annotate clang/lib/Lex/UnicodeCharSets.h @ 209:dd44ba33042e

merged...
author Shinji KONO <kono@ie.u-ryukyu.ac.jp>
date Tue, 08 Jun 2021 06:36:09 +0900
parents 1d019706d866
children c4bab56944e8
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
150
anatofuz
parents:
diff changeset
1 //===--- UnicodeCharSets.h - Contains important sets of characters --------===//
anatofuz
parents:
diff changeset
2 //
anatofuz
parents:
diff changeset
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
anatofuz
parents:
diff changeset
4 // See https://llvm.org/LICENSE.txt for license information.
anatofuz
parents:
diff changeset
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
anatofuz
parents:
diff changeset
6 //
anatofuz
parents:
diff changeset
7 //===----------------------------------------------------------------------===//
anatofuz
parents:
diff changeset
8 #ifndef LLVM_CLANG_LIB_LEX_UNICODECHARSETS_H
anatofuz
parents:
diff changeset
9 #define LLVM_CLANG_LIB_LEX_UNICODECHARSETS_H
anatofuz
parents:
diff changeset
10
anatofuz
parents:
diff changeset
11 #include "llvm/Support/UnicodeCharRanges.h"
anatofuz
parents:
diff changeset
12
anatofuz
parents:
diff changeset
13 // C11 D.1, C++11 [charname.allowed]
anatofuz
parents:
diff changeset
14 static const llvm::sys::UnicodeCharRange C11AllowedIDCharRanges[] = {
anatofuz
parents:
diff changeset
15 // 1
anatofuz
parents:
diff changeset
16 { 0x00A8, 0x00A8 }, { 0x00AA, 0x00AA }, { 0x00AD, 0x00AD },
anatofuz
parents:
diff changeset
17 { 0x00AF, 0x00AF }, { 0x00B2, 0x00B5 }, { 0x00B7, 0x00BA },
anatofuz
parents:
diff changeset
18 { 0x00BC, 0x00BE }, { 0x00C0, 0x00D6 }, { 0x00D8, 0x00F6 },
anatofuz
parents:
diff changeset
19 { 0x00F8, 0x00FF },
anatofuz
parents:
diff changeset
20 // 2
anatofuz
parents:
diff changeset
21 { 0x0100, 0x167F }, { 0x1681, 0x180D }, { 0x180F, 0x1FFF },
anatofuz
parents:
diff changeset
22 // 3
anatofuz
parents:
diff changeset
23 { 0x200B, 0x200D }, { 0x202A, 0x202E }, { 0x203F, 0x2040 },
anatofuz
parents:
diff changeset
24 { 0x2054, 0x2054 }, { 0x2060, 0x206F },
anatofuz
parents:
diff changeset
25 // 4
anatofuz
parents:
diff changeset
26 { 0x2070, 0x218F }, { 0x2460, 0x24FF }, { 0x2776, 0x2793 },
anatofuz
parents:
diff changeset
27 { 0x2C00, 0x2DFF }, { 0x2E80, 0x2FFF },
anatofuz
parents:
diff changeset
28 // 5
anatofuz
parents:
diff changeset
29 { 0x3004, 0x3007 }, { 0x3021, 0x302F }, { 0x3031, 0x303F },
anatofuz
parents:
diff changeset
30 // 6
anatofuz
parents:
diff changeset
31 { 0x3040, 0xD7FF },
anatofuz
parents:
diff changeset
32 // 7
anatofuz
parents:
diff changeset
33 { 0xF900, 0xFD3D }, { 0xFD40, 0xFDCF }, { 0xFDF0, 0xFE44 },
anatofuz
parents:
diff changeset
34 { 0xFE47, 0xFFFD },
anatofuz
parents:
diff changeset
35 // 8
anatofuz
parents:
diff changeset
36 { 0x10000, 0x1FFFD }, { 0x20000, 0x2FFFD }, { 0x30000, 0x3FFFD },
anatofuz
parents:
diff changeset
37 { 0x40000, 0x4FFFD }, { 0x50000, 0x5FFFD }, { 0x60000, 0x6FFFD },
anatofuz
parents:
diff changeset
38 { 0x70000, 0x7FFFD }, { 0x80000, 0x8FFFD }, { 0x90000, 0x9FFFD },
anatofuz
parents:
diff changeset
39 { 0xA0000, 0xAFFFD }, { 0xB0000, 0xBFFFD }, { 0xC0000, 0xCFFFD },
anatofuz
parents:
diff changeset
40 { 0xD0000, 0xDFFFD }, { 0xE0000, 0xEFFFD }
anatofuz
parents:
diff changeset
41 };
anatofuz
parents:
diff changeset
42
anatofuz
parents:
diff changeset
43 // C++03 [extendid]
anatofuz
parents:
diff changeset
44 // Note that this is not the same as C++98, but we don't distinguish C++98
anatofuz
parents:
diff changeset
45 // and C++03 in Clang.
anatofuz
parents:
diff changeset
46 static const llvm::sys::UnicodeCharRange CXX03AllowedIDCharRanges[] = {
anatofuz
parents:
diff changeset
47 // Latin
anatofuz
parents:
diff changeset
48 { 0x00C0, 0x00D6 }, { 0x00D8, 0x00F6 }, { 0x00F8, 0x01F5 },
anatofuz
parents:
diff changeset
49 { 0x01FA, 0x0217 }, { 0x0250, 0x02A8 },
anatofuz
parents:
diff changeset
50
anatofuz
parents:
diff changeset
51 // Greek
anatofuz
parents:
diff changeset
52 { 0x0384, 0x0384 }, { 0x0388, 0x038A }, { 0x038C, 0x038C },
anatofuz
parents:
diff changeset
53 { 0x038E, 0x03A1 }, { 0x03A3, 0x03CE }, { 0x03D0, 0x03D6 },
anatofuz
parents:
diff changeset
54 { 0x03DA, 0x03DA }, { 0x03DC, 0x03DC }, { 0x03DE, 0x03DE },
anatofuz
parents:
diff changeset
55 { 0x03E0, 0x03E0 }, { 0x03E2, 0x03F3 },
anatofuz
parents:
diff changeset
56
anatofuz
parents:
diff changeset
57 // Cyrillic
anatofuz
parents:
diff changeset
58 { 0x0401, 0x040D }, { 0x040F, 0x044F }, { 0x0451, 0x045C },
anatofuz
parents:
diff changeset
59 { 0x045E, 0x0481 }, { 0x0490, 0x04C4 }, { 0x04C7, 0x04C8 },
anatofuz
parents:
diff changeset
60 { 0x04CB, 0x04CC }, { 0x04D0, 0x04EB }, { 0x04EE, 0x04F5 },
anatofuz
parents:
diff changeset
61 { 0x04F8, 0x04F9 },
anatofuz
parents:
diff changeset
62
anatofuz
parents:
diff changeset
63 // Armenian
anatofuz
parents:
diff changeset
64 { 0x0531, 0x0556 }, { 0x0561, 0x0587 },
anatofuz
parents:
diff changeset
65
anatofuz
parents:
diff changeset
66 // Hebrew
anatofuz
parents:
diff changeset
67 { 0x05D0, 0x05EA }, { 0x05F0, 0x05F4 },
anatofuz
parents:
diff changeset
68
anatofuz
parents:
diff changeset
69 // Arabic
anatofuz
parents:
diff changeset
70 { 0x0621, 0x063A }, { 0x0640, 0x0652 }, { 0x0670, 0x06B7 },
anatofuz
parents:
diff changeset
71 { 0x06BA, 0x06BE }, { 0x06C0, 0x06CE }, { 0x06E5, 0x06E7 },
anatofuz
parents:
diff changeset
72
anatofuz
parents:
diff changeset
73 // Devanagari
anatofuz
parents:
diff changeset
74 { 0x0905, 0x0939 }, { 0x0958, 0x0962 },
anatofuz
parents:
diff changeset
75
anatofuz
parents:
diff changeset
76 // Bengali
anatofuz
parents:
diff changeset
77 { 0x0985, 0x098C }, { 0x098F, 0x0990 }, { 0x0993, 0x09A8 },
anatofuz
parents:
diff changeset
78 { 0x09AA, 0x09B0 }, { 0x09B2, 0x09B2 }, { 0x09B6, 0x09B9 },
anatofuz
parents:
diff changeset
79 { 0x09DC, 0x09DD }, { 0x09DF, 0x09E1 }, { 0x09F0, 0x09F1 },
anatofuz
parents:
diff changeset
80
anatofuz
parents:
diff changeset
81 // Gurmukhi
anatofuz
parents:
diff changeset
82 { 0x0A05, 0x0A0A }, { 0x0A0F, 0x0A10 }, { 0x0A13, 0x0A28 },
anatofuz
parents:
diff changeset
83 { 0x0A2A, 0x0A30 }, { 0x0A32, 0x0A33 }, { 0x0A35, 0x0A36 },
anatofuz
parents:
diff changeset
84 { 0x0A38, 0x0A39 }, { 0x0A59, 0x0A5C }, { 0x0A5E, 0x0A5E },
anatofuz
parents:
diff changeset
85
anatofuz
parents:
diff changeset
86 // Gujarti
anatofuz
parents:
diff changeset
87 { 0x0A85, 0x0A8B }, { 0x0A8D, 0x0A8D }, { 0x0A8F, 0x0A91 },
anatofuz
parents:
diff changeset
88 { 0x0A93, 0x0AA8 }, { 0x0AAA, 0x0AB0 }, { 0x0AB2, 0x0AB3 },
anatofuz
parents:
diff changeset
89 { 0x0AB5, 0x0AB9 }, { 0x0AE0, 0x0AE0 },
anatofuz
parents:
diff changeset
90
anatofuz
parents:
diff changeset
91 // Oriya
anatofuz
parents:
diff changeset
92 { 0x0B05, 0x0B0C }, { 0x0B0F, 0x0B10 }, { 0x0B13, 0x0B28 },
anatofuz
parents:
diff changeset
93 { 0x0B2A, 0x0B30 }, { 0x0B32, 0x0B33 }, { 0x0B36, 0x0B39 },
anatofuz
parents:
diff changeset
94 { 0x0B5C, 0x0B5D }, { 0x0B5F, 0x0B61 },
anatofuz
parents:
diff changeset
95
anatofuz
parents:
diff changeset
96 // Tamil
anatofuz
parents:
diff changeset
97 { 0x0B85, 0x0B8A }, { 0x0B8E, 0x0B90 }, { 0x0B92, 0x0B95 },
anatofuz
parents:
diff changeset
98 { 0x0B99, 0x0B9A }, { 0x0B9C, 0x0B9C }, { 0x0B9E, 0x0B9F },
anatofuz
parents:
diff changeset
99 { 0x0BA3, 0x0BA4 }, { 0x0BA8, 0x0BAA }, { 0x0BAE, 0x0BB5 },
anatofuz
parents:
diff changeset
100 { 0x0BB7, 0x0BB9 },
anatofuz
parents:
diff changeset
101
anatofuz
parents:
diff changeset
102 // Telugu
anatofuz
parents:
diff changeset
103 { 0x0C05, 0x0C0C }, { 0x0C0E, 0x0C10 }, { 0x0C12, 0x0C28 },
anatofuz
parents:
diff changeset
104 { 0x0C2A, 0x0C33 }, { 0x0C35, 0x0C39 }, { 0x0C60, 0x0C61 },
anatofuz
parents:
diff changeset
105
anatofuz
parents:
diff changeset
106 // Kannada
anatofuz
parents:
diff changeset
107 { 0x0C85, 0x0C8C }, { 0x0C8E, 0x0C90 }, { 0x0C92, 0x0CA8 },
anatofuz
parents:
diff changeset
108 { 0x0CAA, 0x0CB3 }, { 0x0CB5, 0x0CB9 }, { 0x0CE0, 0x0CE1 },
anatofuz
parents:
diff changeset
109
anatofuz
parents:
diff changeset
110 // Malayam
anatofuz
parents:
diff changeset
111 { 0x0D05, 0x0D0C }, { 0x0D0E, 0x0D10 }, { 0x0D12, 0x0D28 },
anatofuz
parents:
diff changeset
112 { 0x0D2A, 0x0D39 }, { 0x0D60, 0x0D61 },
anatofuz
parents:
diff changeset
113
anatofuz
parents:
diff changeset
114 // Thai
anatofuz
parents:
diff changeset
115 { 0x0E01, 0x0E30 }, { 0x0E32, 0x0E33 }, { 0x0E40, 0x0E46 },
anatofuz
parents:
diff changeset
116 { 0x0E4F, 0x0E5B },
anatofuz
parents:
diff changeset
117
anatofuz
parents:
diff changeset
118 // Lao
anatofuz
parents:
diff changeset
119 { 0x0E81, 0x0E82 }, { 0x0E84, 0x0E84 }, { 0x0E87, 0x0E87 },
anatofuz
parents:
diff changeset
120 { 0x0E88, 0x0E88 }, { 0x0E8A, 0x0E8A }, { 0x0E8D, 0x0E8D },
anatofuz
parents:
diff changeset
121 { 0x0E94, 0x0E97 }, { 0x0E99, 0x0E9F }, { 0x0EA1, 0x0EA3 },
anatofuz
parents:
diff changeset
122 { 0x0EA5, 0x0EA5 }, { 0x0EA7, 0x0EA7 }, { 0x0EAA, 0x0EAA },
anatofuz
parents:
diff changeset
123 { 0x0EAB, 0x0EAB }, { 0x0EAD, 0x0EB0 }, { 0x0EB2, 0x0EB2 },
anatofuz
parents:
diff changeset
124 { 0x0EB3, 0x0EB3 }, { 0x0EBD, 0x0EBD }, { 0x0EC0, 0x0EC4 },
anatofuz
parents:
diff changeset
125 { 0x0EC6, 0x0EC6 },
anatofuz
parents:
diff changeset
126
anatofuz
parents:
diff changeset
127 // Georgian
anatofuz
parents:
diff changeset
128 { 0x10A0, 0x10C5 }, { 0x10D0, 0x10F6 },
anatofuz
parents:
diff changeset
129
anatofuz
parents:
diff changeset
130 // Hangul
anatofuz
parents:
diff changeset
131 { 0x1100, 0x1159 }, { 0x1161, 0x11A2 }, { 0x11A8, 0x11F9 },
anatofuz
parents:
diff changeset
132
anatofuz
parents:
diff changeset
133 // Latin (2)
anatofuz
parents:
diff changeset
134 { 0x1E00, 0x1E9A }, { 0x1EA0, 0x1EF9 },
anatofuz
parents:
diff changeset
135
anatofuz
parents:
diff changeset
136 // Greek (2)
anatofuz
parents:
diff changeset
137 { 0x1F00, 0x1F15 }, { 0x1F18, 0x1F1D }, { 0x1F20, 0x1F45 },
anatofuz
parents:
diff changeset
138 { 0x1F48, 0x1F4D }, { 0x1F50, 0x1F57 }, { 0x1F59, 0x1F59 },
anatofuz
parents:
diff changeset
139 { 0x1F5B, 0x1F5B }, { 0x1F5D, 0x1F5D }, { 0x1F5F, 0x1F7D },
anatofuz
parents:
diff changeset
140 { 0x1F80, 0x1FB4 }, { 0x1FB6, 0x1FBC }, { 0x1FC2, 0x1FC4 },
anatofuz
parents:
diff changeset
141 { 0x1FC6, 0x1FCC }, { 0x1FD0, 0x1FD3 }, { 0x1FD6, 0x1FDB },
anatofuz
parents:
diff changeset
142 { 0x1FE0, 0x1FEC }, { 0x1FF2, 0x1FF4 }, { 0x1FF6, 0x1FFC },
anatofuz
parents:
diff changeset
143
anatofuz
parents:
diff changeset
144 // Hiragana
anatofuz
parents:
diff changeset
145 { 0x3041, 0x3094 }, { 0x309B, 0x309E },
anatofuz
parents:
diff changeset
146
anatofuz
parents:
diff changeset
147 // Katakana
anatofuz
parents:
diff changeset
148 { 0x30A1, 0x30FE },
anatofuz
parents:
diff changeset
149
anatofuz
parents:
diff changeset
150 // Bopmofo [sic]
anatofuz
parents:
diff changeset
151 { 0x3105, 0x312C },
anatofuz
parents:
diff changeset
152
anatofuz
parents:
diff changeset
153 // CJK Unified Ideographs
anatofuz
parents:
diff changeset
154 { 0x4E00, 0x9FA5 }, { 0xF900, 0xFA2D }, { 0xFB1F, 0xFB36 },
anatofuz
parents:
diff changeset
155 { 0xFB38, 0xFB3C }, { 0xFB3E, 0xFB3E }, { 0xFB40, 0xFB41 },
anatofuz
parents:
diff changeset
156 { 0xFB42, 0xFB44 }, { 0xFB46, 0xFBB1 }, { 0xFBD3, 0xFD3F },
anatofuz
parents:
diff changeset
157 { 0xFD50, 0xFD8F }, { 0xFD92, 0xFDC7 }, { 0xFDF0, 0xFDFB },
anatofuz
parents:
diff changeset
158 { 0xFE70, 0xFE72 }, { 0xFE74, 0xFE74 }, { 0xFE76, 0xFEFC },
anatofuz
parents:
diff changeset
159 { 0xFF21, 0xFF3A }, { 0xFF41, 0xFF5A }, { 0xFF66, 0xFFBE },
anatofuz
parents:
diff changeset
160 { 0xFFC2, 0xFFC7 }, { 0xFFCA, 0xFFCF }, { 0xFFD2, 0xFFD7 },
anatofuz
parents:
diff changeset
161 { 0xFFDA, 0xFFDC }
anatofuz
parents:
diff changeset
162 };
anatofuz
parents:
diff changeset
163
anatofuz
parents:
diff changeset
164 // C99 Annex D
anatofuz
parents:
diff changeset
165 static const llvm::sys::UnicodeCharRange C99AllowedIDCharRanges[] = {
anatofuz
parents:
diff changeset
166 // Latin (1)
anatofuz
parents:
diff changeset
167 { 0x00AA, 0x00AA },
anatofuz
parents:
diff changeset
168
anatofuz
parents:
diff changeset
169 // Special characters (1)
anatofuz
parents:
diff changeset
170 { 0x00B5, 0x00B5 }, { 0x00B7, 0x00B7 },
anatofuz
parents:
diff changeset
171
anatofuz
parents:
diff changeset
172 // Latin (2)
anatofuz
parents:
diff changeset
173 { 0x00BA, 0x00BA }, { 0x00C0, 0x00D6 }, { 0x00D8, 0x00F6 },
anatofuz
parents:
diff changeset
174 { 0x00F8, 0x01F5 }, { 0x01FA, 0x0217 }, { 0x0250, 0x02A8 },
anatofuz
parents:
diff changeset
175
anatofuz
parents:
diff changeset
176 // Special characters (2)
anatofuz
parents:
diff changeset
177 { 0x02B0, 0x02B8 }, { 0x02BB, 0x02BB }, { 0x02BD, 0x02C1 },
anatofuz
parents:
diff changeset
178 { 0x02D0, 0x02D1 }, { 0x02E0, 0x02E4 }, { 0x037A, 0x037A },
anatofuz
parents:
diff changeset
179
anatofuz
parents:
diff changeset
180 // Greek (1)
anatofuz
parents:
diff changeset
181 { 0x0386, 0x0386 }, { 0x0388, 0x038A }, { 0x038C, 0x038C },
anatofuz
parents:
diff changeset
182 { 0x038E, 0x03A1 }, { 0x03A3, 0x03CE }, { 0x03D0, 0x03D6 },
anatofuz
parents:
diff changeset
183 { 0x03DA, 0x03DA }, { 0x03DC, 0x03DC }, { 0x03DE, 0x03DE },
anatofuz
parents:
diff changeset
184 { 0x03E0, 0x03E0 }, { 0x03E2, 0x03F3 },
anatofuz
parents:
diff changeset
185
anatofuz
parents:
diff changeset
186 // Cyrillic
anatofuz
parents:
diff changeset
187 { 0x0401, 0x040C }, { 0x040E, 0x044F }, { 0x0451, 0x045C },
anatofuz
parents:
diff changeset
188 { 0x045E, 0x0481 }, { 0x0490, 0x04C4 }, { 0x04C7, 0x04C8 },
anatofuz
parents:
diff changeset
189 { 0x04CB, 0x04CC }, { 0x04D0, 0x04EB }, { 0x04EE, 0x04F5 },
anatofuz
parents:
diff changeset
190 { 0x04F8, 0x04F9 },
anatofuz
parents:
diff changeset
191
anatofuz
parents:
diff changeset
192 // Armenian (1)
anatofuz
parents:
diff changeset
193 { 0x0531, 0x0556 },
anatofuz
parents:
diff changeset
194
anatofuz
parents:
diff changeset
195 // Special characters (3)
anatofuz
parents:
diff changeset
196 { 0x0559, 0x0559 },
anatofuz
parents:
diff changeset
197
anatofuz
parents:
diff changeset
198 // Armenian (2)
anatofuz
parents:
diff changeset
199 { 0x0561, 0x0587 },
anatofuz
parents:
diff changeset
200
anatofuz
parents:
diff changeset
201 // Hebrew
anatofuz
parents:
diff changeset
202 { 0x05B0, 0x05B9 }, { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF },
anatofuz
parents:
diff changeset
203 { 0x05C1, 0x05C2 }, { 0x05D0, 0x05EA }, { 0x05F0, 0x05F2 },
anatofuz
parents:
diff changeset
204
anatofuz
parents:
diff changeset
205 // Arabic (1)
anatofuz
parents:
diff changeset
206 { 0x0621, 0x063A }, { 0x0640, 0x0652 },
anatofuz
parents:
diff changeset
207
anatofuz
parents:
diff changeset
208 // Digits (1)
anatofuz
parents:
diff changeset
209 { 0x0660, 0x0669 },
anatofuz
parents:
diff changeset
210
anatofuz
parents:
diff changeset
211 // Arabic (2)
anatofuz
parents:
diff changeset
212 { 0x0670, 0x06B7 }, { 0x06BA, 0x06BE }, { 0x06C0, 0x06CE },
anatofuz
parents:
diff changeset
213 { 0x06D0, 0x06DC }, { 0x06E5, 0x06E8 }, { 0x06EA, 0x06ED },
anatofuz
parents:
diff changeset
214
anatofuz
parents:
diff changeset
215 // Digits (2)
anatofuz
parents:
diff changeset
216 { 0x06F0, 0x06F9 },
anatofuz
parents:
diff changeset
217
anatofuz
parents:
diff changeset
218 // Devanagari and Special character 0x093D.
anatofuz
parents:
diff changeset
219 { 0x0901, 0x0903 }, { 0x0905, 0x0939 }, { 0x093D, 0x094D },
anatofuz
parents:
diff changeset
220 { 0x0950, 0x0952 }, { 0x0958, 0x0963 },
anatofuz
parents:
diff changeset
221
anatofuz
parents:
diff changeset
222 // Digits (3)
anatofuz
parents:
diff changeset
223 { 0x0966, 0x096F },
anatofuz
parents:
diff changeset
224
anatofuz
parents:
diff changeset
225 // Bengali (1)
anatofuz
parents:
diff changeset
226 { 0x0981, 0x0983 }, { 0x0985, 0x098C }, { 0x098F, 0x0990 },
anatofuz
parents:
diff changeset
227 { 0x0993, 0x09A8 }, { 0x09AA, 0x09B0 }, { 0x09B2, 0x09B2 },
anatofuz
parents:
diff changeset
228 { 0x09B6, 0x09B9 }, { 0x09BE, 0x09C4 }, { 0x09C7, 0x09C8 },
anatofuz
parents:
diff changeset
229 { 0x09CB, 0x09CD }, { 0x09DC, 0x09DD }, { 0x09DF, 0x09E3 },
anatofuz
parents:
diff changeset
230
anatofuz
parents:
diff changeset
231 // Digits (4)
anatofuz
parents:
diff changeset
232 { 0x09E6, 0x09EF },
anatofuz
parents:
diff changeset
233
anatofuz
parents:
diff changeset
234 // Bengali (2)
anatofuz
parents:
diff changeset
235 { 0x09F0, 0x09F1 },
anatofuz
parents:
diff changeset
236
anatofuz
parents:
diff changeset
237 // Gurmukhi (1)
anatofuz
parents:
diff changeset
238 { 0x0A02, 0x0A02 }, { 0x0A05, 0x0A0A }, { 0x0A0F, 0x0A10 },
anatofuz
parents:
diff changeset
239 { 0x0A13, 0x0A28 }, { 0x0A2A, 0x0A30 }, { 0x0A32, 0x0A33 },
anatofuz
parents:
diff changeset
240 { 0x0A35, 0x0A36 }, { 0x0A38, 0x0A39 }, { 0x0A3E, 0x0A42 },
anatofuz
parents:
diff changeset
241 { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D }, { 0x0A59, 0x0A5C },
anatofuz
parents:
diff changeset
242 { 0x0A5E, 0x0A5E },
anatofuz
parents:
diff changeset
243
anatofuz
parents:
diff changeset
244 // Digits (5)
anatofuz
parents:
diff changeset
245 { 0x0A66, 0x0A6F },
anatofuz
parents:
diff changeset
246
anatofuz
parents:
diff changeset
247 // Gurmukhi (2)
anatofuz
parents:
diff changeset
248 { 0x0A74, 0x0A74 },
anatofuz
parents:
diff changeset
249
anatofuz
parents:
diff changeset
250 // Gujarti
anatofuz
parents:
diff changeset
251 { 0x0A81, 0x0A83 }, { 0x0A85, 0x0A8B }, { 0x0A8D, 0x0A8D },
anatofuz
parents:
diff changeset
252 { 0x0A8F, 0x0A91 }, { 0x0A93, 0x0AA8 }, { 0x0AAA, 0x0AB0 },
anatofuz
parents:
diff changeset
253 { 0x0AB2, 0x0AB3 }, { 0x0AB5, 0x0AB9 }, { 0x0ABD, 0x0AC5 },
anatofuz
parents:
diff changeset
254 { 0x0AC7, 0x0AC9 }, { 0x0ACB, 0x0ACD }, { 0x0AD0, 0x0AD0 },
anatofuz
parents:
diff changeset
255 { 0x0AE0, 0x0AE0 },
anatofuz
parents:
diff changeset
256
anatofuz
parents:
diff changeset
257 // Digits (6)
anatofuz
parents:
diff changeset
258 { 0x0AE6, 0x0AEF },
anatofuz
parents:
diff changeset
259
anatofuz
parents:
diff changeset
260 // Oriya and Special character 0x0B3D
anatofuz
parents:
diff changeset
261 { 0x0B01, 0x0B03 }, { 0x0B05, 0x0B0C }, { 0x0B0F, 0x0B10 },
anatofuz
parents:
diff changeset
262 { 0x0B13, 0x0B28 }, { 0x0B2A, 0x0B30 }, { 0x0B32, 0x0B33 },
anatofuz
parents:
diff changeset
263 { 0x0B36, 0x0B39 }, { 0x0B3D, 0x0B43 }, { 0x0B47, 0x0B48 },
anatofuz
parents:
diff changeset
264 { 0x0B4B, 0x0B4D }, { 0x0B5C, 0x0B5D }, { 0x0B5F, 0x0B61 },
anatofuz
parents:
diff changeset
265
anatofuz
parents:
diff changeset
266 // Digits (7)
anatofuz
parents:
diff changeset
267 { 0x0B66, 0x0B6F },
anatofuz
parents:
diff changeset
268
anatofuz
parents:
diff changeset
269 // Tamil
anatofuz
parents:
diff changeset
270 { 0x0B82, 0x0B83 }, { 0x0B85, 0x0B8A }, { 0x0B8E, 0x0B90 },
anatofuz
parents:
diff changeset
271 { 0x0B92, 0x0B95 }, { 0x0B99, 0x0B9A }, { 0x0B9C, 0x0B9C },
anatofuz
parents:
diff changeset
272 { 0x0B9E, 0x0B9F }, { 0x0BA3, 0x0BA4 }, { 0x0BA8, 0x0BAA },
anatofuz
parents:
diff changeset
273 { 0x0BAE, 0x0BB5 }, { 0x0BB7, 0x0BB9 }, { 0x0BBE, 0x0BC2 },
anatofuz
parents:
diff changeset
274 { 0x0BC6, 0x0BC8 }, { 0x0BCA, 0x0BCD },
anatofuz
parents:
diff changeset
275
anatofuz
parents:
diff changeset
276 // Digits (8)
anatofuz
parents:
diff changeset
277 { 0x0BE7, 0x0BEF },
anatofuz
parents:
diff changeset
278
anatofuz
parents:
diff changeset
279 // Telugu
anatofuz
parents:
diff changeset
280 { 0x0C01, 0x0C03 }, { 0x0C05, 0x0C0C }, { 0x0C0E, 0x0C10 },
anatofuz
parents:
diff changeset
281 { 0x0C12, 0x0C28 }, { 0x0C2A, 0x0C33 }, { 0x0C35, 0x0C39 },
anatofuz
parents:
diff changeset
282 { 0x0C3E, 0x0C44 }, { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D },
anatofuz
parents:
diff changeset
283 { 0x0C60, 0x0C61 },
anatofuz
parents:
diff changeset
284
anatofuz
parents:
diff changeset
285 // Digits (9)
anatofuz
parents:
diff changeset
286 { 0x0C66, 0x0C6F },
anatofuz
parents:
diff changeset
287
anatofuz
parents:
diff changeset
288 // Kannada
anatofuz
parents:
diff changeset
289 { 0x0C82, 0x0C83 }, { 0x0C85, 0x0C8C }, { 0x0C8E, 0x0C90 },
anatofuz
parents:
diff changeset
290 { 0x0C92, 0x0CA8 }, { 0x0CAA, 0x0CB3 }, { 0x0CB5, 0x0CB9 },
anatofuz
parents:
diff changeset
291 { 0x0CBE, 0x0CC4 }, { 0x0CC6, 0x0CC8 }, { 0x0CCA, 0x0CCD },
anatofuz
parents:
diff changeset
292 { 0x0CDE, 0x0CDE }, { 0x0CE0, 0x0CE1 },
anatofuz
parents:
diff changeset
293
anatofuz
parents:
diff changeset
294 // Digits (10)
anatofuz
parents:
diff changeset
295 { 0x0CE6, 0x0CEF },
anatofuz
parents:
diff changeset
296
anatofuz
parents:
diff changeset
297 // Malayam
anatofuz
parents:
diff changeset
298 { 0x0D02, 0x0D03 }, { 0x0D05, 0x0D0C }, { 0x0D0E, 0x0D10 },
anatofuz
parents:
diff changeset
299 { 0x0D12, 0x0D28 }, { 0x0D2A, 0x0D39 }, { 0x0D3E, 0x0D43 },
anatofuz
parents:
diff changeset
300 { 0x0D46, 0x0D48 }, { 0x0D4A, 0x0D4D }, { 0x0D60, 0x0D61 },
anatofuz
parents:
diff changeset
301
anatofuz
parents:
diff changeset
302 // Digits (11)
anatofuz
parents:
diff changeset
303 { 0x0D66, 0x0D6F },
anatofuz
parents:
diff changeset
304
anatofuz
parents:
diff changeset
305 // Thai...including Digits { 0x0E50, 0x0E59 }
anatofuz
parents:
diff changeset
306 { 0x0E01, 0x0E3A }, { 0x0E40, 0x0E5B },
anatofuz
parents:
diff changeset
307
anatofuz
parents:
diff changeset
308 // Lao (1)
anatofuz
parents:
diff changeset
309 { 0x0E81, 0x0E82 }, { 0x0E84, 0x0E84 }, { 0x0E87, 0x0E88 },
anatofuz
parents:
diff changeset
310 { 0x0E8A, 0x0E8A }, { 0x0E8D, 0x0E8D }, { 0x0E94, 0x0E97 },
anatofuz
parents:
diff changeset
311 { 0x0E99, 0x0E9F }, { 0x0EA1, 0x0EA3 }, { 0x0EA5, 0x0EA5 },
anatofuz
parents:
diff changeset
312 { 0x0EA7, 0x0EA7 }, { 0x0EAA, 0x0EAB }, { 0x0EAD, 0x0EAE },
anatofuz
parents:
diff changeset
313 { 0x0EB0, 0x0EB9 }, { 0x0EBB, 0x0EBD }, { 0x0EC0, 0x0EC4 },
anatofuz
parents:
diff changeset
314 { 0x0EC6, 0x0EC6 }, { 0x0EC8, 0x0ECD },
anatofuz
parents:
diff changeset
315
anatofuz
parents:
diff changeset
316 // Digits (12)
anatofuz
parents:
diff changeset
317 { 0x0ED0, 0x0ED9 },
anatofuz
parents:
diff changeset
318
anatofuz
parents:
diff changeset
319 // Lao (2)
anatofuz
parents:
diff changeset
320 { 0x0EDC, 0x0EDD },
anatofuz
parents:
diff changeset
321
anatofuz
parents:
diff changeset
322 // Tibetan (1)
anatofuz
parents:
diff changeset
323 { 0x0F00, 0x0F00 }, { 0x0F18, 0x0F19 },
anatofuz
parents:
diff changeset
324
anatofuz
parents:
diff changeset
325 // Digits (13)
anatofuz
parents:
diff changeset
326 { 0x0F20, 0x0F33 },
anatofuz
parents:
diff changeset
327
anatofuz
parents:
diff changeset
328 // Tibetan (2)
anatofuz
parents:
diff changeset
329 { 0x0F35, 0x0F35 }, { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 },
anatofuz
parents:
diff changeset
330 { 0x0F3E, 0x0F47 }, { 0x0F49, 0x0F69 }, { 0x0F71, 0x0F84 },
anatofuz
parents:
diff changeset
331 { 0x0F86, 0x0F8B }, { 0x0F90, 0x0F95 }, { 0x0F97, 0x0F97 },
anatofuz
parents:
diff changeset
332 { 0x0F99, 0x0FAD }, { 0x0FB1, 0x0FB7 }, { 0x0FB9, 0x0FB9 },
anatofuz
parents:
diff changeset
333
anatofuz
parents:
diff changeset
334 // Georgian
anatofuz
parents:
diff changeset
335 { 0x10A0, 0x10C5 }, { 0x10D0, 0x10F6 },
anatofuz
parents:
diff changeset
336
anatofuz
parents:
diff changeset
337 // Latin (3)
anatofuz
parents:
diff changeset
338 { 0x1E00, 0x1E9B }, { 0x1EA0, 0x1EF9 },
anatofuz
parents:
diff changeset
339
anatofuz
parents:
diff changeset
340 // Greek (2)
anatofuz
parents:
diff changeset
341 { 0x1F00, 0x1F15 }, { 0x1F18, 0x1F1D }, { 0x1F20, 0x1F45 },
anatofuz
parents:
diff changeset
342 { 0x1F48, 0x1F4D }, { 0x1F50, 0x1F57 }, { 0x1F59, 0x1F59 },
anatofuz
parents:
diff changeset
343 { 0x1F5B, 0x1F5B }, { 0x1F5D, 0x1F5D }, { 0x1F5F, 0x1F7D },
anatofuz
parents:
diff changeset
344 { 0x1F80, 0x1FB4 }, { 0x1FB6, 0x1FBC },
anatofuz
parents:
diff changeset
345
anatofuz
parents:
diff changeset
346 // Special characters (4)
anatofuz
parents:
diff changeset
347 { 0x1FBE, 0x1FBE },
anatofuz
parents:
diff changeset
348
anatofuz
parents:
diff changeset
349 // Greek (3)
anatofuz
parents:
diff changeset
350 { 0x1FC2, 0x1FC4 }, { 0x1FC6, 0x1FCC }, { 0x1FD0, 0x1FD3 },
anatofuz
parents:
diff changeset
351 { 0x1FD6, 0x1FDB }, { 0x1FE0, 0x1FEC }, { 0x1FF2, 0x1FF4 },
anatofuz
parents:
diff changeset
352 { 0x1FF6, 0x1FFC },
anatofuz
parents:
diff changeset
353
anatofuz
parents:
diff changeset
354 // Special characters (5)
anatofuz
parents:
diff changeset
355 { 0x203F, 0x2040 },
anatofuz
parents:
diff changeset
356
anatofuz
parents:
diff changeset
357 // Latin (4)
anatofuz
parents:
diff changeset
358 { 0x207F, 0x207F },
anatofuz
parents:
diff changeset
359
anatofuz
parents:
diff changeset
360 // Special characters (6)
anatofuz
parents:
diff changeset
361 { 0x2102, 0x2102 }, { 0x2107, 0x2107 }, { 0x210A, 0x2113 },
anatofuz
parents:
diff changeset
362 { 0x2115, 0x2115 }, { 0x2118, 0x211D }, { 0x2124, 0x2124 },
anatofuz
parents:
diff changeset
363 { 0x2126, 0x2126 }, { 0x2128, 0x2128 }, { 0x212A, 0x2131 },
anatofuz
parents:
diff changeset
364 { 0x2133, 0x2138 }, { 0x2160, 0x2182 }, { 0x3005, 0x3007 },
anatofuz
parents:
diff changeset
365 { 0x3021, 0x3029 },
anatofuz
parents:
diff changeset
366
anatofuz
parents:
diff changeset
367 // Hiragana
anatofuz
parents:
diff changeset
368 { 0x3041, 0x3093 }, { 0x309B, 0x309C },
anatofuz
parents:
diff changeset
369
anatofuz
parents:
diff changeset
370 // Katakana
anatofuz
parents:
diff changeset
371 { 0x30A1, 0x30F6 }, { 0x30FB, 0x30FC },
anatofuz
parents:
diff changeset
372
anatofuz
parents:
diff changeset
373 // Bopmofo [sic]
anatofuz
parents:
diff changeset
374 { 0x3105, 0x312C },
anatofuz
parents:
diff changeset
375
anatofuz
parents:
diff changeset
376 // CJK Unified Ideographs
anatofuz
parents:
diff changeset
377 { 0x4E00, 0x9FA5 },
anatofuz
parents:
diff changeset
378
anatofuz
parents:
diff changeset
379 // Hangul,
anatofuz
parents:
diff changeset
380 { 0xAC00, 0xD7A3 }
anatofuz
parents:
diff changeset
381 };
anatofuz
parents:
diff changeset
382
anatofuz
parents:
diff changeset
383 // C11 D.2, C++11 [charname.disallowed]
anatofuz
parents:
diff changeset
384 static const llvm::sys::UnicodeCharRange C11DisallowedInitialIDCharRanges[] = {
anatofuz
parents:
diff changeset
385 { 0x0300, 0x036F }, { 0x1DC0, 0x1DFF }, { 0x20D0, 0x20FF },
anatofuz
parents:
diff changeset
386 { 0xFE20, 0xFE2F }
anatofuz
parents:
diff changeset
387 };
anatofuz
parents:
diff changeset
388
anatofuz
parents:
diff changeset
389 // C99 6.4.2.1p3: The initial character [of an identifier] shall not be a
anatofuz
parents:
diff changeset
390 // universal character name designating a digit.
anatofuz
parents:
diff changeset
391 // C99 Annex D defines these characters as "Digits".
anatofuz
parents:
diff changeset
392 static const llvm::sys::UnicodeCharRange C99DisallowedInitialIDCharRanges[] = {
anatofuz
parents:
diff changeset
393 { 0x0660, 0x0669 }, { 0x06F0, 0x06F9 }, { 0x0966, 0x096F },
anatofuz
parents:
diff changeset
394 { 0x09E6, 0x09EF }, { 0x0A66, 0x0A6F }, { 0x0AE6, 0x0AEF },
anatofuz
parents:
diff changeset
395 { 0x0B66, 0x0B6F }, { 0x0BE7, 0x0BEF }, { 0x0C66, 0x0C6F },
anatofuz
parents:
diff changeset
396 { 0x0CE6, 0x0CEF }, { 0x0D66, 0x0D6F }, { 0x0E50, 0x0E59 },
anatofuz
parents:
diff changeset
397 { 0x0ED0, 0x0ED9 }, { 0x0F20, 0x0F33 }
anatofuz
parents:
diff changeset
398 };
anatofuz
parents:
diff changeset
399
anatofuz
parents:
diff changeset
400 // Unicode v6.2, chapter 6.2, table 6-2.
anatofuz
parents:
diff changeset
401 static const llvm::sys::UnicodeCharRange UnicodeWhitespaceCharRanges[] = {
anatofuz
parents:
diff changeset
402 { 0x0085, 0x0085 }, { 0x00A0, 0x00A0 }, { 0x1680, 0x1680 },
anatofuz
parents:
diff changeset
403 { 0x180E, 0x180E }, { 0x2000, 0x200A }, { 0x2028, 0x2029 },
anatofuz
parents:
diff changeset
404 { 0x202F, 0x202F }, { 0x205F, 0x205F }, { 0x3000, 0x3000 }
anatofuz
parents:
diff changeset
405 };
anatofuz
parents:
diff changeset
406
anatofuz
parents:
diff changeset
407 #endif