LiteralSupport.h revision 3144749f8bf9bbf7c027f2161a930bff80ad6f72 
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/LLVM.h"
19#include "llvm/ADT/APFloat.h"
20#include "llvm/ADT/SmallString.h"
21#include "llvm/Support/DataTypes.h"
22#include "clang/Basic/TokenKinds.h"
23#include <cctype>
24
25namespace clang {
26
27class DiagnosticsEngine;
28class Preprocessor;
29class Token;
30class SourceLocation;
31class TargetInfo;
32class SourceManager;
33class LangOptions;
34
35/// NumericLiteralParser - This performs strict semantic analysis of the content
36/// of a ppnumber, classifying it as either integer, floating, or erroneous,
37/// determines the radix of the value and can convert it to a useful value.
38class NumericLiteralParser {
39  Preprocessor &PP; // needed for diagnostics
40
41  const char *const ThisTokBegin;
42  const char *const ThisTokEnd;
43  const char *DigitsBegin, *SuffixBegin; // markers
44  const char *s; // cursor
45
46  unsigned radix;
47
48  bool saw_exponent, saw_period, saw_ud_suffix;
49
50public:
51  NumericLiteralParser(const char *begin, const char *end,
52                       SourceLocation Loc, Preprocessor &PP);
53  bool hadError;
54  bool isUnsigned;
55  bool isLong;        // This is *not* set for long long.
56  bool isLongLong;
57  bool isFloat;       // 1.0f
58  bool isImaginary;   // 1.0i
59  bool isMicrosoftInteger;  // Microsoft suffix extension i8, i16, i32, or i64.
60
61  bool isIntegerLiteral() const {
62    return !saw_period && !saw_exponent;
63  }
64  bool isFloatingLiteral() const {
65    return saw_period || saw_exponent;
66  }
67
68  bool hasUDSuffix() const {
69    return saw_ud_suffix;
70  }
71  StringRef getUDSuffix() const {
72    assert(saw_ud_suffix);
73    return StringRef(SuffixBegin, ThisTokEnd - SuffixBegin);
74  }
75  unsigned getUDSuffixOffset() const {
76    assert(saw_ud_suffix);
77    return SuffixBegin - ThisTokBegin;
78  }
79
80  unsigned getRadix() const { return radix; }
81
82  /// GetIntegerValue - Convert this numeric literal value to an APInt that
83  /// matches Val's input width.  If there is an overflow (i.e., if the unsigned
84  /// value read is larger than the APInt's bits will hold), set Val to the low
85  /// bits of the result and return true.  Otherwise, return false.
86  bool GetIntegerValue(llvm::APInt &Val);
87
88  /// GetFloatValue - Convert this numeric literal to a floating value, using
89  /// the specified APFloat fltSemantics (specifying float, double, etc).
90  /// The optional bool isExact (passed-by-reference) has its value
91  /// set to true if the returned APFloat can represent the number in the
92  /// literal exactly, and false otherwise.
93  llvm::APFloat::opStatus GetFloatValue(llvm::APFloat &Result);
94
95private:
96
97  void ParseNumberStartingWithZero(SourceLocation TokLoc);
98
99  /// SkipHexDigits - Read and skip over any hex digits, up to End.
100  /// Return a pointer to the first non-hex digit or End.
101  const char *SkipHexDigits(const char *ptr) {
102    while (ptr != ThisTokEnd && isxdigit(*ptr))
103      ptr++;
104    return ptr;
105  }
106
107  /// SkipOctalDigits - Read and skip over any octal digits, up to End.
108  /// Return a pointer to the first non-hex digit or End.
109  const char *SkipOctalDigits(const char *ptr) {
110    while (ptr != ThisTokEnd && ((*ptr >= '0') && (*ptr <= '7')))
111      ptr++;
112    return ptr;
113  }
114
115  /// SkipDigits - Read and skip over any digits, up to End.
116  /// Return a pointer to the first non-hex digit or End.
117  const char *SkipDigits(const char *ptr) {
118    while (ptr != ThisTokEnd && isdigit(*ptr))
119      ptr++;
120    return ptr;
121  }
122
123  /// SkipBinaryDigits - Read and skip over any binary digits, up to End.
124  /// Return a pointer to the first non-binary digit or End.
125  const char *SkipBinaryDigits(const char *ptr) {
126    while (ptr != ThisTokEnd && (*ptr == '0' || *ptr == '1'))
127      ptr++;
128    return ptr;
129  }
130
131};
132
133/// CharLiteralParser - Perform interpretation and semantic analysis of a
134/// character literal.
135class CharLiteralParser {
136  uint64_t Value;
137  tok::TokenKind Kind;
138  bool IsMultiChar;
139  bool HadError;
140  SmallString<32> UDSuffixBuf;
141  unsigned UDSuffixOffset;
142public:
143  CharLiteralParser(const char *begin, const char *end,
144                    SourceLocation Loc, Preprocessor &PP,
145                    tok::TokenKind kind);
146
147  bool hadError() const { return HadError; }
148  bool isAscii() const { return Kind == tok::char_constant; }
149  bool isWide() const { return Kind == tok::wide_char_constant; }
150  bool isUTF16() const { return Kind == tok::utf16_char_constant; }
151  bool isUTF32() const { return Kind == tok::utf32_char_constant; }
152  bool isMultiChar() const { return IsMultiChar; }
153  uint64_t getValue() const { return Value; }
154  StringRef getUDSuffix() const { return UDSuffixBuf; }
155  unsigned getUDSuffixOffset() const {
156    assert(!UDSuffixBuf.empty() && "no ud-suffix");
157    return UDSuffixOffset;
158  }
159};
160
161/// StringLiteralParser - This decodes string escape characters and performs
162/// wide string analysis and Translation Phase #6 (concatenation of string
163/// literals) (C99 5.1.1.2p1).
164class StringLiteralParser {
165  const SourceManager &SM;
166  const LangOptions &Features;
167  const TargetInfo &Target;
168  DiagnosticsEngine *Diags;
169
170  unsigned MaxTokenLength;
171  unsigned SizeBound;
172  unsigned CharByteWidth;
173  tok::TokenKind Kind;
174  SmallString<512> ResultBuf;
175  char *ResultPtr; // cursor
176  SmallString<32> UDSuffixBuf;
177  unsigned UDSuffixToken;
178  unsigned UDSuffixOffset;
179public:
180  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
181                      Preprocessor &PP, bool Complain = true);
182  StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
183                      const SourceManager &sm, const LangOptions &features,
184                      const TargetInfo &target, DiagnosticsEngine *diags = 0)
185    : SM(sm), Features(features), Target(target), Diags(diags),
186      MaxTokenLength(0), SizeBound(0), CharByteWidth(0), Kind(tok::unknown),
187      ResultPtr(ResultBuf.data()), hadError(false), Pascal(false) {
188    init(StringToks, NumStringToks);
189  }
190
191
192  bool hadError;
193  bool Pascal;
194
195  StringRef GetString() const {
196    return StringRef(ResultBuf.data(), GetStringLength());
197  }
198  unsigned GetStringLength() const { return ResultPtr-ResultBuf.data(); }
199
200  unsigned GetNumStringChars() const {
201    return GetStringLength() / CharByteWidth;
202  }
203  /// getOffsetOfStringByte - This function returns the offset of the
204  /// specified byte of the string data represented by Token.  This handles
205  /// advancing over escape sequences in the string.
206  ///
207  /// If the Diagnostics pointer is non-null, then this will do semantic
208  /// checking of the string literal and emit errors and warnings.
209  unsigned getOffsetOfStringByte(const Token &TheTok, unsigned ByteNo) const;
210
211  bool isAscii() const { return Kind == tok::string_literal; }
212  bool isWide() const { return Kind == tok::wide_string_literal; }
213  bool isUTF8() const { return Kind == tok::utf8_string_literal; }
214  bool isUTF16() const { return Kind == tok::utf16_string_literal; }
215  bool isUTF32() const { return Kind == tok::utf32_string_literal; }
216  bool isPascal() const { return Pascal; }
217
218  StringRef getUDSuffix() const { return UDSuffixBuf; }
219
220  /// Get the index of a token containing a ud-suffix.
221  unsigned getUDSuffixToken() const {
222    assert(!UDSuffixBuf.empty() && "no ud-suffix");
223    return UDSuffixToken;
224  }
225  /// Get the spelling offset of the first byte of the ud-suffix.
226  unsigned getUDSuffixOffset() const {
227    assert(!UDSuffixBuf.empty() && "no ud-suffix");
228    return UDSuffixOffset;
229  }
230
231private:
232  void init(const Token *StringToks, unsigned NumStringToks);
233  bool CopyStringFragment(StringRef Fragment);
234  bool DiagnoseBadString(const Token& Tok);
235  void DiagnoseLexingError(SourceLocation Loc);
236};
237
238}  // end namespace clang
239
240#endif
241