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 LLVM_CLANG_LEX_LITERALSUPPORT_H
16#define LLVM_CLANG_LEX_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/ArrayRef.h"
23#include "llvm/ADT/SmallString.h"
24#include "llvm/ADT/StringRef.h"
25#include "llvm/Support/DataTypes.h"
26
27namespace clang {
28
29class DiagnosticsEngine;
30class Preprocessor;
31class Token;
32class SourceLocation;
33class TargetInfo;
34class SourceManager;
35class LangOptions;
36
37/// Copy characters from Input to Buf, expanding any UCNs.
38void expandUCNs(SmallVectorImpl<char> &Buf, StringRef Input);
39
40/// NumericLiteralParser - This performs strict semantic analysis of the content
41/// of a ppnumber, classifying it as either integer, floating, or erroneous,
42/// determines the radix of the value and can convert it to a useful value.
43class NumericLiteralParser {
44  Preprocessor &PP; // needed for diagnostics
45
46  const char *const ThisTokBegin;
47  const char *const ThisTokEnd;
48  const char *DigitsBegin, *SuffixBegin; // markers
49  const char *s; // cursor
50
51  unsigned radix;
52
53  bool saw_exponent, saw_period, saw_ud_suffix;
54
55  SmallString<32> UDSuffixBuf;
56
57public:
58  NumericLiteralParser(StringRef TokSpelling,
59                       SourceLocation TokLoc,
60                       Preprocessor &PP);
61  bool hadError : 1;
62  bool isUnsigned : 1;
63  bool isLong : 1;          // This is *not* set for long long.
64  bool isLongLong : 1;
65  bool isHalf : 1;          // 1.0h
66  bool isFloat : 1;         // 1.0f
67  bool isImaginary : 1;     // 1.0i
68  bool isFloat128 : 1;      // 1.0q
69  uint8_t MicrosoftInteger; // Microsoft suffix extension i8, i16, i32, or i64.
70
71  bool isIntegerLiteral() const {
72    return !saw_period && !saw_exponent;
73  }
74  bool isFloatingLiteral() const {
75    return saw_period || saw_exponent;
76  }
77
78  bool hasUDSuffix() const {
79    return saw_ud_suffix;
80  }
81  StringRef getUDSuffix() const {
82    assert(saw_ud_suffix);
83    return UDSuffixBuf;
84  }
85  unsigned getUDSuffixOffset() const {
86    assert(saw_ud_suffix);
87    return SuffixBegin - ThisTokBegin;
88  }
89
90  static bool isValidUDSuffix(const LangOptions &LangOpts, StringRef Suffix);
91
92  unsigned getRadix() const { return radix; }
93
94  /// GetIntegerValue - Convert this numeric literal value to an APInt that
95  /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
96  /// value read is larger than the APInt's bits will hold), set Val to the low
97  /// bits of the result and return true.  Otherwise, return false.
98  bool GetIntegerValue(llvm::APInt &Val);
99
100  /// GetFloatValue - Convert this numeric literal to a floating value, using
101  /// the specified APFloat fltSemantics (specifying float, double, etc).
102  /// The optional bool isExact (passed-by-reference) has its value
103  /// set to true if the returned APFloat can represent the number in the
104  /// literal exactly, and false otherwise.
105  llvm::APFloat::opStatus GetFloatValue(llvm::APFloat &Result);
106
107private:
108
109  void ParseNumberStartingWithZero(SourceLocation TokLoc);
110  void ParseDecimalOrOctalCommon(SourceLocation TokLoc);
111
112  static bool isDigitSeparator(char C) { return C == '\''; }
113
114  /// \brief Determine whether the sequence of characters [Start, End) contains
115  /// any real digits (not digit separators).
116  bool containsDigits(const char *Start, const char *End) {
117    return Start != End && (Start + 1 != End || !isDigitSeparator(Start[0]));
118  }
119
120  enum CheckSeparatorKind { CSK_BeforeDigits, CSK_AfterDigits };
121
122  /// \brief Ensure that we don't have a digit separator here.
123  void checkSeparator(SourceLocation TokLoc, const char *Pos,
124                      CheckSeparatorKind IsAfterDigits);
125
126  /// SkipHexDigits - Read and skip over any hex digits, up to End.
127  /// Return a pointer to the first non-hex digit or End.
128  const char *SkipHexDigits(const char *ptr) {
129    while (ptr != ThisTokEnd && (isHexDigit(*ptr) || isDigitSeparator(*ptr)))
130      ptr++;
131    return ptr;
132  }
133
134  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
135  /// Return a pointer to the first non-hex digit or End.
136  const char *SkipOctalDigits(const char *ptr) {
137    while (ptr != ThisTokEnd &&
138           ((*ptr >= '0' && *ptr <= '7') || isDigitSeparator(*ptr)))
139      ptr++;
140    return ptr;
141  }
142
143  /// SkipDigits - Read and skip over any digits, up to End.
144  /// Return a pointer to the first non-hex digit or End.
145  const char *SkipDigits(const char *ptr) {
146    while (ptr != ThisTokEnd && (isDigit(*ptr) || isDigitSeparator(*ptr)))
147      ptr++;
148    return ptr;
149  }
150
151  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
152  /// Return a pointer to the first non-binary digit or End.
153  const char *SkipBinaryDigits(const char *ptr) {
154    while (ptr != ThisTokEnd &&
155           (*ptr == '0' || *ptr == '1' || isDigitSeparator(*ptr)))
156      ptr++;
157    return ptr;
158  }
159
160};
161
162/// CharLiteralParser - Perform interpretation and semantic analysis of a
163/// character literal.
164class CharLiteralParser {
165  uint64_t Value;
166  tok::TokenKind Kind;
167  bool IsMultiChar;
168  bool HadError;
169  SmallString<32> UDSuffixBuf;
170  unsigned UDSuffixOffset;
171public:
172  CharLiteralParser(const char *begin, const char *end,
173                    SourceLocation Loc, Preprocessor &PP,
174                    tok::TokenKind kind);
175
176  bool hadError() const { return HadError; }
177  bool isAscii() const { return Kind == tok::char_constant; }
178  bool isWide() const { return Kind == tok::wide_char_constant; }
179  bool isUTF8() const { return Kind == tok::utf8_char_constant; }
180  bool isUTF16() const { return Kind == tok::utf16_char_constant; }
181  bool isUTF32() const { return Kind == tok::utf32_char_constant; }
182  bool isMultiChar() const { return IsMultiChar; }
183  uint64_t getValue() const { return Value; }
184  StringRef getUDSuffix() const { return UDSuffixBuf; }
185  unsigned getUDSuffixOffset() const {
186    assert(!UDSuffixBuf.empty() && "no ud-suffix");
187    return UDSuffixOffset;
188  }
189};
190
191/// StringLiteralParser - This decodes string escape characters and performs
192/// wide string analysis and Translation Phase #6 (concatenation of string
193/// literals) (C99 5.1.1.2p1).
194class StringLiteralParser {
195  const SourceManager &SM;
196  const LangOptions &Features;
197  const TargetInfo &Target;
198  DiagnosticsEngine *Diags;
199
200  unsigned MaxTokenLength;
201  unsigned SizeBound;
202  unsigned CharByteWidth;
203  tok::TokenKind Kind;
204  SmallString<512> ResultBuf;
205  char *ResultPtr; // cursor
206  SmallString<32> UDSuffixBuf;
207  unsigned UDSuffixToken;
208  unsigned UDSuffixOffset;
209public:
210  StringLiteralParser(ArrayRef<Token> StringToks,
211                      Preprocessor &PP, bool Complain = true);
212  StringLiteralParser(ArrayRef<Token> StringToks,
213                      const SourceManager &sm, const LangOptions &features,
214                      const TargetInfo &target,
215                      DiagnosticsEngine *diags = nullptr)
216    : SM(sm), Features(features), Target(target), Diags(diags),
217      MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
218      ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
219    init(StringToks);
220  }
221
222
223  bool hadError;
224  bool Pascal;
225
226  StringRef GetString() const {
227    return StringRef(ResultBuf.data(), GetStringLength());
228  }
229  unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
230
231  unsigned GetNumStringChars() const {
232    return GetStringLength() / CharByteWidth;
233  }
234  /// getOffsetOfStringByte - This function returns the offset of the
235  /// specified byte of the string data represented by Token.  This handles
236  /// advancing over escape sequences in the string.
237  ///
238  /// If the Diagnostics pointer is non-null, then this will do semantic
239  /// checking of the string literal and emit errors and warnings.
240  unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
241
242  bool isAscii() const { return Kind == tok::string_literal; }
243  bool isWide() const { return Kind == tok::wide_string_literal; }
244  bool isUTF8() const { return Kind == tok::utf8_string_literal; }
245  bool isUTF16() const { return Kind == tok::utf16_string_literal; }
246  bool isUTF32() const { return Kind == tok::utf32_string_literal; }
247  bool isPascal() const { return Pascal; }
248
249  StringRef getUDSuffix() const { return UDSuffixBuf; }
250
251  /// Get the index of a token containing a ud-suffix.
252  unsigned getUDSuffixToken() const {
253    assert(!UDSuffixBuf.empty() && "no ud-suffix");
254    return UDSuffixToken;
255  }
256  /// Get the spelling offset of the first byte of the ud-suffix.
257  unsigned getUDSuffixOffset() const {
258    assert(!UDSuffixBuf.empty() && "no ud-suffix");
259    return UDSuffixOffset;
260  }
261
262private:
263  void init(ArrayRef<Token> StringToks);
264  bool CopyStringFragment(const Token &Tok, const char *TokBegin,
265                          StringRef Fragment);
266  void DiagnoseLexingError(SourceLocation Loc);
267};
268
269}  // end namespace clang
270
271#endif
272