1 //===--- LiteralSupport.h ---------------------------------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file defines the NumericLiteralParser, CharLiteralParser, and
10 // StringLiteralParser interfaces.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #ifndef LLVM_CLANG_LEX_LITERALSUPPORT_H
15 #define LLVM_CLANG_LEX_LITERALSUPPORT_H
16 
17 #include "clang/Basic/CharInfo.h"
18 #include "clang/Basic/LLVM.h"
19 #include "clang/Basic/TokenKinds.h"
20 #include "llvm/ADT/APFloat.h"
21 #include "llvm/ADT/ArrayRef.h"
22 #include "llvm/ADT/SmallString.h"
23 #include "llvm/ADT/StringRef.h"
24 #include "llvm/Support/DataTypes.h"
25 
26 namespace clang {
27 
28 class DiagnosticsEngine;
29 class Preprocessor;
30 class Token;
31 class SourceLocation;
32 class TargetInfo;
33 class SourceManager;
34 class LangOptions;
35 
36 /// Copy characters from Input to Buf, expanding any UCNs.
37 void expandUCNs(SmallVectorImpl<char> &Buf, StringRef Input);
38 
39 /// NumericLiteralParser - This performs strict semantic analysis of the content
40 /// of a ppnumber, classifying it as either integer, floating, or erroneous,
41 /// determines the radix of the value and can convert it to a useful value.
42 class NumericLiteralParser {
43   const SourceManager &SM;
44   const LangOptions &LangOpts;
45   DiagnosticsEngine &Diags;
46 
47   const char *const ThisTokBegin;
48   const char *const ThisTokEnd;
49   const char *DigitsBegin, *SuffixBegin; // markers
50   const char *s; // cursor
51 
52   unsigned radix;
53 
54   bool saw_exponent, saw_period, saw_ud_suffix, saw_fixed_point_suffix;
55 
56   SmallString<32> UDSuffixBuf;
57 
58 public:
59   NumericLiteralParser(StringRef TokSpelling, SourceLocation TokLoc,
60                        const SourceManager &SM, const LangOptions &LangOpts,
61                        const TargetInfo &Target, DiagnosticsEngine &Diags);
62   bool hadError : 1;
63   bool isUnsigned : 1;
64   bool isLong : 1;          // This is *not* set for long long.
65   bool isLongLong : 1;
66   bool isSizeT : 1;         // 1z, 1uz (C++2b)
67   bool isHalf : 1;          // 1.0h
68   bool isFloat : 1;         // 1.0f
69   bool isImaginary : 1;     // 1.0i
70   bool isFloat16 : 1;       // 1.0f16
71   bool isFloat128 : 1;      // 1.0q
72   bool isFract : 1;         // 1.0hr/r/lr/uhr/ur/ulr
73   bool isAccum : 1;         // 1.0hk/k/lk/uhk/uk/ulk
74   bool isBitInt : 1;        // 1wb, 1uwb (C2x)
75   uint8_t MicrosoftInteger; // Microsoft suffix extension i8, i16, i32, or i64.
76 
77 
78   bool isFixedPointLiteral() const {
79     return (saw_period || saw_exponent) && saw_fixed_point_suffix;
80   }
81 
82   bool isIntegerLiteral() const {
83     return !saw_period && !saw_exponent && !isFixedPointLiteral();
84   }
85   bool isFloatingLiteral() const {
86     return (saw_period || saw_exponent) && !isFixedPointLiteral();
87   }
88 
89   bool hasUDSuffix() const {
90     return saw_ud_suffix;
91   }
92   StringRef getUDSuffix() const {
93     assert(saw_ud_suffix);
94     return UDSuffixBuf;
95   }
96   unsigned getUDSuffixOffset() const {
97     assert(saw_ud_suffix);
98     return SuffixBegin - ThisTokBegin;
99   }
100 
101   static bool isValidUDSuffix(const LangOptions &LangOpts, StringRef Suffix);
102 
103   unsigned getRadix() const { return radix; }
104 
105   /// GetIntegerValue - Convert this numeric literal value to an APInt that
106   /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
107   /// value read is larger than the APInt's bits will hold), set Val to the low
108   /// bits of the result and return true.  Otherwise, return false.
109   bool GetIntegerValue(llvm::APInt &Val);
110 
111   /// GetFloatValue - Convert this numeric literal to a floating value, using
112   /// the specified APFloat fltSemantics (specifying float, double, etc).
113   /// The optional bool isExact (passed-by-reference) has its value
114   /// set to true if the returned APFloat can represent the number in the
115   /// literal exactly, and false otherwise.
116   llvm::APFloat::opStatus GetFloatValue(llvm::APFloat &Result);
117 
118   /// GetFixedPointValue - Convert this numeric literal value into a
119   /// scaled integer that represents this value. Returns true if an overflow
120   /// occurred when calculating the integral part of the scaled integer or
121   /// calculating the digit sequence of the exponent.
122   bool GetFixedPointValue(llvm::APInt &StoreVal, unsigned Scale);
123 
124   /// Get the digits that comprise the literal. This excludes any prefix or
125   /// suffix associated with the literal.
126   StringRef getLiteralDigits() const {
127     assert(!hadError && "cannot reliably get the literal digits with an error");
128     return StringRef(DigitsBegin, SuffixBegin - DigitsBegin);
129   }
130 
131 private:
132 
133   void ParseNumberStartingWithZero(SourceLocation TokLoc);
134   void ParseDecimalOrOctalCommon(SourceLocation TokLoc);
135 
136   static bool isDigitSeparator(char C) { return C == '\''; }
137 
138   /// Determine whether the sequence of characters [Start, End) contains
139   /// any real digits (not digit separators).
140   bool containsDigits(const char *Start, const char *End) {
141     return Start != End && (Start + 1 != End || !isDigitSeparator(Start[0]));
142   }
143 
144   enum CheckSeparatorKind { CSK_BeforeDigits, CSK_AfterDigits };
145 
146   /// Ensure that we don't have a digit separator here.
147   void checkSeparator(SourceLocation TokLoc, const char *Pos,
148                       CheckSeparatorKind IsAfterDigits);
149 
150   /// SkipHexDigits - Read and skip over any hex digits, up to End.
151   /// Return a pointer to the first non-hex digit or End.
152   const char *SkipHexDigits(const char *ptr) {
153     while (ptr != ThisTokEnd && (isHexDigit(*ptr) || isDigitSeparator(*ptr)))
154       ptr++;
155     return ptr;
156   }
157 
158   /// SkipOctalDigits - Read and skip over any octal digits, up to End.
159   /// Return a pointer to the first non-hex digit or End.
160   const char *SkipOctalDigits(const char *ptr) {
161     while (ptr != ThisTokEnd &&
162            ((*ptr >= '0' && *ptr <= '7') || isDigitSeparator(*ptr)))
163       ptr++;
164     return ptr;
165   }
166 
167   /// SkipDigits - Read and skip over any digits, up to End.
168   /// Return a pointer to the first non-hex digit or End.
169   const char *SkipDigits(const char *ptr) {
170     while (ptr != ThisTokEnd && (isDigit(*ptr) || isDigitSeparator(*ptr)))
171       ptr++;
172     return ptr;
173   }
174 
175   /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
176   /// Return a pointer to the first non-binary digit or End.
177   const char *SkipBinaryDigits(const char *ptr) {
178     while (ptr != ThisTokEnd &&
179            (*ptr == '0' || *ptr == '1' || isDigitSeparator(*ptr)))
180       ptr++;
181     return ptr;
182   }
183 
184 };
185 
186 /// CharLiteralParser - Perform interpretation and semantic analysis of a
187 /// character literal.
188 class CharLiteralParser {
189   uint64_t Value;
190   tok::TokenKind Kind;
191   bool IsMultiChar;
192   bool HadError;
193   SmallString<32> UDSuffixBuf;
194   unsigned UDSuffixOffset;
195 public:
196   CharLiteralParser(const char *begin, const char *end,
197                     SourceLocation Loc, Preprocessor &PP,
198                     tok::TokenKind kind);
199 
200   bool hadError() const { return HadError; }
201   bool isOrdinary() const { return Kind == tok::char_constant; }
202   bool isWide() const { return Kind == tok::wide_char_constant; }
203   bool isUTF8() const { return Kind == tok::utf8_char_constant; }
204   bool isUTF16() const { return Kind == tok::utf16_char_constant; }
205   bool isUTF32() const { return Kind == tok::utf32_char_constant; }
206   bool isMultiChar() const { return IsMultiChar; }
207   uint64_t getValue() const { return Value; }
208   StringRef getUDSuffix() const { return UDSuffixBuf; }
209   unsigned getUDSuffixOffset() const {
210     assert(!UDSuffixBuf.empty() && "no ud-suffix");
211     return UDSuffixOffset;
212   }
213 };
214 
215 /// StringLiteralParser - This decodes string escape characters and performs
216 /// wide string analysis and Translation Phase #6 (concatenation of string
217 /// literals) (C99 5.1.1.2p1).
218 class StringLiteralParser {
219   const SourceManager &SM;
220   const LangOptions &Features;
221   const TargetInfo &Target;
222   DiagnosticsEngine *Diags;
223 
224   unsigned MaxTokenLength;
225   unsigned SizeBound;
226   unsigned CharByteWidth;
227   tok::TokenKind Kind;
228   SmallString<512> ResultBuf;
229   char *ResultPtr; // cursor
230   SmallString<32> UDSuffixBuf;
231   unsigned UDSuffixToken;
232   unsigned UDSuffixOffset;
233 public:
234   StringLiteralParser(ArrayRef<Token> StringToks,
235                       Preprocessor &PP);
236   StringLiteralParser(ArrayRef<Token> StringToks,
237                       const SourceManager &sm, const LangOptions &features,
238                       const TargetInfo &target,
239                       DiagnosticsEngine *diags = nullptr)
240     : SM(sm), Features(features), Target(target), Diags(diags),
241       MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
242       ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
243     init(StringToks);
244   }
245 
246 
247   bool hadError;
248   bool Pascal;
249 
250   StringRef GetString() const {
251     return StringRef(ResultBuf.data(), GetStringLength());
252   }
253   unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
254 
255   unsigned GetNumStringChars() const {
256     return GetStringLength() / CharByteWidth;
257   }
258   /// getOffsetOfStringByte - This function returns the offset of the
259   /// specified byte of the string data represented by Token.  This handles
260   /// advancing over escape sequences in the string.
261   ///
262   /// If the Diagnostics pointer is non-null, then this will do semantic
263   /// checking of the string literal and emit errors and warnings.
264   unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
265 
266   bool isOrdinary() const { return Kind == tok::string_literal; }
267   bool isWide() const { return Kind == tok::wide_string_literal; }
268   bool isUTF8() const { return Kind == tok::utf8_string_literal; }
269   bool isUTF16() const { return Kind == tok::utf16_string_literal; }
270   bool isUTF32() const { return Kind == tok::utf32_string_literal; }
271   bool isPascal() const { return Pascal; }
272 
273   StringRef getUDSuffix() const { return UDSuffixBuf; }
274 
275   /// Get the index of a token containing a ud-suffix.
276   unsigned getUDSuffixToken() const {
277     assert(!UDSuffixBuf.empty() && "no ud-suffix");
278     return UDSuffixToken;
279   }
280   /// Get the spelling offset of the first byte of the ud-suffix.
281   unsigned getUDSuffixOffset() const {
282     assert(!UDSuffixBuf.empty() && "no ud-suffix");
283     return UDSuffixOffset;
284   }
285 
286   static bool isValidUDSuffix(const LangOptions &LangOpts, StringRef Suffix);
287 
288 private:
289   void init(ArrayRef<Token> StringToks);
290   bool CopyStringFragment(const Token &Tok, const char *TokBegin,
291                           StringRef Fragment);
292   void DiagnoseLexingError(SourceLocation Loc);
293 };
294 
295 }  // end namespace clang
296 
297 #endif
298