Coverage Report

Created: 2018-07-19 20:53

/Users/buildslave/jenkins/workspace/clang-stage2-coverage-R/llvm/tools/clang/include/clang/Basic/CharInfo.h
Line
Count
Source
1
//===--- clang/Basic/CharInfo.h - Classifying ASCII Characters ------------===//
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
#ifndef LLVM_CLANG_BASIC_CHARINFO_H
11
#define LLVM_CLANG_BASIC_CHARINFO_H
12
13
#include "clang/Basic/LLVM.h"
14
#include "llvm/ADT/StringRef.h"
15
#include "llvm/Support/Compiler.h"
16
#include "llvm/Support/DataTypes.h"
17
18
namespace clang {
19
namespace charinfo {
20
  extern const uint16_t InfoTable[256];
21
22
  enum {
23
    CHAR_HORZ_WS  = 0x0001,  // '\t', '\f', '\v'.  Note, no '\0'
24
    CHAR_VERT_WS  = 0x0002,  // '\r', '\n'
25
    CHAR_SPACE    = 0x0004,  // ' '
26
    CHAR_DIGIT    = 0x0008,  // 0-9
27
    CHAR_XLETTER  = 0x0010,  // a-f,A-F
28
    CHAR_UPPER    = 0x0020,  // A-Z
29
    CHAR_LOWER    = 0x0040,  // a-z
30
    CHAR_UNDER    = 0x0080,  // _
31
    CHAR_PERIOD   = 0x0100,  // .
32
    CHAR_RAWDEL   = 0x0200,  // {}[]#<>%:;?*+-/^&|~!=,"'
33
    CHAR_PUNCT    = 0x0400   // `$@()
34
  };
35
36
  enum {
37
    CHAR_XUPPER = CHAR_XLETTER | CHAR_UPPER,
38
    CHAR_XLOWER = CHAR_XLETTER | CHAR_LOWER
39
  };
40
} // end namespace charinfo
41
42
/// Returns true if this is an ASCII character.
43
605M
LLVM_READNONE inline bool isASCII(char c) {
44
605M
  return static_cast<unsigned char>(c) <= 127;
45
605M
}
46
47
/// Returns true if this is a valid first character of a C identifier,
48
/// which is [a-zA-Z_].
49
LLVM_READONLY inline bool isIdentifierHead(unsigned char c,
50
3.09M
                                           bool AllowDollar = false) {
51
3.09M
  using namespace charinfo;
52
3.09M
  if (InfoTable[c] & (CHAR_UPPER|CHAR_LOWER|CHAR_UNDER))
53
4.48k
    return true;
54
3.08M
  return AllowDollar && 
c == '$'5
;
55
3.08M
}
56
57
/// Returns true if this is a body character of a C identifier,
58
/// which is [a-zA-Z0-9_].
59
LLVM_READONLY inline bool isIdentifierBody(unsigned char c,
60
9.83G
                                           bool AllowDollar = false) {
61
9.83G
  using namespace charinfo;
62
9.83G
  if (InfoTable[c] & (CHAR_UPPER|CHAR_LOWER|CHAR_DIGIT|CHAR_UNDER))
63
9.27G
    return true;
64
562M
  return AllowDollar && 
c == '$'3.25k
;
65
562M
}
66
67
/// Returns true if this character is horizontal ASCII whitespace:
68
/// ' ', '\\t', '\\f', '\\v'.
69
///
70
/// Note that this returns false for '\\0'.
71
1.23G
LLVM_READONLY inline bool isHorizontalWhitespace(unsigned char c) {
72
1.23G
  using namespace charinfo;
73
1.23G
  return (InfoTable[c] & (CHAR_HORZ_WS|CHAR_SPACE)) != 0;
74
1.23G
}
75
76
/// Returns true if this character is vertical ASCII whitespace: '\\n', '\\r'.
77
///
78
/// Note that this returns false for '\\0'.
79
411M
LLVM_READONLY inline bool isVerticalWhitespace(unsigned char c) {
80
411M
  using namespace charinfo;
81
411M
  return (InfoTable[c] & CHAR_VERT_WS) != 0;
82
411M
}
83
84
/// Return true if this character is horizontal or vertical ASCII whitespace:
85
/// ' ', '\\t', '\\f', '\\v', '\\n', '\\r'.
86
///
87
/// Note that this returns false for '\\0'.
88
8.91M
LLVM_READONLY inline bool isWhitespace(unsigned char c) {
89
8.91M
  using namespace charinfo;
90
8.91M
  return (InfoTable[c] & (CHAR_HORZ_WS|CHAR_VERT_WS|CHAR_SPACE)) != 0;
91
8.91M
}
92
93
/// Return true if this character is an ASCII digit: [0-9]
94
17.4M
LLVM_READONLY inline bool isDigit(unsigned char c) {
95
17.4M
  using namespace charinfo;
96
17.4M
  return (InfoTable[c] & CHAR_DIGIT) != 0;
97
17.4M
}
98
99
/// Return true if this character is a lowercase ASCII letter: [a-z]
100
38.4k
LLVM_READONLY inline bool isLowercase(unsigned char c) {
101
38.4k
  using namespace charinfo;
102
38.4k
  return (InfoTable[c] & CHAR_LOWER) != 0;
103
38.4k
}
104
105
/// Return true if this character is an uppercase ASCII letter: [A-Z]
106
25.5k
LLVM_READONLY inline bool isUppercase(unsigned char c) {
107
25.5k
  using namespace charinfo;
108
25.5k
  return (InfoTable[c] & CHAR_UPPER) != 0;
109
25.5k
}
110
111
/// Return true if this character is an ASCII letter: [a-zA-Z]
112
133k
LLVM_READONLY inline bool isLetter(unsigned char c) {
113
133k
  using namespace charinfo;
114
133k
  return (InfoTable[c] & (CHAR_UPPER|CHAR_LOWER)) != 0;
115
133k
}
116
117
/// Return true if this character is an ASCII letter or digit: [a-zA-Z0-9]
118
1.51M
LLVM_READONLY inline bool isAlphanumeric(unsigned char c) {
119
1.51M
  using namespace charinfo;
120
1.51M
  return (InfoTable[c] & (CHAR_DIGIT|CHAR_UPPER|CHAR_LOWER)) != 0;
121
1.51M
}
122
123
/// Return true if this character is an ASCII hex digit: [0-9a-fA-F]
124
2.48M
LLVM_READONLY inline bool isHexDigit(unsigned char c) {
125
2.48M
  using namespace charinfo;
126
2.48M
  return (InfoTable[c] & (CHAR_DIGIT|CHAR_XLETTER)) != 0;
127
2.48M
}
128
129
/// Return true if this character is an ASCII punctuation character.
130
///
131
/// Note that '_' is both a punctuation character and an identifier character!
132
390k
LLVM_READONLY inline bool isPunctuation(unsigned char c) {
133
390k
  using namespace charinfo;
134
390k
  return (InfoTable[c] & (CHAR_UNDER|CHAR_PERIOD|CHAR_RAWDEL|CHAR_PUNCT)) != 0;
135
390k
}
136
137
/// Return true if this character is an ASCII printable character; that is, a
138
/// character that should take exactly one column to print in a fixed-width
139
/// terminal.
140
7.60k
LLVM_READONLY inline bool isPrintable(unsigned char c) {
141
7.60k
  using namespace charinfo;
142
7.60k
  return (InfoTable[c] & (CHAR_UPPER|CHAR_LOWER|CHAR_PERIOD|CHAR_PUNCT|
143
7.60k
                          CHAR_DIGIT|CHAR_UNDER|CHAR_RAWDEL|CHAR_SPACE)) != 0;
144
7.60k
}
145
146
/// Return true if this is the body character of a C preprocessing number,
147
/// which is [a-zA-Z0-9_.].
148
136M
LLVM_READONLY inline bool isPreprocessingNumberBody(unsigned char c) {
149
136M
  using namespace charinfo;
150
136M
  return (InfoTable[c] &
151
136M
          (CHAR_UPPER|CHAR_LOWER|CHAR_DIGIT|CHAR_UNDER|CHAR_PERIOD)) != 0;
152
136M
}
153
154
/// Return true if this is the body character of a C++ raw string delimiter.
155
1.24k
LLVM_READONLY inline bool isRawStringDelimBody(unsigned char c) {
156
1.24k
  using namespace charinfo;
157
1.24k
  return (InfoTable[c] & (CHAR_UPPER|CHAR_LOWER|CHAR_PERIOD|
158
1.24k
                          CHAR_DIGIT|CHAR_UNDER|CHAR_RAWDEL)) != 0;
159
1.24k
}
160
161
162
/// Converts the given ASCII character to its lowercase equivalent.
163
///
164
/// If the character is not an uppercase character, it is returned as is.
165
529
LLVM_READONLY inline char toLowercase(char c) {
166
529
  if (isUppercase(c))
167
192
    return c + 'a' - 'A';
168
337
  return c;
169
337
}
170
171
/// Converts the given ASCII character to its uppercase equivalent.
172
///
173
/// If the character is not a lowercase character, it is returned as is.
174
26.3k
LLVM_READONLY inline char toUppercase(char c) {
175
26.3k
  if (isLowercase(c))
176
24.0k
    return c + 'A' - 'a';
177
2.26k
  return c;
178
2.26k
}
179
180
181
/// Return true if this is a valid ASCII identifier.
182
///
183
/// Note that this is a very simple check; it does not accept UCNs as valid
184
/// identifier characters.
185
LLVM_READONLY inline bool isValidIdentifier(StringRef S,
186
1.41k
                                            bool AllowDollar = false) {
187
1.41k
  if (S.empty() || 
!isIdentifierHead(S[0], AllowDollar)1.41k
)
188
21
    return false;
189
1.39k
190
11.2k
  
for (StringRef::iterator I = S.begin(), E = S.end(); 1.39k
I != E;
++I9.90k
)
191
9.94k
    if (!isIdentifierBody(*I, AllowDollar))
192
43
      return false;
193
1.39k
194
1.39k
  
return true1.34k
;
195
1.39k
}
196
197
} // end namespace clang
198
199
#endif