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