LiteralSupport.h revision 4ac537b0f07f2efb9fcf081f60d54e6cfb1cf2d5 
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  /// SkipHexDigits - Read and skip over any hex digits, up to End.
104  /// Return a pointer to the first non-hex digit or End.
105  const char *SkipHexDigits(const char *ptr) {
106    while (ptr != ThisTokEnd && isHexDigit(*ptr))
107      ptr++;
108    return ptr;
109  }
110
111  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
112  /// Return a pointer to the first non-hex digit or End.
113  const char *SkipOctalDigits(const char *ptr) {
114    while (ptr != ThisTokEnd && ((*ptr >= '0') && (*ptr <= '7')))
115      ptr++;
116    return ptr;
117  }
118
119  /// SkipDigits - Read and skip over any digits, up to End.
120  /// Return a pointer to the first non-hex digit or End.
121  const char *SkipDigits(const char *ptr) {
122    while (ptr != ThisTokEnd && isDigit(*ptr))
123      ptr++;
124    return ptr;
125  }
126
127  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
128  /// Return a pointer to the first non-binary digit or End.
129  const char *SkipBinaryDigits(const char *ptr) {
130    while (ptr != ThisTokEnd && (*ptr == '0' || *ptr == '1'))
131      ptr++;
132    return ptr;
133  }
134
135};
136
137/// CharLiteralParser - Perform interpretation and semantic analysis of a
138/// character literal.
139class CharLiteralParser {
140  uint64_t Value;
141  tok::TokenKind Kind;
142  bool IsMultiChar;
143  bool HadError;
144  SmallString<32> UDSuffixBuf;
145  unsigned UDSuffixOffset;
146public:
147  CharLiteralParser(const char *begin, const char *end,
148                    SourceLocation Loc, Preprocessor &PP,
149                    tok::TokenKind kind);
150
151  bool hadError() const { return HadError; }
152  bool isAscii() const { return Kind == tok::char_constant; }
153  bool isWide() const { return Kind == tok::wide_char_constant; }
154  bool isUTF16() const { return Kind == tok::utf16_char_constant; }
155  bool isUTF32() const { return Kind == tok::utf32_char_constant; }
156  bool isMultiChar() const { return IsMultiChar; }
157  uint64_t getValue() const { return Value; }
158  StringRef getUDSuffix() const { return UDSuffixBuf; }
159  unsigned getUDSuffixOffset() const {
160    assert(!UDSuffixBuf.empty() && "no ud-suffix");
161    return UDSuffixOffset;
162  }
163};
164
165/// StringLiteralParser - This decodes string escape characters and performs
166/// wide string analysis and Translation Phase #6 (concatenation of string
167/// literals) (C99 5.1.1.2p1).
168class StringLiteralParser {
169  const SourceManager &SM;
170  const LangOptions &Features;
171  const TargetInfo &Target;
172  DiagnosticsEngine *Diags;
173
174  unsigned MaxTokenLength;
175  unsigned SizeBound;
176  unsigned CharByteWidth;
177  tok::TokenKind Kind;
178  SmallString<512> ResultBuf;
179  char *ResultPtr; // cursor
180  SmallString<32> UDSuffixBuf;
181  unsigned UDSuffixToken;
182  unsigned UDSuffixOffset;
183public:
184  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
185                      Preprocessor &PP, bool Complain = true);
186  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
187                      const SourceManager &sm, const LangOptions &features,
188                      const TargetInfo &target, DiagnosticsEngine *diags = 0)
189    : SM(sm), Features(features), Target(target), Diags(diags),
190      MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
191      ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
192    init(StringToks, NumStringToks);
193  }
194
195
196  bool hadError;
197  bool Pascal;
198
199  StringRef GetString() const {
200    return StringRef(ResultBuf.data(), GetStringLength());
201  }
202  unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
203
204  unsigned GetNumStringChars() const {
205    return GetStringLength() / CharByteWidth;
206  }
207  /// getOffsetOfStringByte - This function returns the offset of the
208  /// specified byte of the string data represented by Token.  This handles
209  /// advancing over escape sequences in the string.
210  ///
211  /// If the Diagnostics pointer is non-null, then this will do semantic
212  /// checking of the string literal and emit errors and warnings.
213  unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
214
215  bool isAscii() const { return Kind == tok::string_literal; }
216  bool isWide() const { return Kind == tok::wide_string_literal; }
217  bool isUTF8() const { return Kind == tok::utf8_string_literal; }
218  bool isUTF16() const { return Kind == tok::utf16_string_literal; }
219  bool isUTF32() const { return Kind == tok::utf32_string_literal; }
220  bool isPascal() const { return Pascal; }
221
222  StringRef getUDSuffix() const { return UDSuffixBuf; }
223
224  /// Get the index of a token containing a ud-suffix.
225  unsigned getUDSuffixToken() const {
226    assert(!UDSuffixBuf.empty() && "no ud-suffix");
227    return UDSuffixToken;
228  }
229  /// Get the spelling offset of the first byte of the ud-suffix.
230  unsigned getUDSuffixOffset() const {
231    assert(!UDSuffixBuf.empty() && "no ud-suffix");
232    return UDSuffixOffset;
233  }
234
235private:
236  void init(const Token *StringToks, unsigned NumStringToks);
237  bool CopyStringFragment(const Token &Tok, const char *TokBegin,
238                          StringRef Fragment);
239  void DiagnoseLexingError(SourceLocation Loc);
240};
241
242}  // end namespace clang
243
244#endif
245