1259698Sdim//===- llvm/Support/Unicode.cpp - Unicode character properties  -*- C++ -*-===//
2259698Sdim//
3259698Sdim//                     The LLVM Compiler Infrastructure
4259698Sdim//
5259698Sdim// This file is distributed under the University of Illinois Open Source
6259698Sdim// License. See LICENSE.TXT for details.
7259698Sdim//
8259698Sdim//===----------------------------------------------------------------------===//
9259698Sdim//
10259698Sdim// This file implements functions that allow querying certain properties of
11259698Sdim// Unicode characters.
12259698Sdim//
13259698Sdim//===----------------------------------------------------------------------===//
14259698Sdim
15259698Sdim#include "llvm/Support/Unicode.h"
16259698Sdim#include "llvm/Support/ConvertUTF.h"
17259698Sdim#include "llvm/Support/UnicodeCharRanges.h"
18259698Sdim
19259698Sdimnamespace llvm {
20259698Sdimnamespace sys {
21259698Sdimnamespace unicode {
22259698Sdim
23259698Sdimbool isPrintable(int UCS) {
24259698Sdim  // Sorted list of non-overlapping intervals of code points that are not
25259698Sdim  // supposed to be printable.
26259698Sdim  static const UnicodeCharRange NonPrintableRanges[] = {
27259698Sdim    { 0x0000, 0x001F }, { 0x007F, 0x009F }, { 0x034F, 0x034F },
28259698Sdim    { 0x0378, 0x0379 }, { 0x037F, 0x0383 }, { 0x038B, 0x038B },
29259698Sdim    { 0x038D, 0x038D }, { 0x03A2, 0x03A2 }, { 0x0528, 0x0530 },
30259698Sdim    { 0x0557, 0x0558 }, { 0x0560, 0x0560 }, { 0x0588, 0x0588 },
31259698Sdim    { 0x058B, 0x058E }, { 0x0590, 0x0590 }, { 0x05C8, 0x05CF },
32259698Sdim    { 0x05EB, 0x05EF }, { 0x05F5, 0x0605 }, { 0x061C, 0x061D },
33259698Sdim    { 0x06DD, 0x06DD }, { 0x070E, 0x070F }, { 0x074B, 0x074C },
34259698Sdim    { 0x07B2, 0x07BF }, { 0x07FB, 0x07FF }, { 0x082E, 0x082F },
35259698Sdim    { 0x083F, 0x083F }, { 0x085C, 0x085D }, { 0x085F, 0x089F },
36259698Sdim    { 0x08A1, 0x08A1 }, { 0x08AD, 0x08E3 }, { 0x08FF, 0x08FF },
37259698Sdim    { 0x0978, 0x0978 }, { 0x0980, 0x0980 }, { 0x0984, 0x0984 },
38259698Sdim    { 0x098D, 0x098E }, { 0x0991, 0x0992 }, { 0x09A9, 0x09A9 },
39259698Sdim    { 0x09B1, 0x09B1 }, { 0x09B3, 0x09B5 }, { 0x09BA, 0x09BB },
40259698Sdim    { 0x09C5, 0x09C6 }, { 0x09C9, 0x09CA }, { 0x09CF, 0x09D6 },
41259698Sdim    { 0x09D8, 0x09DB }, { 0x09DE, 0x09DE }, { 0x09E4, 0x09E5 },
42259698Sdim    { 0x09FC, 0x0A00 }, { 0x0A04, 0x0A04 }, { 0x0A0B, 0x0A0E },
43259698Sdim    { 0x0A11, 0x0A12 }, { 0x0A29, 0x0A29 }, { 0x0A31, 0x0A31 },
44259698Sdim    { 0x0A34, 0x0A34 }, { 0x0A37, 0x0A37 }, { 0x0A3A, 0x0A3B },
45259698Sdim    { 0x0A3D, 0x0A3D }, { 0x0A43, 0x0A46 }, { 0x0A49, 0x0A4A },
46259698Sdim    { 0x0A4E, 0x0A50 }, { 0x0A52, 0x0A58 }, { 0x0A5D, 0x0A5D },
47259698Sdim    { 0x0A5F, 0x0A65 }, { 0x0A76, 0x0A80 }, { 0x0A84, 0x0A84 },
48259698Sdim    { 0x0A8E, 0x0A8E }, { 0x0A92, 0x0A92 }, { 0x0AA9, 0x0AA9 },
49259698Sdim    { 0x0AB1, 0x0AB1 }, { 0x0AB4, 0x0AB4 }, { 0x0ABA, 0x0ABB },
50259698Sdim    { 0x0AC6, 0x0AC6 }, { 0x0ACA, 0x0ACA }, { 0x0ACE, 0x0ACF },
51259698Sdim    { 0x0AD1, 0x0ADF }, { 0x0AE4, 0x0AE5 }, { 0x0AF2, 0x0B00 },
52259698Sdim    { 0x0B04, 0x0B04 }, { 0x0B0D, 0x0B0E }, { 0x0B11, 0x0B12 },
53259698Sdim    { 0x0B29, 0x0B29 }, { 0x0B31, 0x0B31 }, { 0x0B34, 0x0B34 },
54259698Sdim    { 0x0B3A, 0x0B3B }, { 0x0B45, 0x0B46 }, { 0x0B49, 0x0B4A },
55259698Sdim    { 0x0B4E, 0x0B55 }, { 0x0B58, 0x0B5B }, { 0x0B5E, 0x0B5E },
56259698Sdim    { 0x0B64, 0x0B65 }, { 0x0B78, 0x0B81 }, { 0x0B84, 0x0B84 },
57259698Sdim    { 0x0B8B, 0x0B8D }, { 0x0B91, 0x0B91 }, { 0x0B96, 0x0B98 },
58259698Sdim    { 0x0B9B, 0x0B9B }, { 0x0B9D, 0x0B9D }, { 0x0BA0, 0x0BA2 },
59259698Sdim    { 0x0BA5, 0x0BA7 }, { 0x0BAB, 0x0BAD }, { 0x0BBA, 0x0BBD },
60259698Sdim    { 0x0BC3, 0x0BC5 }, { 0x0BC9, 0x0BC9 }, { 0x0BCE, 0x0BCF },
61259698Sdim    { 0x0BD1, 0x0BD6 }, { 0x0BD8, 0x0BE5 }, { 0x0BFB, 0x0C00 },
62259698Sdim    { 0x0C04, 0x0C04 }, { 0x0C0D, 0x0C0D }, { 0x0C11, 0x0C11 },
63259698Sdim    { 0x0C29, 0x0C29 }, { 0x0C34, 0x0C34 }, { 0x0C3A, 0x0C3C },
64259698Sdim    { 0x0C45, 0x0C45 }, { 0x0C49, 0x0C49 }, { 0x0C4E, 0x0C54 },
65259698Sdim    { 0x0C57, 0x0C57 }, { 0x0C5A, 0x0C5F }, { 0x0C64, 0x0C65 },
66259698Sdim    { 0x0C70, 0x0C77 }, { 0x0C80, 0x0C81 }, { 0x0C84, 0x0C84 },
67259698Sdim    { 0x0C8D, 0x0C8D }, { 0x0C91, 0x0C91 }, { 0x0CA9, 0x0CA9 },
68259698Sdim    { 0x0CB4, 0x0CB4 }, { 0x0CBA, 0x0CBB }, { 0x0CC5, 0x0CC5 },
69259698Sdim    { 0x0CC9, 0x0CC9 }, { 0x0CCE, 0x0CD4 }, { 0x0CD7, 0x0CDD },
70259698Sdim    { 0x0CDF, 0x0CDF }, { 0x0CE4, 0x0CE5 }, { 0x0CF0, 0x0CF0 },
71259698Sdim    { 0x0CF3, 0x0D01 }, { 0x0D04, 0x0D04 }, { 0x0D0D, 0x0D0D },
72259698Sdim    { 0x0D11, 0x0D11 }, { 0x0D3B, 0x0D3C }, { 0x0D45, 0x0D45 },
73259698Sdim    { 0x0D49, 0x0D49 }, { 0x0D4F, 0x0D56 }, { 0x0D58, 0x0D5F },
74259698Sdim    { 0x0D64, 0x0D65 }, { 0x0D76, 0x0D78 }, { 0x0D80, 0x0D81 },
75259698Sdim    { 0x0D84, 0x0D84 }, { 0x0D97, 0x0D99 }, { 0x0DB2, 0x0DB2 },
76259698Sdim    { 0x0DBC, 0x0DBC }, { 0x0DBE, 0x0DBF }, { 0x0DC7, 0x0DC9 },
77259698Sdim    { 0x0DCB, 0x0DCE }, { 0x0DD5, 0x0DD5 }, { 0x0DD7, 0x0DD7 },
78259698Sdim    { 0x0DE0, 0x0DF1 }, { 0x0DF5, 0x0E00 }, { 0x0E3B, 0x0E3E },
79259698Sdim    { 0x0E5C, 0x0E80 }, { 0x0E83, 0x0E83 }, { 0x0E85, 0x0E86 },
80259698Sdim    { 0x0E89, 0x0E89 }, { 0x0E8B, 0x0E8C }, { 0x0E8E, 0x0E93 },
81259698Sdim    { 0x0E98, 0x0E98 }, { 0x0EA0, 0x0EA0 }, { 0x0EA4, 0x0EA4 },
82259698Sdim    { 0x0EA6, 0x0EA6 }, { 0x0EA8, 0x0EA9 }, { 0x0EAC, 0x0EAC },
83259698Sdim    { 0x0EBA, 0x0EBA }, { 0x0EBE, 0x0EBF }, { 0x0EC5, 0x0EC5 },
84259698Sdim    { 0x0EC7, 0x0EC7 }, { 0x0ECE, 0x0ECF }, { 0x0EDA, 0x0EDB },
85259698Sdim    { 0x0EE0, 0x0EFF }, { 0x0F48, 0x0F48 }, { 0x0F6D, 0x0F70 },
86259698Sdim    { 0x0F98, 0x0F98 }, { 0x0FBD, 0x0FBD }, { 0x0FCD, 0x0FCD },
87259698Sdim    { 0x0FDB, 0x0FFF }, { 0x10C6, 0x10C6 }, { 0x10C8, 0x10CC },
88259698Sdim    { 0x10CE, 0x10CF }, { 0x115F, 0x1160 }, { 0x1249, 0x1249 },
89259698Sdim    { 0x124E, 0x124F }, { 0x1257, 0x1257 }, { 0x1259, 0x1259 },
90259698Sdim    { 0x125E, 0x125F }, { 0x1289, 0x1289 }, { 0x128E, 0x128F },
91259698Sdim    { 0x12B1, 0x12B1 }, { 0x12B6, 0x12B7 }, { 0x12BF, 0x12BF },
92259698Sdim    { 0x12C1, 0x12C1 }, { 0x12C6, 0x12C7 }, { 0x12D7, 0x12D7 },
93259698Sdim    { 0x1311, 0x1311 }, { 0x1316, 0x1317 }, { 0x135B, 0x135C },
94259698Sdim    { 0x137D, 0x137F }, { 0x139A, 0x139F }, { 0x13F5, 0x13FF },
95259698Sdim    { 0x169D, 0x169F }, { 0x16F1, 0x16FF }, { 0x170D, 0x170D },
96259698Sdim    { 0x1715, 0x171F }, { 0x1737, 0x173F }, { 0x1754, 0x175F },
97259698Sdim    { 0x176D, 0x176D }, { 0x1771, 0x1771 }, { 0x1774, 0x177F },
98259698Sdim    { 0x17B4, 0x17B5 }, { 0x17DE, 0x17DF }, { 0x17EA, 0x17EF },
99259698Sdim    { 0x17FA, 0x17FF }, { 0x180B, 0x180D }, { 0x180F, 0x180F },
100259698Sdim    { 0x181A, 0x181F }, { 0x1878, 0x187F }, { 0x18AB, 0x18AF },
101259698Sdim    { 0x18F6, 0x18FF }, { 0x191D, 0x191F }, { 0x192C, 0x192F },
102259698Sdim    { 0x193C, 0x193F }, { 0x1941, 0x1943 }, { 0x196E, 0x196F },
103259698Sdim    { 0x1975, 0x197F }, { 0x19AC, 0x19AF }, { 0x19CA, 0x19CF },
104259698Sdim    { 0x19DB, 0x19DD }, { 0x1A1C, 0x1A1D }, { 0x1A5F, 0x1A5F },
105259698Sdim    { 0x1A7D, 0x1A7E }, { 0x1A8A, 0x1A8F }, { 0x1A9A, 0x1A9F },
106259698Sdim    { 0x1AAE, 0x1AFF }, { 0x1B4C, 0x1B4F }, { 0x1B7D, 0x1B7F },
107259698Sdim    { 0x1BF4, 0x1BFB }, { 0x1C38, 0x1C3A }, { 0x1C4A, 0x1C4C },
108259698Sdim    { 0x1C80, 0x1CBF }, { 0x1CC8, 0x1CCF }, { 0x1CF7, 0x1CFF },
109259698Sdim    { 0x1DE7, 0x1DFB }, { 0x1F16, 0x1F17 }, { 0x1F1E, 0x1F1F },
110259698Sdim    { 0x1F46, 0x1F47 }, { 0x1F4E, 0x1F4F }, { 0x1F58, 0x1F58 },
111259698Sdim    { 0x1F5A, 0x1F5A }, { 0x1F5C, 0x1F5C }, { 0x1F5E, 0x1F5E },
112259698Sdim    { 0x1F7E, 0x1F7F }, { 0x1FB5, 0x1FB5 }, { 0x1FC5, 0x1FC5 },
113259698Sdim    { 0x1FD4, 0x1FD5 }, { 0x1FDC, 0x1FDC }, { 0x1FF0, 0x1FF1 },
114259698Sdim    { 0x1FF5, 0x1FF5 }, { 0x1FFF, 0x1FFF }, { 0x200B, 0x200F },
115259698Sdim    { 0x202A, 0x202E }, { 0x2060, 0x206F }, { 0x2072, 0x2073 },
116259698Sdim    { 0x208F, 0x208F }, { 0x209D, 0x209F }, { 0x20BB, 0x20CF },
117259698Sdim    { 0x20F1, 0x20FF }, { 0x218A, 0x218F }, { 0x23F4, 0x23FF },
118259698Sdim    { 0x2427, 0x243F }, { 0x244B, 0x245F }, { 0x2700, 0x2700 },
119259698Sdim    { 0x2B4D, 0x2B4F }, { 0x2B5A, 0x2BFF }, { 0x2C2F, 0x2C2F },
120259698Sdim    { 0x2C5F, 0x2C5F }, { 0x2CF4, 0x2CF8 }, { 0x2D26, 0x2D26 },
121259698Sdim    { 0x2D28, 0x2D2C }, { 0x2D2E, 0x2D2F }, { 0x2D68, 0x2D6E },
122259698Sdim    { 0x2D71, 0x2D7E }, { 0x2D97, 0x2D9F }, { 0x2DA7, 0x2DA7 },
123259698Sdim    { 0x2DAF, 0x2DAF }, { 0x2DB7, 0x2DB7 }, { 0x2DBF, 0x2DBF },
124259698Sdim    { 0x2DC7, 0x2DC7 }, { 0x2DCF, 0x2DCF }, { 0x2DD7, 0x2DD7 },
125259698Sdim    { 0x2DDF, 0x2DDF }, { 0x2E3C, 0x2E7F }, { 0x2E9A, 0x2E9A },
126259698Sdim    { 0x2EF4, 0x2EFF }, { 0x2FD6, 0x2FEF }, { 0x2FFC, 0x2FFF },
127259698Sdim    { 0x3040, 0x3040 }, { 0x3097, 0x3098 }, { 0x3100, 0x3104 },
128259698Sdim    { 0x312E, 0x3130 }, { 0x3164, 0x3164 }, { 0x318F, 0x318F },
129259698Sdim    { 0x31BB, 0x31BF }, { 0x31E4, 0x31EF }, { 0x321F, 0x321F },
130259698Sdim    { 0x32FF, 0x32FF }, { 0x4DB6, 0x4DBF }, { 0x9FCD, 0x9FFF },
131259698Sdim    { 0xA48D, 0xA48F }, { 0xA4C7, 0xA4CF }, { 0xA62C, 0xA63F },
132259698Sdim    { 0xA698, 0xA69E }, { 0xA6F8, 0xA6FF }, { 0xA78F, 0xA78F },
133259698Sdim    { 0xA794, 0xA79F }, { 0xA7AB, 0xA7F7 }, { 0xA82C, 0xA82F },
134259698Sdim    { 0xA83A, 0xA83F }, { 0xA878, 0xA87F }, { 0xA8C5, 0xA8CD },
135259698Sdim    { 0xA8DA, 0xA8DF }, { 0xA8FC, 0xA8FF }, { 0xA954, 0xA95E },
136259698Sdim    { 0xA97D, 0xA97F }, { 0xA9CE, 0xA9CE }, { 0xA9DA, 0xA9DD },
137259698Sdim    { 0xA9E0, 0xA9FF }, { 0xAA37, 0xAA3F }, { 0xAA4E, 0xAA4F },
138259698Sdim    { 0xAA5A, 0xAA5B }, { 0xAA7C, 0xAA7F }, { 0xAAC3, 0xAADA },
139259698Sdim    { 0xAAF7, 0xAB00 }, { 0xAB07, 0xAB08 }, { 0xAB0F, 0xAB10 },
140259698Sdim    { 0xAB17, 0xAB1F }, { 0xAB27, 0xAB27 }, { 0xAB2F, 0xABBF },
141259698Sdim    { 0xABEE, 0xABEF }, { 0xABFA, 0xABFF }, { 0xD7A4, 0xD7AF },
142259698Sdim    { 0xD7C7, 0xD7CA }, { 0xD7FC, 0xDFFF }, { 0xFA6E, 0xFA6F },
143259698Sdim    { 0xFADA, 0xFAFF }, { 0xFB07, 0xFB12 }, { 0xFB18, 0xFB1C },
144259698Sdim    { 0xFB37, 0xFB37 }, { 0xFB3D, 0xFB3D }, { 0xFB3F, 0xFB3F },
145259698Sdim    { 0xFB42, 0xFB42 }, { 0xFB45, 0xFB45 }, { 0xFBC2, 0xFBD2 },
146259698Sdim    { 0xFD40, 0xFD4F }, { 0xFD90, 0xFD91 }, { 0xFDC8, 0xFDEF },
147259698Sdim    { 0xFDFE, 0xFE0F }, { 0xFE1A, 0xFE1F }, { 0xFE27, 0xFE2F },
148259698Sdim    { 0xFE53, 0xFE53 }, { 0xFE67, 0xFE67 }, { 0xFE6C, 0xFE6F },
149259698Sdim    { 0xFE75, 0xFE75 }, { 0xFEFD, 0xFEFF }, { 0xFF00, 0xFF00 },
150259698Sdim    { 0xFFA0, 0xFFA0 }, { 0xFFBF, 0xFFC1 }, { 0xFFC8, 0xFFC9 },
151259698Sdim    { 0xFFD0, 0xFFD1 }, { 0xFFD8, 0xFFD9 }, { 0xFFDD, 0xFFDF },
152259698Sdim    { 0xFFE7, 0xFFE7 }, { 0xFFEF, 0xFFFB }, { 0xFFFE, 0xFFFF },
153259698Sdim    { 0x1000C, 0x1000C }, { 0x10027, 0x10027 }, { 0x1003B, 0x1003B },
154259698Sdim    { 0x1003E, 0x1003E }, { 0x1004E, 0x1004F }, { 0x1005E, 0x1007F },
155259698Sdim    { 0x100FB, 0x100FF }, { 0x10103, 0x10106 }, { 0x10134, 0x10136 },
156259698Sdim    { 0x1018B, 0x1018F }, { 0x1019C, 0x101CF }, { 0x101FE, 0x1027F },
157259698Sdim    { 0x1029D, 0x1029F }, { 0x102D1, 0x102FF }, { 0x1031F, 0x1031F },
158259698Sdim    { 0x10324, 0x1032F }, { 0x1034B, 0x1037F }, { 0x1039E, 0x1039E },
159259698Sdim    { 0x103C4, 0x103C7 }, { 0x103D6, 0x103FF }, { 0x1049E, 0x1049F },
160259698Sdim    { 0x104AA, 0x107FF }, { 0x10806, 0x10807 }, { 0x10809, 0x10809 },
161259698Sdim    { 0x10836, 0x10836 }, { 0x10839, 0x1083B }, { 0x1083D, 0x1083E },
162259698Sdim    { 0x10856, 0x10856 }, { 0x10860, 0x108FF }, { 0x1091C, 0x1091E },
163259698Sdim    { 0x1093A, 0x1093E }, { 0x10940, 0x1097F }, { 0x109B8, 0x109BD },
164259698Sdim    { 0x109C0, 0x109FF }, { 0x10A04, 0x10A04 }, { 0x10A07, 0x10A0B },
165259698Sdim    { 0x10A14, 0x10A14 }, { 0x10A18, 0x10A18 }, { 0x10A34, 0x10A37 },
166259698Sdim    { 0x10A3B, 0x10A3E }, { 0x10A48, 0x10A4F }, { 0x10A59, 0x10A5F },
167259698Sdim    { 0x10A80, 0x10AFF }, { 0x10B36, 0x10B38 }, { 0x10B56, 0x10B57 },
168259698Sdim    { 0x10B73, 0x10B77 }, { 0x10B80, 0x10BFF }, { 0x10C49, 0x10E5F },
169259698Sdim    { 0x10E7F, 0x10FFF }, { 0x1104E, 0x11051 }, { 0x11070, 0x1107F },
170259698Sdim    { 0x110BD, 0x110BD }, { 0x110C2, 0x110CF }, { 0x110E9, 0x110EF },
171259698Sdim    { 0x110FA, 0x110FF }, { 0x11135, 0x11135 }, { 0x11144, 0x1117F },
172259698Sdim    { 0x111C9, 0x111CF }, { 0x111DA, 0x1167F }, { 0x116B8, 0x116BF },
173259698Sdim    { 0x116CA, 0x11FFF }, { 0x1236F, 0x123FF }, { 0x12463, 0x1246F },
174259698Sdim    { 0x12474, 0x12FFF }, { 0x1342F, 0x167FF }, { 0x16A39, 0x16EFF },
175259698Sdim    { 0x16F45, 0x16F4F }, { 0x16F7F, 0x16F8E }, { 0x16FA0, 0x1AFFF },
176259698Sdim    { 0x1B002, 0x1CFFF }, { 0x1D0F6, 0x1D0FF }, { 0x1D127, 0x1D128 },
177259698Sdim    { 0x1D173, 0x1D17A }, { 0x1D1DE, 0x1D1FF }, { 0x1D246, 0x1D2FF },
178259698Sdim    { 0x1D357, 0x1D35F }, { 0x1D372, 0x1D3FF }, { 0x1D455, 0x1D455 },
179259698Sdim    { 0x1D49D, 0x1D49D }, { 0x1D4A0, 0x1D4A1 }, { 0x1D4A3, 0x1D4A4 },
180259698Sdim    { 0x1D4A7, 0x1D4A8 }, { 0x1D4AD, 0x1D4AD }, { 0x1D4BA, 0x1D4BA },
181259698Sdim    { 0x1D4BC, 0x1D4BC }, { 0x1D4C4, 0x1D4C4 }, { 0x1D506, 0x1D506 },
182259698Sdim    { 0x1D50B, 0x1D50C }, { 0x1D515, 0x1D515 }, { 0x1D51D, 0x1D51D },
183259698Sdim    { 0x1D53A, 0x1D53A }, { 0x1D53F, 0x1D53F }, { 0x1D545, 0x1D545 },
184259698Sdim    { 0x1D547, 0x1D549 }, { 0x1D551, 0x1D551 }, { 0x1D6A6, 0x1D6A7 },
185259698Sdim    { 0x1D7CC, 0x1D7CD }, { 0x1D800, 0x1EDFF }, { 0x1EE04, 0x1EE04 },
186259698Sdim    { 0x1EE20, 0x1EE20 }, { 0x1EE23, 0x1EE23 }, { 0x1EE25, 0x1EE26 },
187259698Sdim    { 0x1EE28, 0x1EE28 }, { 0x1EE33, 0x1EE33 }, { 0x1EE38, 0x1EE38 },
188259698Sdim    { 0x1EE3A, 0x1EE3A }, { 0x1EE3C, 0x1EE41 }, { 0x1EE43, 0x1EE46 },
189259698Sdim    { 0x1EE48, 0x1EE48 }, { 0x1EE4A, 0x1EE4A }, { 0x1EE4C, 0x1EE4C },
190259698Sdim    { 0x1EE50, 0x1EE50 }, { 0x1EE53, 0x1EE53 }, { 0x1EE55, 0x1EE56 },
191259698Sdim    { 0x1EE58, 0x1EE58 }, { 0x1EE5A, 0x1EE5A }, { 0x1EE5C, 0x1EE5C },
192259698Sdim    { 0x1EE5E, 0x1EE5E }, { 0x1EE60, 0x1EE60 }, { 0x1EE63, 0x1EE63 },
193259698Sdim    { 0x1EE65, 0x1EE66 }, { 0x1EE6B, 0x1EE6B }, { 0x1EE73, 0x1EE73 },
194259698Sdim    { 0x1EE78, 0x1EE78 }, { 0x1EE7D, 0x1EE7D }, { 0x1EE7F, 0x1EE7F },
195259698Sdim    { 0x1EE8A, 0x1EE8A }, { 0x1EE9C, 0x1EEA0 }, { 0x1EEA4, 0x1EEA4 },
196259698Sdim    { 0x1EEAA, 0x1EEAA }, { 0x1EEBC, 0x1EEEF }, { 0x1EEF2, 0x1EFFF },
197259698Sdim    { 0x1F02C, 0x1F02F }, { 0x1F094, 0x1F09F }, { 0x1F0AF, 0x1F0B0 },
198259698Sdim    { 0x1F0BF, 0x1F0C0 }, { 0x1F0D0, 0x1F0D0 }, { 0x1F0E0, 0x1F0FF },
199259698Sdim    { 0x1F10B, 0x1F10F }, { 0x1F12F, 0x1F12F }, { 0x1F16C, 0x1F16F },
200259698Sdim    { 0x1F19B, 0x1F1E5 }, { 0x1F203, 0x1F20F }, { 0x1F23B, 0x1F23F },
201259698Sdim    { 0x1F249, 0x1F24F }, { 0x1F252, 0x1F2FF }, { 0x1F321, 0x1F32F },
202259698Sdim    { 0x1F336, 0x1F336 }, { 0x1F37D, 0x1F37F }, { 0x1F394, 0x1F39F },
203259698Sdim    { 0x1F3C5, 0x1F3C5 }, { 0x1F3CB, 0x1F3DF }, { 0x1F3F1, 0x1F3FF },
204259698Sdim    { 0x1F43F, 0x1F43F }, { 0x1F441, 0x1F441 }, { 0x1F4F8, 0x1F4F8 },
205259698Sdim    { 0x1F4FD, 0x1F4FF }, { 0x1F53E, 0x1F53F }, { 0x1F544, 0x1F54F },
206259698Sdim    { 0x1F568, 0x1F5FA }, { 0x1F641, 0x1F644 }, { 0x1F650, 0x1F67F },
207259698Sdim    { 0x1F6C6, 0x1F6FF }, { 0x1F774, 0x1FFFF }, { 0x2A6D7, 0x2A6FF },
208259698Sdim    { 0x2B735, 0x2B73F }, { 0x2B81E, 0x2F7FF }, { 0x2FA1E, 0xF0000 },
209259698Sdim    { 0xFFFFE, 0xFFFFF }, { 0x10FFFE, 0x10FFFF }
210259698Sdim  };
211259698Sdim  static const UnicodeCharSet NonPrintables(NonPrintableRanges);
212259698Sdim
213259698Sdim  return UCS >= 0 && UCS <= 0x10FFFF && !NonPrintables.contains(UCS);
214259698Sdim}
215259698Sdim
216259698Sdim/// Gets the number of positions a character is likely to occupy when output
217259698Sdim/// on a terminal ("character width"). This depends on the implementation of the
218259698Sdim/// terminal, and there's no standard definition of character width.
219259698Sdim/// The implementation defines it in a way that is expected to be compatible
220259698Sdim/// with a generic Unicode-capable terminal.
221259698Sdim/// \return Character width:
222259698Sdim///   * ErrorNonPrintableCharacter (-1) for non-printable characters (as
223259698Sdim///     identified by isPrintable);
224259698Sdim///   * 0 for non-spacing and enclosing combining marks;
225259698Sdim///   * 2 for CJK characters excluding halfwidth forms;
226259698Sdim///   * 1 for all remaining characters.
227259698Sdimstatic inline int charWidth(int UCS)
228259698Sdim{
229259698Sdim  if (!isPrintable(UCS))
230259698Sdim    return ErrorNonPrintableCharacter;
231259698Sdim
232259698Sdim  // Sorted list of non-spacing and enclosing combining mark intervals as
233259698Sdim  // defined in "3.6 Combination" of
234259698Sdim  // http://www.unicode.org/versions/Unicode6.2.0/UnicodeStandard-6.2.pdf
235259698Sdim  static const UnicodeCharRange CombiningCharacterRanges[] = {
236259698Sdim    { 0x0300, 0x036F }, { 0x0483, 0x0489 }, { 0x0591, 0x05BD },
237259698Sdim    { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 }, { 0x05C4, 0x05C5 },
238259698Sdim    { 0x05C7, 0x05C7 }, { 0x0610, 0x061A }, { 0x064B, 0x065F },
239259698Sdim    { 0x0670, 0x0670 }, { 0x06D6, 0x06DC }, { 0x06DF, 0x06E4 },
240259698Sdim    { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, { 0x0711, 0x0711 },
241259698Sdim    { 0x0730, 0x074A }, { 0x07A6, 0x07B0 }, { 0x07EB, 0x07F3 },
242259698Sdim    { 0x0816, 0x0819 }, { 0x081B, 0x0823 }, { 0x0825, 0x0827 },
243259698Sdim    { 0x0829, 0x082D }, { 0x0859, 0x085B }, { 0x08E4, 0x08FE },
244259698Sdim    { 0x0900, 0x0902 }, { 0x093A, 0x093A }, { 0x093C, 0x093C },
245259698Sdim    { 0x0941, 0x0948 }, { 0x094D, 0x094D }, { 0x0951, 0x0957 },
246259698Sdim    { 0x0962, 0x0963 }, { 0x0981, 0x0981 }, { 0x09BC, 0x09BC },
247259698Sdim    { 0x09C1, 0x09C4 }, { 0x09CD, 0x09CD }, { 0x09E2, 0x09E3 },
248259698Sdim    { 0x0A01, 0x0A02 }, { 0x0A3C, 0x0A3C }, { 0x0A41, 0x0A42 },
249259698Sdim    { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D }, { 0x0A51, 0x0A51 },
250259698Sdim    { 0x0A70, 0x0A71 }, { 0x0A75, 0x0A75 }, { 0x0A81, 0x0A82 },
251259698Sdim    { 0x0ABC, 0x0ABC }, { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 },
252259698Sdim    { 0x0ACD, 0x0ACD }, { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 },
253259698Sdim    { 0x0B3C, 0x0B3C }, { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B44 },
254259698Sdim    { 0x0B4D, 0x0B4D }, { 0x0B56, 0x0B56 }, { 0x0B62, 0x0B63 },
255259698Sdim    { 0x0B82, 0x0B82 }, { 0x0BC0, 0x0BC0 }, { 0x0BCD, 0x0BCD },
256259698Sdim    { 0x0C3E, 0x0C40 }, { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D },
257259698Sdim    { 0x0C55, 0x0C56 }, { 0x0C62, 0x0C63 }, { 0x0CBC, 0x0CBC },
258259698Sdim    { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 }, { 0x0CCC, 0x0CCD },
259259698Sdim    { 0x0CE2, 0x0CE3 }, { 0x0D41, 0x0D44 }, { 0x0D4D, 0x0D4D },
260259698Sdim    { 0x0D62, 0x0D63 }, { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 },
261259698Sdim    { 0x0DD6, 0x0DD6 }, { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A },
262259698Sdim    { 0x0E47, 0x0E4E }, { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 },
263259698Sdim    { 0x0EBB, 0x0EBC }, { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 },
264259698Sdim    { 0x0F35, 0x0F35 }, { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 },
265259698Sdim    { 0x0F71, 0x0F7E }, { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 },
266259698Sdim    { 0x0F8D, 0x0F97 }, { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 },
267259698Sdim    { 0x102D, 0x1030 }, { 0x1032, 0x1037 }, { 0x1039, 0x103A },
268259698Sdim    { 0x103D, 0x103E }, { 0x1058, 0x1059 }, { 0x105E, 0x1060 },
269259698Sdim    { 0x1071, 0x1074 }, { 0x1082, 0x1082 }, { 0x1085, 0x1086 },
270259698Sdim    { 0x108D, 0x108D }, { 0x109D, 0x109D }, { 0x135D, 0x135F },
271259698Sdim    { 0x1712, 0x1714 }, { 0x1732, 0x1734 }, { 0x1752, 0x1753 },
272259698Sdim    { 0x1772, 0x1773 }, { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD },
273259698Sdim    { 0x17C6, 0x17C6 }, { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD },
274259698Sdim    { 0x180B, 0x180D }, { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 },
275259698Sdim    { 0x1927, 0x1928 }, { 0x1932, 0x1932 }, { 0x1939, 0x193B },
276259698Sdim    { 0x1A17, 0x1A18 }, { 0x1A56, 0x1A56 }, { 0x1A58, 0x1A5E },
277259698Sdim    { 0x1A60, 0x1A60 }, { 0x1A62, 0x1A62 }, { 0x1A65, 0x1A6C },
278259698Sdim    { 0x1A73, 0x1A7C }, { 0x1A7F, 0x1A7F }, { 0x1B00, 0x1B03 },
279259698Sdim    { 0x1B34, 0x1B34 }, { 0x1B36, 0x1B3A }, { 0x1B3C, 0x1B3C },
280259698Sdim    { 0x1B42, 0x1B42 }, { 0x1B6B, 0x1B73 }, { 0x1B80, 0x1B81 },
281259698Sdim    { 0x1BA2, 0x1BA5 }, { 0x1BA8, 0x1BA9 }, { 0x1BAB, 0x1BAB },
282259698Sdim    { 0x1BE6, 0x1BE6 }, { 0x1BE8, 0x1BE9 }, { 0x1BED, 0x1BED },
283259698Sdim    { 0x1BEF, 0x1BF1 }, { 0x1C2C, 0x1C33 }, { 0x1C36, 0x1C37 },
284259698Sdim    { 0x1CD0, 0x1CD2 }, { 0x1CD4, 0x1CE0 }, { 0x1CE2, 0x1CE8 },
285259698Sdim    { 0x1CED, 0x1CED }, { 0x1CF4, 0x1CF4 }, { 0x1DC0, 0x1DE6 },
286259698Sdim    { 0x1DFC, 0x1DFF }, { 0x20D0, 0x20F0 }, { 0x2CEF, 0x2CF1 },
287259698Sdim    { 0x2D7F, 0x2D7F }, { 0x2DE0, 0x2DFF }, { 0x302A, 0x302D },
288259698Sdim    { 0x3099, 0x309A }, { 0xA66F, 0xA672 }, { 0xA674, 0xA67D },
289259698Sdim    { 0xA69F, 0xA69F }, { 0xA6F0, 0xA6F1 }, { 0xA802, 0xA802 },
290259698Sdim    { 0xA806, 0xA806 }, { 0xA80B, 0xA80B }, { 0xA825, 0xA826 },
291259698Sdim    { 0xA8C4, 0xA8C4 }, { 0xA8E0, 0xA8F1 }, { 0xA926, 0xA92D },
292259698Sdim    { 0xA947, 0xA951 }, { 0xA980, 0xA982 }, { 0xA9B3, 0xA9B3 },
293259698Sdim    { 0xA9B6, 0xA9B9 }, { 0xA9BC, 0xA9BC }, { 0xAA29, 0xAA2E },
294259698Sdim    { 0xAA31, 0xAA32 }, { 0xAA35, 0xAA36 }, { 0xAA43, 0xAA43 },
295259698Sdim    { 0xAA4C, 0xAA4C }, { 0xAAB0, 0xAAB0 }, { 0xAAB2, 0xAAB4 },
296259698Sdim    { 0xAAB7, 0xAAB8 }, { 0xAABE, 0xAABF }, { 0xAAC1, 0xAAC1 },
297259698Sdim    { 0xAAEC, 0xAAED }, { 0xAAF6, 0xAAF6 }, { 0xABE5, 0xABE5 },
298259698Sdim    { 0xABE8, 0xABE8 }, { 0xABED, 0xABED }, { 0xFB1E, 0xFB1E },
299259698Sdim    { 0xFE00, 0xFE0F }, { 0xFE20, 0xFE26 }, { 0x101FD, 0x101FD },
300259698Sdim    { 0x10A01, 0x10A03 }, { 0x10A05, 0x10A06 }, { 0x10A0C, 0x10A0F },
301259698Sdim    { 0x10A38, 0x10A3A }, { 0x10A3F, 0x10A3F }, { 0x11001, 0x11001 },
302259698Sdim    { 0x11038, 0x11046 }, { 0x11080, 0x11081 }, { 0x110B3, 0x110B6 },
303259698Sdim    { 0x110B9, 0x110BA }, { 0x11100, 0x11102 }, { 0x11127, 0x1112B },
304259698Sdim    { 0x1112D, 0x11134 }, { 0x11180, 0x11181 }, { 0x111B6, 0x111BE },
305259698Sdim    { 0x116AB, 0x116AB }, { 0x116AD, 0x116AD }, { 0x116B0, 0x116B5 },
306259698Sdim    { 0x116B7, 0x116B7 }, { 0x16F8F, 0x16F92 }, { 0x1D167, 0x1D169 },
307259698Sdim    { 0x1D17B, 0x1D182 }, { 0x1D185, 0x1D18B }, { 0x1D1AA, 0x1D1AD },
308259698Sdim    { 0x1D242, 0x1D244 }, { 0xE0100, 0xE01EF },
309259698Sdim  };
310259698Sdim  static const UnicodeCharSet CombiningCharacters(CombiningCharacterRanges);
311259698Sdim
312259698Sdim  if (CombiningCharacters.contains(UCS))
313259698Sdim    return 0;
314259698Sdim
315259698Sdim  static const UnicodeCharRange DoubleWidthCharacterRanges[] = {
316259698Sdim    // Hangul Jamo
317259698Sdim    { 0x1100, 0x11FF },
318259698Sdim    // Deprecated fullwidth angle brackets
319259698Sdim    { 0x2329, 0x232A },
320259698Sdim    // CJK Misc, CJK Unified Ideographs, Yijing Hexagrams, Yi
321259698Sdim    // excluding U+303F (IDEOGRAPHIC HALF FILL SPACE)
322259698Sdim    { 0x2E80, 0x303E }, { 0x3040, 0xA4CF },
323259698Sdim    // Hangul
324259698Sdim    { 0xAC00, 0xD7A3 }, { 0xD7B0, 0xD7C6 }, { 0xD7CB, 0xD7FB },
325259698Sdim    // CJK Unified Ideographs
326259698Sdim    { 0xF900, 0xFAFF },
327259698Sdim    // Vertical forms
328259698Sdim    { 0xFE10, 0xFE19 },
329259698Sdim    // CJK Compatibility Forms + Small Form Variants
330259698Sdim    { 0xFE30, 0xFE6F },
331259698Sdim    // Fullwidth forms
332259698Sdim    { 0xFF01, 0xFF60 }, { 0xFFE0, 0xFFE6 },
333259698Sdim    // CJK Unified Ideographs
334259698Sdim    { 0x20000, 0x2A6DF }, { 0x2A700, 0x2B81F }, { 0x2F800, 0x2FA1F }
335259698Sdim  };
336259698Sdim  static const UnicodeCharSet DoubleWidthCharacters(DoubleWidthCharacterRanges);
337259698Sdim
338259698Sdim  if (DoubleWidthCharacters.contains(UCS))
339259698Sdim    return 2;
340259698Sdim  return 1;
341259698Sdim}
342259698Sdim
343259698Sdimint columnWidthUTF8(StringRef Text) {
344259698Sdim  unsigned ColumnWidth = 0;
345259698Sdim  unsigned Length;
346259698Sdim  for (size_t i = 0, e = Text.size(); i < e; i += Length) {
347259698Sdim    Length = getNumBytesForUTF8(Text[i]);
348259698Sdim    if (Length <= 0 || i + Length > Text.size())
349259698Sdim      return ErrorInvalidUTF8;
350259698Sdim    UTF32 buf[1];
351259698Sdim    const UTF8 *Start = reinterpret_cast<const UTF8 *>(Text.data() + i);
352259698Sdim    UTF32 *Target = &buf[0];
353259698Sdim    if (conversionOK != ConvertUTF8toUTF32(&Start, Start + Length, &Target,
354259698Sdim                                           Target + 1, strictConversion))
355259698Sdim      return ErrorInvalidUTF8;
356259698Sdim    int Width = charWidth(buf[0]);
357259698Sdim    if (Width < 0)
358259698Sdim      return ErrorNonPrintableCharacter;
359259698Sdim    ColumnWidth += Width;
360259698Sdim  }
361259698Sdim  return ColumnWidth;
362259698Sdim}
363259698Sdim
364259698Sdim} // namespace unicode
365259698Sdim} // namespace sys
366259698Sdim} // namespace llvm
367259698Sdim
368