LiteralSupport.h revision 859b6227694033dd6eaf3991a2b80877a406c382 
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
26namespace clang {
27
28class DiagnosticsEngine;
29class Preprocessor;
30class Token;
31class SourceLocation;
32class TargetInfo;
33class SourceManager;
34class 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.
39class 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
51public:
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  static bool isValidUDSuffix(const LangOptions &LangOpts, StringRef Suffix);
83
84  unsigned getRadix() const { return radix; }
85
86  /// GetIntegerValue - Convert this numeric literal value to an APInt that
87  /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
88  /// value read is larger than the APInt's bits will hold), set Val to the low
89  /// bits of the result and return true.  Otherwise, return false.
90  bool GetIntegerValue(llvm::APInt &Val);
91
92  /// GetFloatValue - Convert this numeric literal to a floating value, using
93  /// the specified APFloat fltSemantics (specifying float, double, etc).
94  /// The optional bool isExact (passed-by-reference) has its value
95  /// set to true if the returned APFloat can represent the number in the
96  /// literal exactly, and false otherwise.
97  llvm::APFloat::opStatus GetFloatValue(llvm::APFloat &Result);
98
99private:
100
101  void ParseNumberStartingWithZero(SourceLocation TokLoc);
102
103  static bool isDigitSeparator(char C) { return C == '\''; }
104
105  /// \brief Ensure that we don't have a digit separator here.
106  void checkSeparator(SourceLocation TokLoc, const char *Pos,
107                      bool IsAfterDigits);
108
109  /// SkipHexDigits - Read and skip over any hex digits, up to End.
110  /// Return a pointer to the first non-hex digit or End.
111  const char *SkipHexDigits(const char *ptr) {
112    while (ptr != ThisTokEnd && (isHexDigit(*ptr) || isDigitSeparator(*ptr)))
113      ptr++;
114    return ptr;
115  }
116
117  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
118  /// Return a pointer to the first non-hex digit or End.
119  const char *SkipOctalDigits(const char *ptr) {
120    while (ptr != ThisTokEnd &&
121           ((*ptr >= '0' && *ptr <= '7') || isDigitSeparator(*ptr)))
122      ptr++;
123    return ptr;
124  }
125
126  /// SkipDigits - Read and skip over any digits, up to End.
127  /// Return a pointer to the first non-hex digit or End.
128  const char *SkipDigits(const char *ptr) {
129    while (ptr != ThisTokEnd && (isDigit(*ptr) || isDigitSeparator(*ptr)))
130      ptr++;
131    return ptr;
132  }
133
134  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
135  /// Return a pointer to the first non-binary digit or End.
136  const char *SkipBinaryDigits(const char *ptr) {
137    while (ptr != ThisTokEnd &&
138           (*ptr == '0' || *ptr == '1' || isDigitSeparator(*ptr)))
139      ptr++;
140    return ptr;
141  }
142
143};
144
145/// CharLiteralParser - Perform interpretation and semantic analysis of a
146/// character literal.
147class CharLiteralParser {
148  uint64_t Value;
149  tok::TokenKind Kind;
150  bool IsMultiChar;
151  bool HadError;
152  SmallString<32> UDSuffixBuf;
153  unsigned UDSuffixOffset;
154public:
155  CharLiteralParser(const char *begin, const char *end,
156                    SourceLocation Loc, Preprocessor &PP,
157                    tok::TokenKind kind);
158
159  bool hadError() const { return HadError; }
160  bool isAscii() const { return Kind == tok::char_constant; }
161  bool isWide() const { return Kind == tok::wide_char_constant; }
162  bool isUTF16() const { return Kind == tok::utf16_char_constant; }
163  bool isUTF32() const { return Kind == tok::utf32_char_constant; }
164  bool isMultiChar() const { return IsMultiChar; }
165  uint64_t getValue() const { return Value; }
166  StringRef getUDSuffix() const { return UDSuffixBuf; }
167  unsigned getUDSuffixOffset() const {
168    assert(!UDSuffixBuf.empty() && "no ud-suffix");
169    return UDSuffixOffset;
170  }
171};
172
173/// StringLiteralParser - This decodes string escape characters and performs
174/// wide string analysis and Translation Phase #6 (concatenation of string
175/// literals) (C99 5.1.1.2p1).
176class StringLiteralParser {
177  const SourceManager &SM;
178  const LangOptions &Features;
179  const TargetInfo &Target;
180  DiagnosticsEngine *Diags;
181
182  unsigned MaxTokenLength;
183  unsigned SizeBound;
184  unsigned CharByteWidth;
185  tok::TokenKind Kind;
186  SmallString<512> ResultBuf;
187  char *ResultPtr; // cursor
188  SmallString<32> UDSuffixBuf;
189  unsigned UDSuffixToken;
190  unsigned UDSuffixOffset;
191public:
192  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
193                      Preprocessor &PP, bool Complain = true);
194  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
195                      const SourceManager &sm, const LangOptions &features,
196                      const TargetInfo &target, DiagnosticsEngine *diags = 0)
197    : SM(sm), Features(features), Target(target), Diags(diags),
198      MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
199      ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
200    init(StringToks, NumStringToks);
201  }
202
203
204  bool hadError;
205  bool Pascal;
206
207  StringRef GetString() const {
208    return StringRef(ResultBuf.data(), GetStringLength());
209  }
210  unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
211
212  unsigned GetNumStringChars() const {
213    return GetStringLength() / CharByteWidth;
214  }
215  /// getOffsetOfStringByte - This function returns the offset of the
216  /// specified byte of the string data represented by Token.  This handles
217  /// advancing over escape sequences in the string.
218  ///
219  /// If the Diagnostics pointer is non-null, then this will do semantic
220  /// checking of the string literal and emit errors and warnings.
221  unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
222
223  bool isAscii() const { return Kind == tok::string_literal; }
224  bool isWide() const { return Kind == tok::wide_string_literal; }
225  bool isUTF8() const { return Kind == tok::utf8_string_literal; }
226  bool isUTF16() const { return Kind == tok::utf16_string_literal; }
227  bool isUTF32() const { return Kind == tok::utf32_string_literal; }
228  bool isPascal() const { return Pascal; }
229
230  StringRef getUDSuffix() const { return UDSuffixBuf; }
231
232  /// Get the index of a token containing a ud-suffix.
233  unsigned getUDSuffixToken() const {
234    assert(!UDSuffixBuf.empty() && "no ud-suffix");
235    return UDSuffixToken;
236  }
237  /// Get the spelling offset of the first byte of the ud-suffix.
238  unsigned getUDSuffixOffset() const {
239    assert(!UDSuffixBuf.empty() && "no ud-suffix");
240    return UDSuffixOffset;
241  }
242
243private:
244  void init(const Token *StringToks, unsigned NumStringToks);
245  bool CopyStringFragment(const Token &Tok, const char *TokBegin,
246                          StringRef Fragment);
247  void DiagnoseLexingError(SourceLocation Loc);
248};
249
250}  // end namespace clang
251
252#endif
253