1
#ifndef UNICODE_BLOCKS_H
2
#define UNICODE_BLOCKS_H
4
typedef struct _UnicodeBlock
11
static const UnicodeBlock unicodeBlocks[] =
13
{0x0000, 0x007F, "Basic Latin"},
14
{0x0080, 0x00FF, "Latin-1 Supplement"},
15
{0x0100, 0x017F, "Latin Extended-A"},
16
{0x0180, 0x024F, "Latin Extended-B"},
17
{0x0250, 0x02AF, "IPA Extensions"},
18
{0x02B0, 0x02FF, "Spacing Modifier Letters"},
19
{0x0300, 0x036F, "Combining Diacritical Marks"},
20
{0x0370, 0x03FF, "Greek and Coptic"},
21
{0x0400, 0x04FF, "Cyrillic"},
22
{0x0500, 0x052F, "Cyrillic Supplement"},
23
{0x0530, 0x058F, "Armenian"},
24
{0x0590, 0x05FF, "Hebrew"},
25
{0x0600, 0x06FF, "Arabic"},
26
{0x0700, 0x074F, "Syriac"},
27
{0x0750, 0x077F, "Arabic Supplement"},
28
{0x0780, 0x07BF, "Thaana"},
29
{0x07C0, 0x07FF, "NKo"},
30
{0x0800, 0x083F, "Samaritan"},
31
{0x0840, 0x085F, "Mandaic"},
32
{0x0900, 0x097F, "Devanagari"},
33
{0x0980, 0x09FF, "Bengali"},
34
{0x0A00, 0x0A7F, "Gurmukhi"},
35
{0x0A80, 0x0AFF, "Gujarati"},
36
{0x0B00, 0x0B7F, "Oriya"},
37
{0x0B80, 0x0BFF, "Tamil"},
38
{0x0C00, 0x0C7F, "Telugu"},
39
{0x0C80, 0x0CFF, "Kannada"},
40
{0x0D00, 0x0D7F, "Malayalam"},
41
{0x0D80, 0x0DFF, "Sinhala"},
42
{0x0E00, 0x0E7F, "Thai"},
43
{0x0E80, 0x0EFF, "Lao"},
44
{0x0F00, 0x0FFF, "Tibetan"},
45
{0x1000, 0x109F, "Myanmar"},
46
{0x10A0, 0x10FF, "Georgian"},
47
{0x1100, 0x11FF, "Hangul Jamo"},
48
{0x1200, 0x137F, "Ethiopic"},
49
{0x1380, 0x139F, "Ethiopic Supplement"},
50
{0x13A0, 0x13FF, "Cherokee"},
51
{0x1400, 0x167F, "Unified Canadian Aboriginal Syllabics"},
52
{0x1680, 0x169F, "Ogham"},
53
{0x16A0, 0x16FF, "Runic"},
54
{0x1700, 0x171F, "Tagalog"},
55
{0x1720, 0x173F, "Hanunoo"},
56
{0x1740, 0x175F, "Buhid"},
57
{0x1760, 0x177F, "Tagbanwa"},
58
{0x1780, 0x17FF, "Khmer"},
59
{0x1800, 0x18AF, "Mongolian"},
60
{0x18B0, 0x18FF, "Unified Canadian Aboriginal Syllabics Extended"},
61
{0x1900, 0x194F, "Limbu"},
62
{0x1950, 0x197F, "Tai Le"},
63
{0x1980, 0x19DF, "New Tai Lue"},
64
{0x19E0, 0x19FF, "Khmer Symbols"},
65
{0x1A00, 0x1A1F, "Buginese"},
66
{0x1A20, 0x1AAF, "Tai Tham"},
67
{0x1B00, 0x1B7F, "Balinese"},
68
{0x1B80, 0x1BBF, "Sundanese"},
69
{0x1BC0, 0x1BFF, "Batak"},
70
{0x1C00, 0x1C4F, "Lepcha"},
71
{0x1C50, 0x1C7F, "Ol Chiki"},
72
{0x1CD0, 0x1CFF, "Vedic Extensions"},
73
{0x1D00, 0x1D7F, "Phonetic Extensions"},
74
{0x1D80, 0x1DBF, "Phonetic Extensions Supplement"},
75
{0x1DC0, 0x1DFF, "Combining Diacritical Marks Supplement"},
76
{0x1E00, 0x1EFF, "Latin Extended Additional"},
77
{0x1F00, 0x1FFF, "Greek Extended"},
78
{0x2000, 0x206F, "General Punctuation"},
79
{0x2070, 0x209F, "Superscripts and Subscripts"},
80
{0x20A0, 0x20CF, "Currency Symbols"},
81
{0x20D0, 0x20FF, "Combining Diacritical Marks for Symbols"},
82
{0x2100, 0x214F, "Letterlike Symbols"},
83
{0x2150, 0x218F, "Number Forms"},
84
{0x2190, 0x21FF, "Arrows"},
85
{0x2200, 0x22FF, "Mathematical Operators"},
86
{0x2300, 0x23FF, "Miscellaneous Technical"},
87
{0x2400, 0x243F, "Control Pictures"},
88
{0x2440, 0x245F, "Optical Character Recognition"},
89
{0x2460, 0x24FF, "Enclosed Alphanumerics"},
90
{0x2500, 0x257F, "Box Drawing"},
91
{0x2580, 0x259F, "Block Elements"},
92
{0x25A0, 0x25FF, "Geometric Shapes"},
93
{0x2600, 0x26FF, "Miscellaneous Symbols"},
94
{0x2700, 0x27BF, "Dingbats"},
95
{0x27C0, 0x27EF, "Miscellaneous Mathematical Symbols-A"},
96
{0x27F0, 0x27FF, "Supplemental Arrows-A"},
97
{0x2800, 0x28FF, "Braille Patterns"},
98
{0x2900, 0x297F, "Supplemental Arrows-B"},
99
{0x2980, 0x29FF, "Miscellaneous Mathematical Symbols-B"},
100
{0x2A00, 0x2AFF, "Supplemental Mathematical Operators"},
101
{0x2B00, 0x2BFF, "Miscellaneous Symbols and Arrows"},
102
{0x2C00, 0x2C5F, "Glagolitic"},
103
{0x2C60, 0x2C7F, "Latin Extended-C"},
104
{0x2C80, 0x2CFF, "Coptic"},
105
{0x2D00, 0x2D2F, "Georgian Supplement"},
106
{0x2D30, 0x2D7F, "Tifinagh"},
107
{0x2D80, 0x2DDF, "Ethiopic Extended"},
108
{0x2DE0, 0x2DFF, "Cyrillic Extended-A"},
109
{0x2E00, 0x2E7F, "Supplemental Punctuation"},
110
{0x2E80, 0x2EFF, "CJK Radicals Supplement"},
111
{0x2F00, 0x2FDF, "Kangxi Radicals"},
112
{0x2FF0, 0x2FFF, "Ideographic Description Characters"},
113
{0x3000, 0x303F, "CJK Symbols and Punctuation"},
114
{0x3040, 0x309F, "Hiragana"},
115
{0x30A0, 0x30FF, "Katakana"},
116
{0x3100, 0x312F, "Bopomofo"},
117
{0x3130, 0x318F, "Hangul Compatibility Jamo"},
118
{0x3190, 0x319F, "Kanbun"},
119
{0x31A0, 0x31BF, "Bopomofo Extended"},
120
{0x31C0, 0x31EF, "CJK Strokes"},
121
{0x31F0, 0x31FF, "Katakana Phonetic Extensions"},
122
{0x3200, 0x32FF, "Enclosed CJK Letters and Months"},
123
{0x3300, 0x33FF, "CJK Compatibility"},
124
{0x3400, 0x4DBF, "CJK Unified Ideographs Extension A"},
125
{0x4DC0, 0x4DFF, "Yijing Hexagram Symbols"},
126
{0x4E00, 0x9FFF, "CJK Unified Ideographs"},
127
{0xA000, 0xA48F, "Yi Syllables"},
128
{0xA490, 0xA4CF, "Yi Radicals"},
129
{0xA4D0, 0xA4FF, "Lisu"},
130
{0xA500, 0xA63F, "Vai"},
131
{0xA640, 0xA69F, "Cyrillic Extended-B"},
132
{0xA6A0, 0xA6FF, "Bamum"},
133
{0xA700, 0xA71F, "Modifier Tone Letters"},
134
{0xA720, 0xA7FF, "Latin Extended-D"},
135
{0xA800, 0xA82F, "Syloti Nagri"},
136
{0xA830, 0xA83F, "Common Indic Number Forms"},
137
{0xA840, 0xA87F, "Phags-pa"},
138
{0xA880, 0xA8DF, "Saurashtra"},
139
{0xA8E0, 0xA8FF, "Devanagari Extended"},
140
{0xA900, 0xA92F, "Kayah Li"},
141
{0xA930, 0xA95F, "Rejang"},
142
{0xA960, 0xA97F, "Hangul Jamo Extended-A"},
143
{0xA980, 0xA9DF, "Javanese"},
144
{0xAA00, 0xAA5F, "Cham"},
145
{0xAA60, 0xAA7F, "Myanmar Extended-A"},
146
{0xAA80, 0xAADF, "Tai Viet"},
147
{0xAB00, 0xAB2F, "Ethiopic Extended-A"},
148
{0xABC0, 0xABFF, "Meetei Mayek"},
149
{0xAC00, 0xD7AF, "Hangul Syllables"},
150
{0xD7B0, 0xD7FF, "Hangul Jamo Extended-B"},
151
{0xD800, 0xDB7F, "High Surrogates"},
152
{0xDB80, 0xDBFF, "High Private Use Surrogates"},
153
{0xDC00, 0xDFFF, "Low Surrogates"},
154
{0xE000, 0xF8FF, "Private Use Area"},
155
{0xF900, 0xFAFF, "CJK Compatibility Ideographs"},
156
{0xFB00, 0xFB4F, "Alphabetic Presentation Forms"},
157
{0xFB50, 0xFDFF, "Arabic Presentation Forms-A"},
158
{0xFE00, 0xFE0F, "Variation Selectors"},
159
{0xFE10, 0xFE1F, "Vertical Forms"},
160
{0xFE20, 0xFE2F, "Combining Half Marks"},
161
{0xFE30, 0xFE4F, "CJK Compatibility Forms"},
162
{0xFE50, 0xFE6F, "Small Form Variants"},
163
{0xFE70, 0xFEFF, "Arabic Presentation Forms-B"},
164
{0xFF00, 0xFFEF, "Halfwidth and Fullwidth Forms"},
165
{0xFFF0, 0xFFFF, "Specials"},
166
{0x10000, 0x1007F, "Linear B Syllabary"},
167
{0x10080, 0x100FF, "Linear B Ideograms"},
168
{0x10100, 0x1013F, "Aegean Numbers"},
169
{0x10140, 0x1018F, "Ancient Greek Numbers"},
170
{0x10190, 0x101CF, "Ancient Symbols"},
171
{0x101D0, 0x101FF, "Phaistos Disc"},
172
{0x10280, 0x1029F, "Lycian"},
173
{0x102A0, 0x102DF, "Carian"},
174
{0x10300, 0x1032F, "Old Italic"},
175
{0x10330, 0x1034F, "Gothic"},
176
{0x10380, 0x1039F, "Ugaritic"},
177
{0x103A0, 0x103DF, "Old Persian"},
178
{0x10400, 0x1044F, "Deseret"},
179
{0x10450, 0x1047F, "Shavian"},
180
{0x10480, 0x104AF, "Osmanya"},
181
{0x10800, 0x1083F, "Cypriot Syllabary"},
182
{0x10840, 0x1085F, "Imperial Aramaic"},
183
{0x10900, 0x1091F, "Phoenician"},
184
{0x10920, 0x1093F, "Lydian"},
185
{0x10A00, 0x10A5F, "Kharoshthi"},
186
{0x10A60, 0x10A7F, "Old South Arabian"},
187
{0x10B00, 0x10B3F, "Avestan"},
188
{0x10B40, 0x10B5F, "Inscriptional Parthian"},
189
{0x10B60, 0x10B7F, "Inscriptional Pahlavi"},
190
{0x10C00, 0x10C4F, "Old Turkic"},
191
{0x10E60, 0x10E7F, "Rumi Numeral Symbols"},
192
{0x11000, 0x1107F, "Brahmi"},
193
{0x11080, 0x110CF, "Kaithi"},
194
{0x12000, 0x123FF, "Cuneiform"},
195
{0x12400, 0x1247F, "Cuneiform Numbers and Punctuation"},
196
{0x13000, 0x1342F, "Egyptian Hieroglyphs"},
197
{0x16800, 0x16A3F, "Bamum Supplement"},
198
{0x1B000, 0x1B0FF, "Kana Supplement"},
199
{0x1D000, 0x1D0FF, "Byzantine Musical Symbols"},
200
{0x1D100, 0x1D1FF, "Musical Symbols"},
201
{0x1D200, 0x1D24F, "Ancient Greek Musical Notation"},
202
{0x1D300, 0x1D35F, "Tai Xuan Jing Symbols"},
203
{0x1D360, 0x1D37F, "Counting Rod Numerals"},
204
{0x1D400, 0x1D7FF, "Mathematical Alphanumeric Symbols"},
205
{0x1F000, 0x1F02F, "Mahjong Tiles"},
206
{0x1F030, 0x1F09F, "Domino Tiles"},
207
{0x1F0A0, 0x1F0FF, "Playing Cards"},
208
{0x1F100, 0x1F1FF, "Enclosed Alphanumeric Supplement"},
209
{0x1F200, 0x1F2FF, "Enclosed Ideographic Supplement"},
210
{0x1F300, 0x1F5FF, "Miscellaneous Symbols And Pictographs"},
211
{0x1F600, 0x1F64F, "Emoticons"},
212
{0x1F680, 0x1F6FF, "Transport And Map Symbols"},
213
{0x1F700, 0x1F77F, "Alchemical Symbols"},
214
{0x20000, 0x2A6DF, "CJK Unified Ideographs Extension B"},
215
{0x2A700, 0x2B73F, "CJK Unified Ideographs Extension C"},
216
{0x2B740, 0x2B81F, "CJK Unified Ideographs Extension D"},
217
{0x2F800, 0x2FA1F, "CJK Compatibility Ideographs Supplement"},
218
{0xE0000, 0xE007F, "Tags"},
219
{0xE0100, 0xE01EF, "Variation Selectors Supplement"},
220
{0xF0000, 0xFFFFF, "Supplementary Private Use Area-A"},
221
{0x100000, 0x10FFFF, "Supplementary Private Use Area-B"},