexDigits - Read and skip over any hex digits, up to End.
90  /// Return a pointer to the first non-hex digit or End.
91  const char *SkipHexDigits(const char *ptr) {
92    while (ptr != ThisTokEnd && isxdigit(*ptr))
93      ptr++;
94    return ptr;
95  }
96
97  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
98  /// Return a pointer to the first non-hex digit or End.
99  const char *SkipOctalDigits(const char *ptr) {
100    while (ptr != ThisTokEnd && ((*ptr >= '0') && (*ptr <= '7')))
101      ptr++;
102    return ptr;
103  }
104
105  /// SkipDigits - Read and skip over any digits, up to End.
106  /// Return a pointer to the first non-hex digit or End.
107  const char *SkipDigits(const char *ptr) {
108    while (ptr != ThisTokEnd && isdigit(*ptr))
109      ptr++;
110    return ptr;
111  }
112
113  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
114  /// Return a pointer to the first non-binary digit or End.
115  const char *SkipBinaryDigits(const char *ptr) {
116    while (ptr != ThisTokEnd && (*ptr == '0' || *ptr == '1'))
117      ptr++;
118    return ptr;
119  }
120
121};
122
123/// CharLiteralParser - Perform interpretation and semantic analysis of a
124/// character literal.
125class CharLiteralParser {
126  uint64_t Value;
127  bool IsWide;
128  bool IsMultiChar;
129  bool HadError;
130public:
131  CharLiteralParser(const char *begin, const char *end,
132                    SourceLocation Loc, Preprocessor &PP);
133
134  bool hadError() const { return HadError; }
135  bool isWide() const { return IsWide; }
136  bool isMultiChar() const { return IsMultiChar; }
137  uint64_t getValue() const { return Value; }
138};
139
140/// StringLiteralParser - This decodes string escape characters and performs
141/// wide string analysis and Translation Phase #6 (concatenation of string
142/// literals) (C99 5.1.1.2p1).
143class StringLiteralParser {
144  const SourceManager &SM;
145  const LangOptions &Features;
146  const TargetInfo &Target;
147  Diagnostic *Diags;
148
149  unsigned MaxTokenLength;
150  unsigned SizeBound;
151  unsigned wchar_tByteWidth;
152  llvm::SmallString<512> ResultBuf;
153  char *ResultPtr; // cursor
154public:
155  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
156                      Preprocessor &PP, bool Complain = true);
157  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
158                      const SourceManager &sm, const LangOptions &features,
159                      const TargetInfo &target, Diagnostic *diags = 0)
160    : SM(sm), Features(features), Target(target), Diags(diags),
161      MaxTokenLength(0), SizeBound(0), wchar_tByteWidth(0),
162      ResultPtr(ResultBuf.data()), hadError(false), AnyWide(false), Pascal(false) {
163    init(StringToks, NumStringToks);
164  }
165
166
167  bool hadError;
168  bool AnyWide;
169  bool Pascal;
170
171  StringRef GetString() const {
172    return StringRef(ResultBuf.data(), GetStringLength());
173  }
174  unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
175
176  unsigned GetNumStringChars() const {
177    if (AnyWide)
178      return GetStringLength() / wchar_tByteWidth;
179    return GetStringLength();
180  }
181  /// getOffsetOfStringByte - This function returns the offset of the
182  /// specified byte of the string data represented by Token.  This handles
183  /// advancing over escape sequences in the string.
184  ///
185  /// If the Diagnostics pointer is non-null, then this will do semantic
186  /// checking of the string literal and emit errors and warnings.
187  unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
188
189private:
190  void init(const Token *StringToks, unsigned NumStringToks);
191};
192
193}  // end namespace clang
194
195#endif
196