LiteralSupport.h revision 064c3f1ed782eb641df6bc00beec069b3c80fa8a 
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  enum CheckSeparatorKind { CSK_BeforeDigits, CSK_AfterDigits };
106
107  /// \brief Ensure that we don't have a digit separator here.
108  void checkSeparator(SourceLocation TokLoc, const char *Pos,
109                      CheckSeparatorKind IsAfterDigits);
110
111  /// SkipHexDigits - Read and skip over any hex digits, up to End.
112  /// Return a pointer to the first non-hex digit or End.
113  const char *SkipHexDigits(const char *ptr) {
114    while (ptr != ThisTokEnd && (isHexDigit(*ptr) || isDigitSeparator(*ptr)))
115      ptr++;
116    return ptr;
117  }
118
119  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
120  /// Return a pointer to the first non-hex digit or End.
121  const char *SkipOctalDigits(const char *ptr) {
122    while (ptr != ThisTokEnd &&
123           ((*ptr >= '0' && *ptr <= '7') || isDigitSeparator(*ptr)))
124      ptr++;
125    return ptr;
126  }
127
128  /// SkipDigits - Read and skip over any digits, up to End.
129  /// Return a pointer to the first non-hex digit or End.
130  const char *SkipDigits(const char *ptr) {
131    while (ptr != ThisTokEnd && (isDigit(*ptr) || isDigitSeparator(*ptr)))
132      ptr++;
133    return ptr;
134  }
135
136  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
137  /// Return a pointer to the first non-binary digit or End.
138  const char *SkipBinaryDigits(const char *ptr) {
139    while (ptr != ThisTokEnd &&
140           (*ptr == '0' || *ptr == '1' || isDigitSeparator(*ptr)))
141      ptr++;
142    return ptr;
143  }
144
145};
146
147/// CharLiteralParser - Perform interpretation and semantic analysis of a
148/// character literal.
149class CharLiteralParser {
150  uint64_t Value;
151  tok::TokenKind Kind;
152  bool IsMultiChar;
153  bool HadError;
154  SmallString<32> UDSuffixBuf;
155  unsigned UDSuffixOffset;
156public:
157  CharLiteralParser(const char *begin, const char *end,
158                    SourceLocation Loc, Preprocessor &PP,
159                    tok::TokenKind kind);
160
161  bool hadError() const { return HadError; }
162  bool isAscii() const { return Kind == tok::char_constant; }
163  bool isWide() const { return Kind == tok::wide_char_constant; }
164  bool isUTF16() const { return Kind == tok::utf16_char_constant; }
165  bool isUTF32() const { return Kind == tok::utf32_char_constant; }
166  bool isMultiChar() const { return IsMultiChar; }
167  uint64_t getValue() const { return Value; }
168  StringRef getUDSuffix() const { return UDSuffixBuf; }
169  unsigned getUDSuffixOffset() const {
170    assert(!UDSuffixBuf.empty() && "no ud-suffix");
171    return UDSuffixOffset;
172  }
173};
174
175/// StringLiteralParser - This decodes string escape characters and performs
176/// wide string analysis and Translation Phase #6 (concatenation of string
177/// literals) (C99 5.1.1.2p1).
178class StringLiteralParser {
179  const SourceManager &SM;
180  const LangOptions &Features;
181  const TargetInfo &Target;
182  DiagnosticsEngine *Diags;
183
184  unsigned MaxTokenLength;
185  unsigned SizeBound;
186  unsigned CharByteWidth;
187  tok::TokenKind Kind;
188  SmallString<512> ResultBuf;
189  char *ResultPtr; // cursor
190  SmallString<32> UDSuffixBuf;
191  unsigned UDSuffixToken;
192  unsigned UDSuffixOffset;
193public:
194  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
195                      Preprocessor &PP, bool Complain = true);
196  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
197                      const SourceManager &sm, const LangOptions &features,
198                      const TargetInfo &target, DiagnosticsEngine *diags = 0)
199    : SM(sm), Features(features), Target(target), Diags(diags),
200      MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
201      ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
202    init(StringToks, NumStringToks);
203  }
204
205
206  bool hadError;
207  bool Pascal;
208
209  StringRef GetString() const {
210    return StringRef(ResultBuf.data(), GetStringLength());
211  }
212  unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
213
214  unsigned GetNumStringChars() const {
215    return GetStringLength() / CharByteWidth;
216  }
217  /// getOffsetOfStringByte - This function returns the offset of the
218  /// specified byte of the string data represented by Token.  This handles
219  /// advancing over escape sequences in the string.
220  ///
221  /// If the Diagnostics pointer is non-null, then this will do semantic
222  /// checking of the string literal and emit errors and warnings.
223  unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
224
225  bool isAscii() const { return Kind == tok::string_literal; }
226  bool isWide() const { return Kind == tok::wide_string_literal; }
227  bool isUTF8() const { return Kind == tok::utf8_string_literal; }
228  bool isUTF16() const { return Kind == tok::utf16_string_literal; }
229  bool isUTF32() const { return Kind == tok::utf32_string_literal; }
230  bool isPascal() const { return Pascal; }
231
232  StringRef getUDSuffix() const { return UDSuffixBuf; }
233
234  /// Get the index of a token containing a ud-suffix.
235  unsigned getUDSuffixToken() const {
236    assert(!UDSuffixBuf.empty() && "no ud-suffix");
237    return UDSuffixToken;
238  }
239  /// Get the spelling offset of the first byte of the ud-suffix.
240  unsigned getUDSuffixOffset() const {
241    assert(!UDSuffixBuf.empty() && "no ud-suffix");
242    return UDSuffixOffset;
243  }
244
245private:
246  void init(const Token *StringToks, unsigned NumStringToks);
247  bool CopyStringFragment(const Token &Tok, const char *TokBegin,
248                          StringRef Fragment);
249  void DiagnoseLexingError(SourceLocation Loc);
250};
251
252}  // end namespace clang
253
254#endif
255