xref: /freebsd/contrib/llvm-project/clang/lib/Lex/UnicodeCharSets.h (revision a90b9d0159070121c221b966469c3e36d912bf82)
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 // Unicode 15.1 XID_Start
14 static const llvm::sys::UnicodeCharRange XIDStartRanges[] = {
15     {0x0041, 0x005A},   {0x0061, 0x007A},   {0x00AA, 0x00AA},
16     {0x00B5, 0x00B5},   {0x00BA, 0x00BA},   {0x00C0, 0x00D6},
17     {0x00D8, 0x00F6},   {0x00F8, 0x02C1},   {0x02C6, 0x02D1},
18     {0x02E0, 0x02E4},   {0x02EC, 0x02EC},   {0x02EE, 0x02EE},
19     {0x0370, 0x0374},   {0x0376, 0x0377},   {0x037B, 0x037D},
20     {0x037F, 0x037F},   {0x0386, 0x0386},   {0x0388, 0x038A},
21     {0x038C, 0x038C},   {0x038E, 0x03A1},   {0x03A3, 0x03F5},
22     {0x03F7, 0x0481},   {0x048A, 0x052F},   {0x0531, 0x0556},
23     {0x0559, 0x0559},   {0x0560, 0x0588},   {0x05D0, 0x05EA},
24     {0x05EF, 0x05F2},   {0x0620, 0x064A},   {0x066E, 0x066F},
25     {0x0671, 0x06D3},   {0x06D5, 0x06D5},   {0x06E5, 0x06E6},
26     {0x06EE, 0x06EF},   {0x06FA, 0x06FC},   {0x06FF, 0x06FF},
27     {0x0710, 0x0710},   {0x0712, 0x072F},   {0x074D, 0x07A5},
28     {0x07B1, 0x07B1},   {0x07CA, 0x07EA},   {0x07F4, 0x07F5},
29     {0x07FA, 0x07FA},   {0x0800, 0x0815},   {0x081A, 0x081A},
30     {0x0824, 0x0824},   {0x0828, 0x0828},   {0x0840, 0x0858},
31     {0x0860, 0x086A},   {0x0870, 0x0887},   {0x0889, 0x088E},
32     {0x08A0, 0x08C9},   {0x0904, 0x0939},   {0x093D, 0x093D},
33     {0x0950, 0x0950},   {0x0958, 0x0961},   {0x0971, 0x0980},
34     {0x0985, 0x098C},   {0x098F, 0x0990},   {0x0993, 0x09A8},
35     {0x09AA, 0x09B0},   {0x09B2, 0x09B2},   {0x09B6, 0x09B9},
36     {0x09BD, 0x09BD},   {0x09CE, 0x09CE},   {0x09DC, 0x09DD},
37     {0x09DF, 0x09E1},   {0x09F0, 0x09F1},   {0x09FC, 0x09FC},
38     {0x0A05, 0x0A0A},   {0x0A0F, 0x0A10},   {0x0A13, 0x0A28},
39     {0x0A2A, 0x0A30},   {0x0A32, 0x0A33},   {0x0A35, 0x0A36},
40     {0x0A38, 0x0A39},   {0x0A59, 0x0A5C},   {0x0A5E, 0x0A5E},
41     {0x0A72, 0x0A74},   {0x0A85, 0x0A8D},   {0x0A8F, 0x0A91},
42     {0x0A93, 0x0AA8},   {0x0AAA, 0x0AB0},   {0x0AB2, 0x0AB3},
43     {0x0AB5, 0x0AB9},   {0x0ABD, 0x0ABD},   {0x0AD0, 0x0AD0},
44     {0x0AE0, 0x0AE1},   {0x0AF9, 0x0AF9},   {0x0B05, 0x0B0C},
45     {0x0B0F, 0x0B10},   {0x0B13, 0x0B28},   {0x0B2A, 0x0B30},
46     {0x0B32, 0x0B33},   {0x0B35, 0x0B39},   {0x0B3D, 0x0B3D},
47     {0x0B5C, 0x0B5D},   {0x0B5F, 0x0B61},   {0x0B71, 0x0B71},
48     {0x0B83, 0x0B83},   {0x0B85, 0x0B8A},   {0x0B8E, 0x0B90},
49     {0x0B92, 0x0B95},   {0x0B99, 0x0B9A},   {0x0B9C, 0x0B9C},
50     {0x0B9E, 0x0B9F},   {0x0BA3, 0x0BA4},   {0x0BA8, 0x0BAA},
51     {0x0BAE, 0x0BB9},   {0x0BD0, 0x0BD0},   {0x0C05, 0x0C0C},
52     {0x0C0E, 0x0C10},   {0x0C12, 0x0C28},   {0x0C2A, 0x0C39},
53     {0x0C3D, 0x0C3D},   {0x0C58, 0x0C5A},   {0x0C5D, 0x0C5D},
54     {0x0C60, 0x0C61},   {0x0C80, 0x0C80},   {0x0C85, 0x0C8C},
55     {0x0C8E, 0x0C90},   {0x0C92, 0x0CA8},   {0x0CAA, 0x0CB3},
56     {0x0CB5, 0x0CB9},   {0x0CBD, 0x0CBD},   {0x0CDD, 0x0CDE},
57     {0x0CE0, 0x0CE1},   {0x0CF1, 0x0CF2},   {0x0D04, 0x0D0C},
58     {0x0D0E, 0x0D10},   {0x0D12, 0x0D3A},   {0x0D3D, 0x0D3D},
59     {0x0D4E, 0x0D4E},   {0x0D54, 0x0D56},   {0x0D5F, 0x0D61},
60     {0x0D7A, 0x0D7F},   {0x0D85, 0x0D96},   {0x0D9A, 0x0DB1},
61     {0x0DB3, 0x0DBB},   {0x0DBD, 0x0DBD},   {0x0DC0, 0x0DC6},
62     {0x0E01, 0x0E30},   {0x0E32, 0x0E32},   {0x0E40, 0x0E46},
63     {0x0E81, 0x0E82},   {0x0E84, 0x0E84},   {0x0E86, 0x0E8A},
64     {0x0E8C, 0x0EA3},   {0x0EA5, 0x0EA5},   {0x0EA7, 0x0EB0},
65     {0x0EB2, 0x0EB2},   {0x0EBD, 0x0EBD},   {0x0EC0, 0x0EC4},
66     {0x0EC6, 0x0EC6},   {0x0EDC, 0x0EDF},   {0x0F00, 0x0F00},
67     {0x0F40, 0x0F47},   {0x0F49, 0x0F6C},   {0x0F88, 0x0F8C},
68     {0x1000, 0x102A},   {0x103F, 0x103F},   {0x1050, 0x1055},
69     {0x105A, 0x105D},   {0x1061, 0x1061},   {0x1065, 0x1066},
70     {0x106E, 0x1070},   {0x1075, 0x1081},   {0x108E, 0x108E},
71     {0x10A0, 0x10C5},   {0x10C7, 0x10C7},   {0x10CD, 0x10CD},
72     {0x10D0, 0x10FA},   {0x10FC, 0x1248},   {0x124A, 0x124D},
73     {0x1250, 0x1256},   {0x1258, 0x1258},   {0x125A, 0x125D},
74     {0x1260, 0x1288},   {0x128A, 0x128D},   {0x1290, 0x12B0},
75     {0x12B2, 0x12B5},   {0x12B8, 0x12BE},   {0x12C0, 0x12C0},
76     {0x12C2, 0x12C5},   {0x12C8, 0x12D6},   {0x12D8, 0x1310},
77     {0x1312, 0x1315},   {0x1318, 0x135A},   {0x1380, 0x138F},
78     {0x13A0, 0x13F5},   {0x13F8, 0x13FD},   {0x1401, 0x166C},
79     {0x166F, 0x167F},   {0x1681, 0x169A},   {0x16A0, 0x16EA},
80     {0x16EE, 0x16F8},   {0x1700, 0x1711},   {0x171F, 0x1731},
81     {0x1740, 0x1751},   {0x1760, 0x176C},   {0x176E, 0x1770},
82     {0x1780, 0x17B3},   {0x17D7, 0x17D7},   {0x17DC, 0x17DC},
83     {0x1820, 0x1878},   {0x1880, 0x18A8},   {0x18AA, 0x18AA},
84     {0x18B0, 0x18F5},   {0x1900, 0x191E},   {0x1950, 0x196D},
85     {0x1970, 0x1974},   {0x1980, 0x19AB},   {0x19B0, 0x19C9},
86     {0x1A00, 0x1A16},   {0x1A20, 0x1A54},   {0x1AA7, 0x1AA7},
87     {0x1B05, 0x1B33},   {0x1B45, 0x1B4C},   {0x1B83, 0x1BA0},
88     {0x1BAE, 0x1BAF},   {0x1BBA, 0x1BE5},   {0x1C00, 0x1C23},
89     {0x1C4D, 0x1C4F},   {0x1C5A, 0x1C7D},   {0x1C80, 0x1C88},
90     {0x1C90, 0x1CBA},   {0x1CBD, 0x1CBF},   {0x1CE9, 0x1CEC},
91     {0x1CEE, 0x1CF3},   {0x1CF5, 0x1CF6},   {0x1CFA, 0x1CFA},
92     {0x1D00, 0x1DBF},   {0x1E00, 0x1F15},   {0x1F18, 0x1F1D},
93     {0x1F20, 0x1F45},   {0x1F48, 0x1F4D},   {0x1F50, 0x1F57},
94     {0x1F59, 0x1F59},   {0x1F5B, 0x1F5B},   {0x1F5D, 0x1F5D},
95     {0x1F5F, 0x1F7D},   {0x1F80, 0x1FB4},   {0x1FB6, 0x1FBC},
96     {0x1FBE, 0x1FBE},   {0x1FC2, 0x1FC4},   {0x1FC6, 0x1FCC},
97     {0x1FD0, 0x1FD3},   {0x1FD6, 0x1FDB},   {0x1FE0, 0x1FEC},
98     {0x1FF2, 0x1FF4},   {0x1FF6, 0x1FFC},   {0x2071, 0x2071},
99     {0x207F, 0x207F},   {0x2090, 0x209C},   {0x2102, 0x2102},
100     {0x2107, 0x2107},   {0x210A, 0x2113},   {0x2115, 0x2115},
101     {0x2118, 0x211D},   {0x2124, 0x2124},   {0x2126, 0x2126},
102     {0x2128, 0x2128},   {0x212A, 0x2139},   {0x213C, 0x213F},
103     {0x2145, 0x2149},   {0x214E, 0x214E},   {0x2160, 0x2188},
104     {0x2C00, 0x2CE4},   {0x2CEB, 0x2CEE},   {0x2CF2, 0x2CF3},
105     {0x2D00, 0x2D25},   {0x2D27, 0x2D27},   {0x2D2D, 0x2D2D},
106     {0x2D30, 0x2D67},   {0x2D6F, 0x2D6F},   {0x2D80, 0x2D96},
107     {0x2DA0, 0x2DA6},   {0x2DA8, 0x2DAE},   {0x2DB0, 0x2DB6},
108     {0x2DB8, 0x2DBE},   {0x2DC0, 0x2DC6},   {0x2DC8, 0x2DCE},
109     {0x2DD0, 0x2DD6},   {0x2DD8, 0x2DDE},   {0x3005, 0x3007},
110     {0x3021, 0x3029},   {0x3031, 0x3035},   {0x3038, 0x303C},
111     {0x3041, 0x3096},   {0x309D, 0x309F},   {0x30A1, 0x30FA},
112     {0x30FC, 0x30FF},   {0x3105, 0x312F},   {0x3131, 0x318E},
113     {0x31A0, 0x31BF},   {0x31F0, 0x31FF},   {0x3400, 0x4DBF},
114     {0x4E00, 0xA48C},   {0xA4D0, 0xA4FD},   {0xA500, 0xA60C},
115     {0xA610, 0xA61F},   {0xA62A, 0xA62B},   {0xA640, 0xA66E},
116     {0xA67F, 0xA69D},   {0xA6A0, 0xA6EF},   {0xA717, 0xA71F},
117     {0xA722, 0xA788},   {0xA78B, 0xA7CA},   {0xA7D0, 0xA7D1},
118     {0xA7D3, 0xA7D3},   {0xA7D5, 0xA7D9},   {0xA7F2, 0xA801},
119     {0xA803, 0xA805},   {0xA807, 0xA80A},   {0xA80C, 0xA822},
120     {0xA840, 0xA873},   {0xA882, 0xA8B3},   {0xA8F2, 0xA8F7},
121     {0xA8FB, 0xA8FB},   {0xA8FD, 0xA8FE},   {0xA90A, 0xA925},
122     {0xA930, 0xA946},   {0xA960, 0xA97C},   {0xA984, 0xA9B2},
123     {0xA9CF, 0xA9CF},   {0xA9E0, 0xA9E4},   {0xA9E6, 0xA9EF},
124     {0xA9FA, 0xA9FE},   {0xAA00, 0xAA28},   {0xAA40, 0xAA42},
125     {0xAA44, 0xAA4B},   {0xAA60, 0xAA76},   {0xAA7A, 0xAA7A},
126     {0xAA7E, 0xAAAF},   {0xAAB1, 0xAAB1},   {0xAAB5, 0xAAB6},
127     {0xAAB9, 0xAABD},   {0xAAC0, 0xAAC0},   {0xAAC2, 0xAAC2},
128     {0xAADB, 0xAADD},   {0xAAE0, 0xAAEA},   {0xAAF2, 0xAAF4},
129     {0xAB01, 0xAB06},   {0xAB09, 0xAB0E},   {0xAB11, 0xAB16},
130     {0xAB20, 0xAB26},   {0xAB28, 0xAB2E},   {0xAB30, 0xAB5A},
131     {0xAB5C, 0xAB69},   {0xAB70, 0xABE2},   {0xAC00, 0xD7A3},
132     {0xD7B0, 0xD7C6},   {0xD7CB, 0xD7FB},   {0xF900, 0xFA6D},
133     {0xFA70, 0xFAD9},   {0xFB00, 0xFB06},   {0xFB13, 0xFB17},
134     {0xFB1D, 0xFB1D},   {0xFB1F, 0xFB28},   {0xFB2A, 0xFB36},
135     {0xFB38, 0xFB3C},   {0xFB3E, 0xFB3E},   {0xFB40, 0xFB41},
136     {0xFB43, 0xFB44},   {0xFB46, 0xFBB1},   {0xFBD3, 0xFC5D},
137     {0xFC64, 0xFD3D},   {0xFD50, 0xFD8F},   {0xFD92, 0xFDC7},
138     {0xFDF0, 0xFDF9},   {0xFE71, 0xFE71},   {0xFE73, 0xFE73},
139     {0xFE77, 0xFE77},   {0xFE79, 0xFE79},   {0xFE7B, 0xFE7B},
140     {0xFE7D, 0xFE7D},   {0xFE7F, 0xFEFC},   {0xFF21, 0xFF3A},
141     {0xFF41, 0xFF5A},   {0xFF66, 0xFF9D},   {0xFFA0, 0xFFBE},
142     {0xFFC2, 0xFFC7},   {0xFFCA, 0xFFCF},   {0xFFD2, 0xFFD7},
143     {0xFFDA, 0xFFDC},   {0x10000, 0x1000B}, {0x1000D, 0x10026},
144     {0x10028, 0x1003A}, {0x1003C, 0x1003D}, {0x1003F, 0x1004D},
145     {0x10050, 0x1005D}, {0x10080, 0x100FA}, {0x10140, 0x10174},
146     {0x10280, 0x1029C}, {0x102A0, 0x102D0}, {0x10300, 0x1031F},
147     {0x1032D, 0x1034A}, {0x10350, 0x10375}, {0x10380, 0x1039D},
148     {0x103A0, 0x103C3}, {0x103C8, 0x103CF}, {0x103D1, 0x103D5},
149     {0x10400, 0x1049D}, {0x104B0, 0x104D3}, {0x104D8, 0x104FB},
150     {0x10500, 0x10527}, {0x10530, 0x10563}, {0x10570, 0x1057A},
151     {0x1057C, 0x1058A}, {0x1058C, 0x10592}, {0x10594, 0x10595},
152     {0x10597, 0x105A1}, {0x105A3, 0x105B1}, {0x105B3, 0x105B9},
153     {0x105BB, 0x105BC}, {0x10600, 0x10736}, {0x10740, 0x10755},
154     {0x10760, 0x10767}, {0x10780, 0x10785}, {0x10787, 0x107B0},
155     {0x107B2, 0x107BA}, {0x10800, 0x10805}, {0x10808, 0x10808},
156     {0x1080A, 0x10835}, {0x10837, 0x10838}, {0x1083C, 0x1083C},
157     {0x1083F, 0x10855}, {0x10860, 0x10876}, {0x10880, 0x1089E},
158     {0x108E0, 0x108F2}, {0x108F4, 0x108F5}, {0x10900, 0x10915},
159     {0x10920, 0x10939}, {0x10980, 0x109B7}, {0x109BE, 0x109BF},
160     {0x10A00, 0x10A00}, {0x10A10, 0x10A13}, {0x10A15, 0x10A17},
161     {0x10A19, 0x10A35}, {0x10A60, 0x10A7C}, {0x10A80, 0x10A9C},
162     {0x10AC0, 0x10AC7}, {0x10AC9, 0x10AE4}, {0x10B00, 0x10B35},
163     {0x10B40, 0x10B55}, {0x10B60, 0x10B72}, {0x10B80, 0x10B91},
164     {0x10C00, 0x10C48}, {0x10C80, 0x10CB2}, {0x10CC0, 0x10CF2},
165     {0x10D00, 0x10D23}, {0x10E80, 0x10EA9}, {0x10EB0, 0x10EB1},
166     {0x10F00, 0x10F1C}, {0x10F27, 0x10F27}, {0x10F30, 0x10F45},
167     {0x10F70, 0x10F81}, {0x10FB0, 0x10FC4}, {0x10FE0, 0x10FF6},
168     {0x11003, 0x11037}, {0x11071, 0x11072}, {0x11075, 0x11075},
169     {0x11083, 0x110AF}, {0x110D0, 0x110E8}, {0x11103, 0x11126},
170     {0x11144, 0x11144}, {0x11147, 0x11147}, {0x11150, 0x11172},
171     {0x11176, 0x11176}, {0x11183, 0x111B2}, {0x111C1, 0x111C4},
172     {0x111DA, 0x111DA}, {0x111DC, 0x111DC}, {0x11200, 0x11211},
173     {0x11213, 0x1122B}, {0x1123F, 0x11240}, {0x11280, 0x11286},
174     {0x11288, 0x11288}, {0x1128A, 0x1128D}, {0x1128F, 0x1129D},
175     {0x1129F, 0x112A8}, {0x112B0, 0x112DE}, {0x11305, 0x1130C},
176     {0x1130F, 0x11310}, {0x11313, 0x11328}, {0x1132A, 0x11330},
177     {0x11332, 0x11333}, {0x11335, 0x11339}, {0x1133D, 0x1133D},
178     {0x11350, 0x11350}, {0x1135D, 0x11361}, {0x11400, 0x11434},
179     {0x11447, 0x1144A}, {0x1145F, 0x11461}, {0x11480, 0x114AF},
180     {0x114C4, 0x114C5}, {0x114C7, 0x114C7}, {0x11580, 0x115AE},
181     {0x115D8, 0x115DB}, {0x11600, 0x1162F}, {0x11644, 0x11644},
182     {0x11680, 0x116AA}, {0x116B8, 0x116B8}, {0x11700, 0x1171A},
183     {0x11740, 0x11746}, {0x11800, 0x1182B}, {0x118A0, 0x118DF},
184     {0x118FF, 0x11906}, {0x11909, 0x11909}, {0x1190C, 0x11913},
185     {0x11915, 0x11916}, {0x11918, 0x1192F}, {0x1193F, 0x1193F},
186     {0x11941, 0x11941}, {0x119A0, 0x119A7}, {0x119AA, 0x119D0},
187     {0x119E1, 0x119E1}, {0x119E3, 0x119E3}, {0x11A00, 0x11A00},
188     {0x11A0B, 0x11A32}, {0x11A3A, 0x11A3A}, {0x11A50, 0x11A50},
189     {0x11A5C, 0x11A89}, {0x11A9D, 0x11A9D}, {0x11AB0, 0x11AF8},
190     {0x11C00, 0x11C08}, {0x11C0A, 0x11C2E}, {0x11C40, 0x11C40},
191     {0x11C72, 0x11C8F}, {0x11D00, 0x11D06}, {0x11D08, 0x11D09},
192     {0x11D0B, 0x11D30}, {0x11D46, 0x11D46}, {0x11D60, 0x11D65},
193     {0x11D67, 0x11D68}, {0x11D6A, 0x11D89}, {0x11D98, 0x11D98},
194     {0x11EE0, 0x11EF2}, {0x11F02, 0x11F02}, {0x11F04, 0x11F10},
195     {0x11F12, 0x11F33}, {0x11FB0, 0x11FB0}, {0x12000, 0x12399},
196     {0x12400, 0x1246E}, {0x12480, 0x12543}, {0x12F90, 0x12FF0},
197     {0x13000, 0x1342F}, {0x13441, 0x13446}, {0x14400, 0x14646},
198     {0x16800, 0x16A38}, {0x16A40, 0x16A5E}, {0x16A70, 0x16ABE},
199     {0x16AD0, 0x16AED}, {0x16B00, 0x16B2F}, {0x16B40, 0x16B43},
200     {0x16B63, 0x16B77}, {0x16B7D, 0x16B8F}, {0x16E40, 0x16E7F},
201     {0x16F00, 0x16F4A}, {0x16F50, 0x16F50}, {0x16F93, 0x16F9F},
202     {0x16FE0, 0x16FE1}, {0x16FE3, 0x16FE3}, {0x17000, 0x187F7},
203     {0x18800, 0x18CD5}, {0x18D00, 0x18D08}, {0x1AFF0, 0x1AFF3},
204     {0x1AFF5, 0x1AFFB}, {0x1AFFD, 0x1AFFE}, {0x1B000, 0x1B122},
205     {0x1B132, 0x1B132}, {0x1B150, 0x1B152}, {0x1B155, 0x1B155},
206     {0x1B164, 0x1B167}, {0x1B170, 0x1B2FB}, {0x1BC00, 0x1BC6A},
207     {0x1BC70, 0x1BC7C}, {0x1BC80, 0x1BC88}, {0x1BC90, 0x1BC99},
208     {0x1D400, 0x1D454}, {0x1D456, 0x1D49C}, {0x1D49E, 0x1D49F},
209     {0x1D4A2, 0x1D4A2}, {0x1D4A5, 0x1D4A6}, {0x1D4A9, 0x1D4AC},
210     {0x1D4AE, 0x1D4B9}, {0x1D4BB, 0x1D4BB}, {0x1D4BD, 0x1D4C3},
211     {0x1D4C5, 0x1D505}, {0x1D507, 0x1D50A}, {0x1D50D, 0x1D514},
212     {0x1D516, 0x1D51C}, {0x1D51E, 0x1D539}, {0x1D53B, 0x1D53E},
213     {0x1D540, 0x1D544}, {0x1D546, 0x1D546}, {0x1D54A, 0x1D550},
214     {0x1D552, 0x1D6A5}, {0x1D6A8, 0x1D6C0}, {0x1D6C2, 0x1D6DA},
215     {0x1D6DC, 0x1D6FA}, {0x1D6FC, 0x1D714}, {0x1D716, 0x1D734},
216     {0x1D736, 0x1D74E}, {0x1D750, 0x1D76E}, {0x1D770, 0x1D788},
217     {0x1D78A, 0x1D7A8}, {0x1D7AA, 0x1D7C2}, {0x1D7C4, 0x1D7CB},
218     {0x1DF00, 0x1DF1E}, {0x1DF25, 0x1DF2A}, {0x1E030, 0x1E06D},
219     {0x1E100, 0x1E12C}, {0x1E137, 0x1E13D}, {0x1E14E, 0x1E14E},
220     {0x1E290, 0x1E2AD}, {0x1E2C0, 0x1E2EB}, {0x1E4D0, 0x1E4EB},
221     {0x1E7E0, 0x1E7E6}, {0x1E7E8, 0x1E7EB}, {0x1E7ED, 0x1E7EE},
222     {0x1E7F0, 0x1E7FE}, {0x1E800, 0x1E8C4}, {0x1E900, 0x1E943},
223     {0x1E94B, 0x1E94B}, {0x1EE00, 0x1EE03}, {0x1EE05, 0x1EE1F},
224     {0x1EE21, 0x1EE22}, {0x1EE24, 0x1EE24}, {0x1EE27, 0x1EE27},
225     {0x1EE29, 0x1EE32}, {0x1EE34, 0x1EE37}, {0x1EE39, 0x1EE39},
226     {0x1EE3B, 0x1EE3B}, {0x1EE42, 0x1EE42}, {0x1EE47, 0x1EE47},
227     {0x1EE49, 0x1EE49}, {0x1EE4B, 0x1EE4B}, {0x1EE4D, 0x1EE4F},
228     {0x1EE51, 0x1EE52}, {0x1EE54, 0x1EE54}, {0x1EE57, 0x1EE57},
229     {0x1EE59, 0x1EE59}, {0x1EE5B, 0x1EE5B}, {0x1EE5D, 0x1EE5D},
230     {0x1EE5F, 0x1EE5F}, {0x1EE61, 0x1EE62}, {0x1EE64, 0x1EE64},
231     {0x1EE67, 0x1EE6A}, {0x1EE6C, 0x1EE72}, {0x1EE74, 0x1EE77},
232     {0x1EE79, 0x1EE7C}, {0x1EE7E, 0x1EE7E}, {0x1EE80, 0x1EE89},
233     {0x1EE8B, 0x1EE9B}, {0x1EEA1, 0x1EEA3}, {0x1EEA5, 0x1EEA9},
234     {0x1EEAB, 0x1EEBB}, {0x20000, 0x2A6DF}, {0x2A700, 0x2B739},
235     {0x2B740, 0x2B81D}, {0x2B820, 0x2CEA1}, {0x2CEB0, 0x2EBE0},
236     {0x2EBF0, 0x2EE5D}, {0x2F800, 0x2FA1D}, {0x30000, 0x3134A},
237     {0x31350, 0x323AF}};
238 
239 // Unicode 15.1 XID_Continue, excluding XID_Start
240 // The Unicode Property XID_Continue is a super set of XID_Start.
241 // To save Space, the table below only contains the codepoints
242 // that are not also in XID_Start.
243 static const llvm::sys::UnicodeCharRange XIDContinueRanges[] = {
244     {0x0030, 0x0039},   {0x005F, 0x005F},   {0x00B7, 0x00B7},
245     {0x0300, 0x036F},   {0x0387, 0x0387},   {0x0483, 0x0487},
246     {0x0591, 0x05BD},   {0x05BF, 0x05BF},   {0x05C1, 0x05C2},
247     {0x05C4, 0x05C5},   {0x05C7, 0x05C7},   {0x0610, 0x061A},
248     {0x064B, 0x0669},   {0x0670, 0x0670},   {0x06D6, 0x06DC},
249     {0x06DF, 0x06E4},   {0x06E7, 0x06E8},   {0x06EA, 0x06ED},
250     {0x06F0, 0x06F9},   {0x0711, 0x0711},   {0x0730, 0x074A},
251     {0x07A6, 0x07B0},   {0x07C0, 0x07C9},   {0x07EB, 0x07F3},
252     {0x07FD, 0x07FD},   {0x0816, 0x0819},   {0x081B, 0x0823},
253     {0x0825, 0x0827},   {0x0829, 0x082D},   {0x0859, 0x085B},
254     {0x0898, 0x089F},   {0x08CA, 0x08E1},   {0x08E3, 0x0903},
255     {0x093A, 0x093C},   {0x093E, 0x094F},   {0x0951, 0x0957},
256     {0x0962, 0x0963},   {0x0966, 0x096F},   {0x0981, 0x0983},
257     {0x09BC, 0x09BC},   {0x09BE, 0x09C4},   {0x09C7, 0x09C8},
258     {0x09CB, 0x09CD},   {0x09D7, 0x09D7},   {0x09E2, 0x09E3},
259     {0x09E6, 0x09EF},   {0x09FE, 0x09FE},   {0x0A01, 0x0A03},
260     {0x0A3C, 0x0A3C},   {0x0A3E, 0x0A42},   {0x0A47, 0x0A48},
261     {0x0A4B, 0x0A4D},   {0x0A51, 0x0A51},   {0x0A66, 0x0A71},
262     {0x0A75, 0x0A75},   {0x0A81, 0x0A83},   {0x0ABC, 0x0ABC},
263     {0x0ABE, 0x0AC5},   {0x0AC7, 0x0AC9},   {0x0ACB, 0x0ACD},
264     {0x0AE2, 0x0AE3},   {0x0AE6, 0x0AEF},   {0x0AFA, 0x0AFF},
265     {0x0B01, 0x0B03},   {0x0B3C, 0x0B3C},   {0x0B3E, 0x0B44},
266     {0x0B47, 0x0B48},   {0x0B4B, 0x0B4D},   {0x0B55, 0x0B57},
267     {0x0B62, 0x0B63},   {0x0B66, 0x0B6F},   {0x0B82, 0x0B82},
268     {0x0BBE, 0x0BC2},   {0x0BC6, 0x0BC8},   {0x0BCA, 0x0BCD},
269     {0x0BD7, 0x0BD7},   {0x0BE6, 0x0BEF},   {0x0C00, 0x0C04},
270     {0x0C3C, 0x0C3C},   {0x0C3E, 0x0C44},   {0x0C46, 0x0C48},
271     {0x0C4A, 0x0C4D},   {0x0C55, 0x0C56},   {0x0C62, 0x0C63},
272     {0x0C66, 0x0C6F},   {0x0C81, 0x0C83},   {0x0CBC, 0x0CBC},
273     {0x0CBE, 0x0CC4},   {0x0CC6, 0x0CC8},   {0x0CCA, 0x0CCD},
274     {0x0CD5, 0x0CD6},   {0x0CE2, 0x0CE3},   {0x0CE6, 0x0CEF},
275     {0x0CF3, 0x0CF3},   {0x0D00, 0x0D03},   {0x0D3B, 0x0D3C},
276     {0x0D3E, 0x0D44},   {0x0D46, 0x0D48},   {0x0D4A, 0x0D4D},
277     {0x0D57, 0x0D57},   {0x0D62, 0x0D63},   {0x0D66, 0x0D6F},
278     {0x0D81, 0x0D83},   {0x0DCA, 0x0DCA},   {0x0DCF, 0x0DD4},
279     {0x0DD6, 0x0DD6},   {0x0DD8, 0x0DDF},   {0x0DE6, 0x0DEF},
280     {0x0DF2, 0x0DF3},   {0x0E31, 0x0E31},   {0x0E33, 0x0E3A},
281     {0x0E47, 0x0E4E},   {0x0E50, 0x0E59},   {0x0EB1, 0x0EB1},
282     {0x0EB3, 0x0EBC},   {0x0EC8, 0x0ECE},   {0x0ED0, 0x0ED9},
283     {0x0F18, 0x0F19},   {0x0F20, 0x0F29},   {0x0F35, 0x0F35},
284     {0x0F37, 0x0F37},   {0x0F39, 0x0F39},   {0x0F3E, 0x0F3F},
285     {0x0F71, 0x0F84},   {0x0F86, 0x0F87},   {0x0F8D, 0x0F97},
286     {0x0F99, 0x0FBC},   {0x0FC6, 0x0FC6},   {0x102B, 0x103E},
287     {0x1040, 0x1049},   {0x1056, 0x1059},   {0x105E, 0x1060},
288     {0x1062, 0x1064},   {0x1067, 0x106D},   {0x1071, 0x1074},
289     {0x1082, 0x108D},   {0x108F, 0x109D},   {0x135D, 0x135F},
290     {0x1369, 0x1371},   {0x1712, 0x1715},   {0x1732, 0x1734},
291     {0x1752, 0x1753},   {0x1772, 0x1773},   {0x17B4, 0x17D3},
292     {0x17DD, 0x17DD},   {0x17E0, 0x17E9},   {0x180B, 0x180D},
293     {0x180F, 0x1819},   {0x18A9, 0x18A9},   {0x1920, 0x192B},
294     {0x1930, 0x193B},   {0x1946, 0x194F},   {0x19D0, 0x19DA},
295     {0x1A17, 0x1A1B},   {0x1A55, 0x1A5E},   {0x1A60, 0x1A7C},
296     {0x1A7F, 0x1A89},   {0x1A90, 0x1A99},   {0x1AB0, 0x1ABD},
297     {0x1ABF, 0x1ACE},   {0x1B00, 0x1B04},   {0x1B34, 0x1B44},
298     {0x1B50, 0x1B59},   {0x1B6B, 0x1B73},   {0x1B80, 0x1B82},
299     {0x1BA1, 0x1BAD},   {0x1BB0, 0x1BB9},   {0x1BE6, 0x1BF3},
300     {0x1C24, 0x1C37},   {0x1C40, 0x1C49},   {0x1C50, 0x1C59},
301     {0x1CD0, 0x1CD2},   {0x1CD4, 0x1CE8},   {0x1CED, 0x1CED},
302     {0x1CF4, 0x1CF4},   {0x1CF7, 0x1CF9},   {0x1DC0, 0x1DFF},
303     {0x200C, 0x200D},   {0x203F, 0x2040},   {0x2054, 0x2054},
304     {0x20D0, 0x20DC},   {0x20E1, 0x20E1},   {0x20E5, 0x20F0},
305     {0x2CEF, 0x2CF1},   {0x2D7F, 0x2D7F},   {0x2DE0, 0x2DFF},
306     {0x302A, 0x302F},   {0x3099, 0x309A},   {0x30FB, 0x30FB},
307     {0xA620, 0xA629},   {0xA66F, 0xA66F},   {0xA674, 0xA67D},
308     {0xA69E, 0xA69F},   {0xA6F0, 0xA6F1},   {0xA802, 0xA802},
309     {0xA806, 0xA806},   {0xA80B, 0xA80B},   {0xA823, 0xA827},
310     {0xA82C, 0xA82C},   {0xA880, 0xA881},   {0xA8B4, 0xA8C5},
311     {0xA8D0, 0xA8D9},   {0xA8E0, 0xA8F1},   {0xA8FF, 0xA909},
312     {0xA926, 0xA92D},   {0xA947, 0xA953},   {0xA980, 0xA983},
313     {0xA9B3, 0xA9C0},   {0xA9D0, 0xA9D9},   {0xA9E5, 0xA9E5},
314     {0xA9F0, 0xA9F9},   {0xAA29, 0xAA36},   {0xAA43, 0xAA43},
315     {0xAA4C, 0xAA4D},   {0xAA50, 0xAA59},   {0xAA7B, 0xAA7D},
316     {0xAAB0, 0xAAB0},   {0xAAB2, 0xAAB4},   {0xAAB7, 0xAAB8},
317     {0xAABE, 0xAABF},   {0xAAC1, 0xAAC1},   {0xAAEB, 0xAAEF},
318     {0xAAF5, 0xAAF6},   {0xABE3, 0xABEA},   {0xABEC, 0xABED},
319     {0xABF0, 0xABF9},   {0xFB1E, 0xFB1E},   {0xFE00, 0xFE0F},
320     {0xFE20, 0xFE2F},   {0xFE33, 0xFE34},   {0xFE4D, 0xFE4F},
321     {0xFF10, 0xFF19},   {0xFF3F, 0xFF3F},   {0xFF65, 0xFF65},
322     {0xFF9E, 0xFF9F},   {0x101FD, 0x101FD}, {0x102E0, 0x102E0},
323     {0x10376, 0x1037A}, {0x104A0, 0x104A9}, {0x10A01, 0x10A03},
324     {0x10A05, 0x10A06}, {0x10A0C, 0x10A0F}, {0x10A38, 0x10A3A},
325     {0x10A3F, 0x10A3F}, {0x10AE5, 0x10AE6}, {0x10D24, 0x10D27},
326     {0x10D30, 0x10D39}, {0x10EAB, 0x10EAC}, {0x10EFD, 0x10EFF},
327     {0x10F46, 0x10F50}, {0x10F82, 0x10F85}, {0x11000, 0x11002},
328     {0x11038, 0x11046}, {0x11066, 0x11070}, {0x11073, 0x11074},
329     {0x1107F, 0x11082}, {0x110B0, 0x110BA}, {0x110C2, 0x110C2},
330     {0x110F0, 0x110F9}, {0x11100, 0x11102}, {0x11127, 0x11134},
331     {0x11136, 0x1113F}, {0x11145, 0x11146}, {0x11173, 0x11173},
332     {0x11180, 0x11182}, {0x111B3, 0x111C0}, {0x111C9, 0x111CC},
333     {0x111CE, 0x111D9}, {0x1122C, 0x11237}, {0x1123E, 0x1123E},
334     {0x11241, 0x11241}, {0x112DF, 0x112EA}, {0x112F0, 0x112F9},
335     {0x11300, 0x11303}, {0x1133B, 0x1133C}, {0x1133E, 0x11344},
336     {0x11347, 0x11348}, {0x1134B, 0x1134D}, {0x11357, 0x11357},
337     {0x11362, 0x11363}, {0x11366, 0x1136C}, {0x11370, 0x11374},
338     {0x11435, 0x11446}, {0x11450, 0x11459}, {0x1145E, 0x1145E},
339     {0x114B0, 0x114C3}, {0x114D0, 0x114D9}, {0x115AF, 0x115B5},
340     {0x115B8, 0x115C0}, {0x115DC, 0x115DD}, {0x11630, 0x11640},
341     {0x11650, 0x11659}, {0x116AB, 0x116B7}, {0x116C0, 0x116C9},
342     {0x1171D, 0x1172B}, {0x11730, 0x11739}, {0x1182C, 0x1183A},
343     {0x118E0, 0x118E9}, {0x11930, 0x11935}, {0x11937, 0x11938},
344     {0x1193B, 0x1193E}, {0x11940, 0x11940}, {0x11942, 0x11943},
345     {0x11950, 0x11959}, {0x119D1, 0x119D7}, {0x119DA, 0x119E0},
346     {0x119E4, 0x119E4}, {0x11A01, 0x11A0A}, {0x11A33, 0x11A39},
347     {0x11A3B, 0x11A3E}, {0x11A47, 0x11A47}, {0x11A51, 0x11A5B},
348     {0x11A8A, 0x11A99}, {0x11C2F, 0x11C36}, {0x11C38, 0x11C3F},
349     {0x11C50, 0x11C59}, {0x11C92, 0x11CA7}, {0x11CA9, 0x11CB6},
350     {0x11D31, 0x11D36}, {0x11D3A, 0x11D3A}, {0x11D3C, 0x11D3D},
351     {0x11D3F, 0x11D45}, {0x11D47, 0x11D47}, {0x11D50, 0x11D59},
352     {0x11D8A, 0x11D8E}, {0x11D90, 0x11D91}, {0x11D93, 0x11D97},
353     {0x11DA0, 0x11DA9}, {0x11EF3, 0x11EF6}, {0x11F00, 0x11F01},
354     {0x11F03, 0x11F03}, {0x11F34, 0x11F3A}, {0x11F3E, 0x11F42},
355     {0x11F50, 0x11F59}, {0x13440, 0x13440}, {0x13447, 0x13455},
356     {0x16A60, 0x16A69}, {0x16AC0, 0x16AC9}, {0x16AF0, 0x16AF4},
357     {0x16B30, 0x16B36}, {0x16B50, 0x16B59}, {0x16F4F, 0x16F4F},
358     {0x16F51, 0x16F87}, {0x16F8F, 0x16F92}, {0x16FE4, 0x16FE4},
359     {0x16FF0, 0x16FF1}, {0x1BC9D, 0x1BC9E}, {0x1CF00, 0x1CF2D},
360     {0x1CF30, 0x1CF46}, {0x1D165, 0x1D169}, {0x1D16D, 0x1D172},
361     {0x1D17B, 0x1D182}, {0x1D185, 0x1D18B}, {0x1D1AA, 0x1D1AD},
362     {0x1D242, 0x1D244}, {0x1D7CE, 0x1D7FF}, {0x1DA00, 0x1DA36},
363     {0x1DA3B, 0x1DA6C}, {0x1DA75, 0x1DA75}, {0x1DA84, 0x1DA84},
364     {0x1DA9B, 0x1DA9F}, {0x1DAA1, 0x1DAAF}, {0x1E000, 0x1E006},
365     {0x1E008, 0x1E018}, {0x1E01B, 0x1E021}, {0x1E023, 0x1E024},
366     {0x1E026, 0x1E02A}, {0x1E08F, 0x1E08F}, {0x1E130, 0x1E136},
367     {0x1E140, 0x1E149}, {0x1E2AE, 0x1E2AE}, {0x1E2EC, 0x1E2F9},
368     {0x1E4EC, 0x1E4F9}, {0x1E8D0, 0x1E8D6}, {0x1E944, 0x1E94A},
369     {0x1E950, 0x1E959}, {0x1FBF0, 0x1FBF9}, {0xE0100, 0xE01EF},
370 };
371 
372 // Clang supports the "Mathematical notation profile" as an extension,
373 // as described in https://www.unicode.org/L2/L2022/22230-math-profile.pdf
374 // Math_Start
375 static const llvm::sys::UnicodeCharRange
376     MathematicalNotationProfileIDStartRanges[] = {
377         {0x02202, 0x02202}, // ∂
378         {0x02207, 0x02207}, // ∇
379         {0x0221E, 0x0221E}, // ∞
380         {0x1D6C1, 0x1D6C1}, // ��
381         {0x1D6DB, 0x1D6DB}, // ��
382         {0x1D6FB, 0x1D6FB}, // ��
383         {0x1D715, 0x1D715}, // ��
384         {0x1D735, 0x1D735}, // ��
385         {0x1D74F, 0x1D74F}, // ��
386         {0x1D76F, 0x1D76F}, // ��
387         {0x1D789, 0x1D789}, // ��
388         {0x1D7A9, 0x1D7A9}, // ��
389         {0x1D7C3, 0x1D7C3}, // ��
390 };
391 
392 // Math_Continue
393 static const llvm::sys::UnicodeCharRange
394     MathematicalNotationProfileIDContinueRanges[] = {
395         {0x000B2, 0x000B3}, // ²-³
396         {0x000B9, 0x000B9}, // ¹
397         {0x02070, 0x02070}, // ⁰
398         {0x02074, 0x0207E}, // ⁴-⁾
399         {0x02080, 0x0208E}, // ₀-₎
400 };
401 
402 // C11 D.1, C++11 [charname.allowed]
403 static const llvm::sys::UnicodeCharRange C11AllowedIDCharRanges[] = {
404   // 1
405   { 0x00A8, 0x00A8 }, { 0x00AA, 0x00AA }, { 0x00AD, 0x00AD },
406   { 0x00AF, 0x00AF }, { 0x00B2, 0x00B5 }, { 0x00B7, 0x00BA },
407   { 0x00BC, 0x00BE }, { 0x00C0, 0x00D6 }, { 0x00D8, 0x00F6 },
408   { 0x00F8, 0x00FF },
409   // 2
410   { 0x0100, 0x167F }, { 0x1681, 0x180D }, { 0x180F, 0x1FFF },
411   // 3
412   { 0x200B, 0x200D }, { 0x202A, 0x202E }, { 0x203F, 0x2040 },
413   { 0x2054, 0x2054 }, { 0x2060, 0x206F },
414   // 4
415   { 0x2070, 0x218F }, { 0x2460, 0x24FF }, { 0x2776, 0x2793 },
416   { 0x2C00, 0x2DFF }, { 0x2E80, 0x2FFF },
417   // 5
418   { 0x3004, 0x3007 }, { 0x3021, 0x302F }, { 0x3031, 0x303F },
419   // 6
420   { 0x3040, 0xD7FF },
421   // 7
422   { 0xF900, 0xFD3D }, { 0xFD40, 0xFDCF }, { 0xFDF0, 0xFE44 },
423   { 0xFE47, 0xFFFD },
424   // 8
425   { 0x10000, 0x1FFFD }, { 0x20000, 0x2FFFD }, { 0x30000, 0x3FFFD },
426   { 0x40000, 0x4FFFD }, { 0x50000, 0x5FFFD }, { 0x60000, 0x6FFFD },
427   { 0x70000, 0x7FFFD }, { 0x80000, 0x8FFFD }, { 0x90000, 0x9FFFD },
428   { 0xA0000, 0xAFFFD }, { 0xB0000, 0xBFFFD }, { 0xC0000, 0xCFFFD },
429   { 0xD0000, 0xDFFFD }, { 0xE0000, 0xEFFFD }
430 };
431 
432 // C99 Annex D
433 static const llvm::sys::UnicodeCharRange C99AllowedIDCharRanges[] = {
434   // Latin (1)
435   { 0x00AA, 0x00AA },
436 
437   // Special characters (1)
438   { 0x00B5, 0x00B5 }, { 0x00B7, 0x00B7 },
439 
440   // Latin (2)
441   { 0x00BA, 0x00BA }, { 0x00C0, 0x00D6 }, { 0x00D8, 0x00F6 },
442   { 0x00F8, 0x01F5 }, { 0x01FA, 0x0217 }, { 0x0250, 0x02A8 },
443 
444   // Special characters (2)
445   { 0x02B0, 0x02B8 }, { 0x02BB, 0x02BB }, { 0x02BD, 0x02C1 },
446   { 0x02D0, 0x02D1 }, { 0x02E0, 0x02E4 }, { 0x037A, 0x037A },
447 
448   // Greek (1)
449   { 0x0386, 0x0386 }, { 0x0388, 0x038A }, { 0x038C, 0x038C },
450   { 0x038E, 0x03A1 }, { 0x03A3, 0x03CE }, { 0x03D0, 0x03D6 },
451   { 0x03DA, 0x03DA }, { 0x03DC, 0x03DC }, { 0x03DE, 0x03DE },
452   { 0x03E0, 0x03E0 }, { 0x03E2, 0x03F3 },
453 
454   // Cyrillic
455   { 0x0401, 0x040C }, { 0x040E, 0x044F }, { 0x0451, 0x045C },
456   { 0x045E, 0x0481 }, { 0x0490, 0x04C4 }, { 0x04C7, 0x04C8 },
457   { 0x04CB, 0x04CC }, { 0x04D0, 0x04EB }, { 0x04EE, 0x04F5 },
458   { 0x04F8, 0x04F9 },
459 
460   // Armenian (1)
461   { 0x0531, 0x0556 },
462 
463   // Special characters (3)
464   { 0x0559, 0x0559 },
465 
466   // Armenian (2)
467   { 0x0561, 0x0587 },
468 
469   // Hebrew
470   { 0x05B0, 0x05B9 }, { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF },
471   { 0x05C1, 0x05C2 }, { 0x05D0, 0x05EA }, { 0x05F0, 0x05F2 },
472 
473   // Arabic (1)
474   { 0x0621, 0x063A }, { 0x0640, 0x0652 },
475 
476   // Digits (1)
477   { 0x0660, 0x0669 },
478 
479   // Arabic (2)
480   { 0x0670, 0x06B7 }, { 0x06BA, 0x06BE }, { 0x06C0, 0x06CE },
481   { 0x06D0, 0x06DC }, { 0x06E5, 0x06E8 }, { 0x06EA, 0x06ED },
482 
483   // Digits (2)
484   { 0x06F0, 0x06F9 },
485 
486   // Devanagari and Special character 0x093D.
487   { 0x0901, 0x0903 }, { 0x0905, 0x0939 }, { 0x093D, 0x094D },
488   { 0x0950, 0x0952 }, { 0x0958, 0x0963 },
489 
490   // Digits (3)
491   { 0x0966, 0x096F },
492 
493   // Bengali (1)
494   { 0x0981, 0x0983 }, { 0x0985, 0x098C }, { 0x098F, 0x0990 },
495   { 0x0993, 0x09A8 }, { 0x09AA, 0x09B0 }, { 0x09B2, 0x09B2 },
496   { 0x09B6, 0x09B9 }, { 0x09BE, 0x09C4 }, { 0x09C7, 0x09C8 },
497   { 0x09CB, 0x09CD }, { 0x09DC, 0x09DD }, { 0x09DF, 0x09E3 },
498 
499   // Digits (4)
500   { 0x09E6, 0x09EF },
501 
502   // Bengali (2)
503   { 0x09F0, 0x09F1 },
504 
505   // Gurmukhi (1)
506   { 0x0A02, 0x0A02 }, { 0x0A05, 0x0A0A }, { 0x0A0F, 0x0A10 },
507   { 0x0A13, 0x0A28 }, { 0x0A2A, 0x0A30 }, { 0x0A32, 0x0A33 },
508   { 0x0A35, 0x0A36 }, { 0x0A38, 0x0A39 }, { 0x0A3E, 0x0A42 },
509   { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D }, { 0x0A59, 0x0A5C },
510   { 0x0A5E, 0x0A5E },
511 
512   // Digits (5)
513   { 0x0A66, 0x0A6F },
514 
515   // Gurmukhi (2)
516   { 0x0A74, 0x0A74 },
517 
518   // Gujarti
519   { 0x0A81, 0x0A83 }, { 0x0A85, 0x0A8B }, { 0x0A8D, 0x0A8D },
520   { 0x0A8F, 0x0A91 }, { 0x0A93, 0x0AA8 }, { 0x0AAA, 0x0AB0 },
521   { 0x0AB2, 0x0AB3 }, { 0x0AB5, 0x0AB9 }, { 0x0ABD, 0x0AC5 },
522   { 0x0AC7, 0x0AC9 }, { 0x0ACB, 0x0ACD }, { 0x0AD0, 0x0AD0 },
523   { 0x0AE0, 0x0AE0 },
524 
525   // Digits (6)
526   { 0x0AE6, 0x0AEF },
527 
528   // Oriya and Special character 0x0B3D
529   { 0x0B01, 0x0B03 }, { 0x0B05, 0x0B0C }, { 0x0B0F, 0x0B10 },
530   { 0x0B13, 0x0B28 }, { 0x0B2A, 0x0B30 }, { 0x0B32, 0x0B33 },
531   { 0x0B36, 0x0B39 }, { 0x0B3D, 0x0B43 }, { 0x0B47, 0x0B48 },
532   { 0x0B4B, 0x0B4D }, { 0x0B5C, 0x0B5D }, { 0x0B5F, 0x0B61 },
533 
534   // Digits (7)
535   { 0x0B66, 0x0B6F },
536 
537   // Tamil
538   { 0x0B82, 0x0B83 }, { 0x0B85, 0x0B8A }, { 0x0B8E, 0x0B90 },
539   { 0x0B92, 0x0B95 }, { 0x0B99, 0x0B9A }, { 0x0B9C, 0x0B9C },
540   { 0x0B9E, 0x0B9F }, { 0x0BA3, 0x0BA4 }, { 0x0BA8, 0x0BAA },
541   { 0x0BAE, 0x0BB5 }, { 0x0BB7, 0x0BB9 }, { 0x0BBE, 0x0BC2 },
542   { 0x0BC6, 0x0BC8 }, { 0x0BCA, 0x0BCD },
543 
544   // Digits (8)
545   { 0x0BE7, 0x0BEF },
546 
547   // Telugu
548   { 0x0C01, 0x0C03 }, { 0x0C05, 0x0C0C }, { 0x0C0E, 0x0C10 },
549   { 0x0C12, 0x0C28 }, { 0x0C2A, 0x0C33 }, { 0x0C35, 0x0C39 },
550   { 0x0C3E, 0x0C44 }, { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D },
551   { 0x0C60, 0x0C61 },
552 
553   // Digits (9)
554   { 0x0C66, 0x0C6F },
555 
556   // Kannada
557   { 0x0C82, 0x0C83 }, { 0x0C85, 0x0C8C }, { 0x0C8E, 0x0C90 },
558   { 0x0C92, 0x0CA8 }, { 0x0CAA, 0x0CB3 }, { 0x0CB5, 0x0CB9 },
559   { 0x0CBE, 0x0CC4 }, { 0x0CC6, 0x0CC8 }, { 0x0CCA, 0x0CCD },
560   { 0x0CDE, 0x0CDE }, { 0x0CE0, 0x0CE1 },
561 
562   // Digits (10)
563   { 0x0CE6, 0x0CEF },
564 
565   // Malayam
566   { 0x0D02, 0x0D03 }, { 0x0D05, 0x0D0C }, { 0x0D0E, 0x0D10 },
567   { 0x0D12, 0x0D28 }, { 0x0D2A, 0x0D39 }, { 0x0D3E, 0x0D43 },
568   { 0x0D46, 0x0D48 }, { 0x0D4A, 0x0D4D }, { 0x0D60, 0x0D61 },
569 
570   // Digits (11)
571   { 0x0D66, 0x0D6F },
572 
573   // Thai...including Digits { 0x0E50, 0x0E59 }
574   { 0x0E01, 0x0E3A }, { 0x0E40, 0x0E5B },
575 
576   // Lao (1)
577   { 0x0E81, 0x0E82 }, { 0x0E84, 0x0E84 }, { 0x0E87, 0x0E88 },
578   { 0x0E8A, 0x0E8A }, { 0x0E8D, 0x0E8D }, { 0x0E94, 0x0E97 },
579   { 0x0E99, 0x0E9F }, { 0x0EA1, 0x0EA3 }, { 0x0EA5, 0x0EA5 },
580   { 0x0EA7, 0x0EA7 }, { 0x0EAA, 0x0EAB }, { 0x0EAD, 0x0EAE },
581   { 0x0EB0, 0x0EB9 }, { 0x0EBB, 0x0EBD }, { 0x0EC0, 0x0EC4 },
582   { 0x0EC6, 0x0EC6 }, { 0x0EC8, 0x0ECD },
583 
584   // Digits (12)
585   { 0x0ED0, 0x0ED9 },
586 
587   // Lao (2)
588   { 0x0EDC, 0x0EDD },
589 
590   // Tibetan (1)
591   { 0x0F00, 0x0F00 }, { 0x0F18, 0x0F19 },
592 
593   // Digits (13)
594   { 0x0F20, 0x0F33 },
595 
596   // Tibetan (2)
597   { 0x0F35, 0x0F35 }, { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 },
598   { 0x0F3E, 0x0F47 }, { 0x0F49, 0x0F69 }, { 0x0F71, 0x0F84 },
599   { 0x0F86, 0x0F8B }, { 0x0F90, 0x0F95 }, { 0x0F97, 0x0F97 },
600   { 0x0F99, 0x0FAD }, { 0x0FB1, 0x0FB7 }, { 0x0FB9, 0x0FB9 },
601 
602   // Georgian
603   { 0x10A0, 0x10C5 }, { 0x10D0, 0x10F6 },
604 
605   // Latin (3)
606   { 0x1E00, 0x1E9B }, { 0x1EA0, 0x1EF9 },
607 
608   // Greek (2)
609   { 0x1F00, 0x1F15 }, { 0x1F18, 0x1F1D }, { 0x1F20, 0x1F45 },
610   { 0x1F48, 0x1F4D }, { 0x1F50, 0x1F57 }, { 0x1F59, 0x1F59 },
611   { 0x1F5B, 0x1F5B }, { 0x1F5D, 0x1F5D }, { 0x1F5F, 0x1F7D },
612   { 0x1F80, 0x1FB4 }, { 0x1FB6, 0x1FBC },
613 
614   // Special characters (4)
615   { 0x1FBE, 0x1FBE },
616 
617   // Greek (3)
618   { 0x1FC2, 0x1FC4 }, { 0x1FC6, 0x1FCC }, { 0x1FD0, 0x1FD3 },
619   { 0x1FD6, 0x1FDB }, { 0x1FE0, 0x1FEC }, { 0x1FF2, 0x1FF4 },
620   { 0x1FF6, 0x1FFC },
621 
622   // Special characters (5)
623   { 0x203F, 0x2040 },
624 
625   // Latin (4)
626   { 0x207F, 0x207F },
627 
628   // Special characters (6)
629   { 0x2102, 0x2102 }, { 0x2107, 0x2107 }, { 0x210A, 0x2113 },
630   { 0x2115, 0x2115 }, { 0x2118, 0x211D }, { 0x2124, 0x2124 },
631   { 0x2126, 0x2126 }, { 0x2128, 0x2128 }, { 0x212A, 0x2131 },
632   { 0x2133, 0x2138 }, { 0x2160, 0x2182 }, { 0x3005, 0x3007 },
633   { 0x3021, 0x3029 },
634 
635   // Hiragana
636   { 0x3041, 0x3093 }, { 0x309B, 0x309C },
637 
638   // Katakana
639   { 0x30A1, 0x30F6 }, { 0x30FB, 0x30FC },
640 
641   // Bopmofo [sic]
642   { 0x3105, 0x312C },
643 
644   // CJK Unified Ideographs
645   { 0x4E00, 0x9FA5 },
646 
647   // Hangul,
648   { 0xAC00, 0xD7A3 }
649 };
650 
651 // C11 D.2, C++11 [charname.disallowed]
652 static const llvm::sys::UnicodeCharRange C11DisallowedInitialIDCharRanges[] = {
653   { 0x0300, 0x036F }, { 0x1DC0, 0x1DFF }, { 0x20D0, 0x20FF },
654   { 0xFE20, 0xFE2F }
655 };
656 
657 // C99 6.4.2.1p3: The initial character [of an identifier] shall not be a
658 // universal character name designating a digit.
659 // C99 Annex D defines these characters as "Digits".
660 static const llvm::sys::UnicodeCharRange C99DisallowedInitialIDCharRanges[] = {
661   { 0x0660, 0x0669 }, { 0x06F0, 0x06F9 }, { 0x0966, 0x096F },
662   { 0x09E6, 0x09EF }, { 0x0A66, 0x0A6F }, { 0x0AE6, 0x0AEF },
663   { 0x0B66, 0x0B6F }, { 0x0BE7, 0x0BEF }, { 0x0C66, 0x0C6F },
664   { 0x0CE6, 0x0CEF }, { 0x0D66, 0x0D6F }, { 0x0E50, 0x0E59 },
665   { 0x0ED0, 0x0ED9 }, { 0x0F20, 0x0F33 }
666 };
667 
668 // Unicode v6.2, chapter 6.2, table 6-2.
669 static const llvm::sys::UnicodeCharRange UnicodeWhitespaceCharRanges[] = {
670   { 0x0085, 0x0085 }, { 0x00A0, 0x00A0 }, { 0x1680, 0x1680 },
671   { 0x180E, 0x180E }, { 0x2000, 0x200A }, { 0x2028, 0x2029 },
672   { 0x202F, 0x202F }, { 0x205F, 0x205F }, { 0x3000, 0x3000 }
673 };
674 
675 #endif
676