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 isFloat16 : 1;       // 1.0f16
69  bool isFloat128 : 1;      // 1.0q
70  uint8_t MicrosoftInteger; // Microsoft suffix extension i8, i16, i32, or i64.
71
72  bool isIntegerLiteral() const {
73    return !saw_period && !saw_exponent;
74  }
75  bool isFloatingLiteral() const {
76    return saw_period || saw_exponent;
77  }
78
79  bool hasUDSuffix() const {
80    return saw_ud_suffix;
81  }
82  StringRef getUDSuffix() const {
83    assert(saw_ud_suffix);
84    return UDSuffixBuf;
85  }
86  unsigned getUDSuffixOffset() const {
87    assert(saw_ud_suffix);
88    return SuffixBegin - ThisTokBegin;
89  }
90
91  static bool isValidUDSuffix(const LangOptions &LangOpts, StringRef Suffix);
92
93  unsigned getRadix() const { return radix; }
94
95  /// GetIntegerValue - Convert this numeric literal value to an APInt that
96  /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
97  /// value read is larger than the APInt's bits will hold), set Val to the low
98  /// bits of the result and return true.  Otherwise, return false.
99  bool GetIntegerValue(llvm::APInt &Val);
100
101  /// GetFloatValue - Convert this numeric literal to a floating value, using
102  /// the specified APFloat fltSemantics (specifying float, double, etc).
103  /// The optional bool isExact (passed-by-reference) has its value
104  /// set to true if the returned APFloat can represent the number in the
105  /// literal exactly, and false otherwise.
106  llvm::APFloat::opStatus GetFloatValue(llvm::APFloat &Result);
107
108private:
109
110  void ParseNumberStartingWithZero(SourceLocation TokLoc);
111  void ParseDecimalOrOctalCommon(SourceLocation TokLoc);
112
113  static bool isDigitSeparator(char C) { return C == '\''; }
114
115  /// \brief Determine whether the sequence of characters [Start, End) contains
116  /// any real digits (not digit separators).
117  bool containsDigits(const char *Start, const char *End) {
118    return Start != End && (Start + 1 != End || !isDigitSeparator(Start[0]));
119  }
120
121  enum CheckSeparatorKind { CSK_BeforeDigits, CSK_AfterDigits };
122
123  /// \brief Ensure that we don't have a digit separator here.
124  void checkSeparator(SourceLocation TokLoc, const char *Pos,
125                      CheckSeparatorKind IsAfterDigits);
126
127  /// SkipHexDigits - Read and skip over any hex digits, up to End.
128  /// Return a pointer to the first non-hex digit or End.
129  const char *SkipHexDigits(const char *ptr) {
130    while (ptr != ThisTokEnd && (isHexDigit(*ptr) || isDigitSeparator(*ptr)))
131      ptr++;
132    return ptr;
133  }
134
135  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
136  /// Return a pointer to the first non-hex digit or End.
137  const char *SkipOctalDigits(const char *ptr) {
138    while (ptr != ThisTokEnd &&
139           ((*ptr >= '0' && *ptr <= '7') || isDigitSeparator(*ptr)))
140      ptr++;
141    return ptr;
142  }
143
144  /// SkipDigits - Read and skip over any digits, up to End.
145  /// Return a pointer to the first non-hex digit or End.
146  const char *SkipDigits(const char *ptr) {
147    while (ptr != ThisTokEnd && (isDigit(*ptr) || isDigitSeparator(*ptr)))
148      ptr++;
149    return ptr;
150  }
151
152  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
153  /// Return a pointer to the first non-binary digit or End.
154  const char *SkipBinaryDigits(const char *ptr) {
155    while (ptr != ThisTokEnd &&
156           (*ptr == '0' || *ptr == '1' || isDigitSeparator(*ptr)))
157      ptr++;
158    return ptr;
159  }
160
161};
162
163/// CharLiteralParser - Perform interpretation and semantic analysis of a
164/// character literal.
165class CharLiteralParser {
166  uint64_t Value;
167  tok::TokenKind Kind;
168  bool IsMultiChar;
169  bool HadError;
170  SmallString<32> UDSuffixBuf;
171  unsigned UDSuffixOffset;
172public:
173  CharLiteralParser(const char *begin, const char *end,
174                    SourceLocation Loc, Preprocessor &PP,
175                    tok::TokenKind kind);
176
177  bool hadError() const { return HadError; }
178  bool isAscii() const { return Kind == tok::char_constant; }
179  bool isWide() const { return Kind == tok::wide_char_constant; }
180  bool isUTF8() const { return Kind == tok::utf8_char_constant; }
181  bool isUTF16() const { return Kind == tok::utf16_char_constant; }
182  bool isUTF32() const { return Kind == tok::utf32_char_constant; }
183  bool isMultiChar() const { return IsMultiChar; }
184  uint64_t getValue() const { return Value; }
185  StringRef getUDSuffix() const { return UDSuffixBuf; }
186  unsigned getUDSuffixOffset() const {
187    assert(!UDSuffixBuf.empty() && "no ud-suffix");
188    return UDSuffixOffset;
189  }
190};
191
192/// StringLiteralParser - This decodes string escape characters and performs
193/// wide string analysis and Translation Phase #6 (concatenation of string
194/// literals) (C99 5.1.1.2p1).
195class StringLiteralParser {
196  const SourceManager &SM;
197  const LangOptions &Features;
198  const TargetInfo &Target;
199  DiagnosticsEngine *Diags;
200
201  unsigned MaxTokenLength;
202  unsigned SizeBound;
203  unsigned CharByteWidth;
204  tok::TokenKind Kind;
205  SmallString<512> ResultBuf;
206  char *ResultPtr; // cursor
207  SmallString<32> UDSuffixBuf;
208  unsigned UDSuffixToken;
209  unsigned UDSuffixOffset;
210public:
211  StringLiteralParser(ArrayRef<Token> StringToks,
212                      Preprocessor &PP, bool Complain = true);
213  StringLiteralParser(ArrayRef<Token> StringToks,
214                      const SourceManager &sm, const LangOptions &features,
215                      const TargetInfo &target,
216                      DiagnosticsEngine *diags = nullptr)
217    : SM(sm), Features(features), Target(target), Diags(diags),
218      MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
219      ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
220    init(StringToks);
221  }
222
223
224  bool hadError;
225  bool Pascal;
226
227  StringRef GetString() const {
228    return StringRef(ResultBuf.data(), GetStringLength());
229  }
230  unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
231
232  unsigned GetNumStringChars() const {
233    return GetStringLength() / CharByteWidth;
234  }
235  /// getOffsetOfStringByte - This function returns the offset of the
236  /// specified byte of the string data represented by Token.  This handles
237  /// advancing over escape sequences in the string.
238  ///
239  /// If the Diagnostics pointer is non-null, then this will do semantic
240  /// checking of the string literal and emit errors and warnings.
241  unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
242
243  bool isAscii() const { return Kind == tok::string_literal; }
244  bool isWide() const { return Kind == tok::wide_string_literal; }
245  bool isUTF8() const { return Kind == tok::utf8_string_literal; }
246  bool isUTF16() const { return Kind == tok::utf16_string_literal; }
247  bool isUTF32() const { return Kind == tok::utf32_string_literal; }
248  bool isPascal() const { return Pascal; }
249
250  StringRef getUDSuffix() const { return UDSuffixBuf; }
251
252  /// Get the index of a token containing a ud-suffix.
253  unsigned getUDSuffixToken() const {
254    assert(!UDSuffixBuf.empty() && "no ud-suffix");
255    return UDSuffixToken;
256  }
257  /// Get the spelling offset of the first byte of the ud-suffix.
258  unsigned getUDSuffixOffset() const {
259    assert(!UDSuffixBuf.empty() && "no ud-suffix");
260    return UDSuffixOffset;
261  }
262
263  static bool isValidUDSuffix(const LangOptions &LangOpts, StringRef Suffix);
264
265private:
266  void init(ArrayRef<Token> StringToks);
267  bool CopyStringFragment(const Token &Tok, const char *TokBegin,
268                          StringRef Fragment);
269  void DiagnoseLexingError(SourceLocation Loc);
270};
271
272}  // end namespace clang
273
274#endif
275