1 /* unicode glyph intervals to include */
2 static int chars
[][2] = {
3 {0x0000, 0x007f}, /* Basic Latin */
4 {0x0080, 0x00ff}, /* Latin-1 Supplement */
5 {0x0100, 0x017f}, /* Latin Extended-A */
6 {0x0180, 0x024f}, /* Latin Extended-B */
7 {0x0250, 0x02af}, /* IPA Extensions */
8 {0x02b0, 0x02ff}, /* Spacing Modifier Letters */
9 {0x0300, 0x036f}, /* Combining Diacritical Marks */
10 {0x0370, 0x03ff}, /* Greek and Coptic */
11 {0x0400, 0x04ff}, /* Cyrillic */
12 {0x0500, 0x052f}, /* Cyrillic Supplement */
13 {0x0530, 0x058f}, /* Armenian */
14 {0x0590, 0x05ff}, /* Hebrew */
15 {0x0600, 0x06ff}, /* Arabic */
16 {0x0700, 0x074f}, /* Syriac */
17 {0x0750, 0x077f}, /* Arabic Supplement */
18 {0x0780, 0x07bf}, /* Thaana */
19 {0x07c0, 0x07ff}, /* NKo */
20 {0x0800, 0x083f}, /* Samaritan */
21 {0x0900, 0x097f}, /* Devanagari */
22 {0x0980, 0x09ff}, /* Bengali */
23 {0x0a00, 0x0a7f}, /* Gurmukhi */
24 {0x0a80, 0x0aff}, /* Gujarati */
25 {0x0b00, 0x0b7f}, /* Oriya */
26 {0x0b80, 0x0bff}, /* Tamil */
27 {0x0c00, 0x0c7f}, /* Telugu */
28 {0x0c80, 0x0cff}, /* Kannada */
29 {0x0d00, 0x0d7f}, /* Malayalam */
30 {0x0d80, 0x0dff}, /* Sinhala */
31 {0x0e00, 0x0e7f}, /* Thai */
32 {0x0e80, 0x0eff}, /* Lao */
33 {0x0f00, 0x0fff}, /* Tibetan */
34 {0x1000, 0x109f}, /* Myanmar */
35 {0x10a0, 0x10ff}, /* Georgian */
36 {0x1100, 0x11ff}, /* Hangul Jamo */
37 {0x1200, 0x137f}, /* Ethiopic */
38 {0x1380, 0x139f}, /* Ethiopic Supplement */
39 {0x13a0, 0x13ff}, /* Cherokee */
40 {0x1400, 0x167f}, /* Unified Canadian Aboriginal Syllabics */
41 {0x1680, 0x169f}, /* Ogham */
42 {0x16a0, 0x16ff}, /* Runic */
43 {0x1700, 0x171f}, /* Tagalog */
44 {0x1720, 0x173f}, /* Hanunoo */
45 {0x1740, 0x175f}, /* Buhid */
46 {0x1760, 0x177f}, /* Tagbanwa */
47 {0x1780, 0x17ff}, /* Khmer */
48 {0x1800, 0x18af}, /* Mongolian */
49 {0x18b0, 0x18ff}, /* Unified Canadian Aboriginal Syllabics Extended */
50 {0x1900, 0x194f}, /* Limbu */
51 {0x1950, 0x197f}, /* Tai Le */
52 {0x1980, 0x19df}, /* New Tai Lue */
53 {0x19e0, 0x19ff}, /* Khmer Symbols */
54 {0x1a00, 0x1a1f}, /* Buginese */
55 {0x1a20, 0x1aaf}, /* Tai Tham */
56 {0x1b00, 0x1b7f}, /* Balinese */
57 {0x1b80, 0x1bbf}, /* Sundanese */
58 {0x1c00, 0x1c4f}, /* Lepcha */
59 {0x1c50, 0x1c7f}, /* Ol Chiki */
60 {0x1cd0, 0x1cff}, /* Vedic Extensions */
61 {0x1d00, 0x1d7f}, /* Phonetic Extensions */
62 {0x1d80, 0x1dbf}, /* Phonetic Extensions Supplement */
63 {0x1dc0, 0x1dff}, /* Combining Diacritical Marks Supplement */
64 {0x1e00, 0x1eff}, /* Latin Extended Additional */
65 {0x1f00, 0x1fff}, /* Greek Extended */
66 {0x2000, 0x206f}, /* General Punctuation */
67 {0x2070, 0x209f}, /* Superscripts and Subscripts */
68 {0x20a0, 0x20cf}, /* Currency Symbols */
69 {0x20d0, 0x20ff}, /* Combining Diacritical Marks for Symbols */
70 {0x2100, 0x214f}, /* Letterlike Symbols */
71 {0x2150, 0x218f}, /* Number Forms */
72 {0x2190, 0x21ff}, /* Arrows */
73 {0x2200, 0x22ff}, /* Mathematical Operators */
74 {0x2300, 0x23ff}, /* Miscellaneous Technical */
75 {0x2400, 0x243f}, /* Control Pictures */
76 {0x2440, 0x245f}, /* Optical Character Recognition */
77 {0x2460, 0x24ff}, /* Enclosed Alphanumerics */
78 {0x2500, 0x257f}, /* Box Drawing */
79 {0x2580, 0x259f}, /* Block Elements */
80 {0x25a0, 0x25ff}, /* Geometric Shapes */
81 {0x2600, 0x26ff}, /* Miscellaneous Symbols */
82 {0x2700, 0x27bf}, /* Dingbats */
83 {0x27c0, 0x27ef}, /* Miscellaneous Mathematical Symbols-A */
84 {0x27f0, 0x27ff}, /* Supplemental Arrows-A */
85 {0x2800, 0x28ff}, /* Braille Patterns */
86 {0x2900, 0x297f}, /* Supplemental Arrows-B */
87 {0x2980, 0x29ff}, /* Miscellaneous Mathematical Symbols-B */
88 {0x2a00, 0x2aff}, /* Supplemental Mathematical Operators */
89 {0x2b00, 0x2bff}, /* Miscellaneous Symbols and Arrows */
90 {0x2c00, 0x2c5f}, /* Glagolitic */
91 {0x2c60, 0x2c7f}, /* Latin Extended-C */
92 {0x2c80, 0x2cff}, /* Coptic */
93 {0x2d00, 0x2d2f}, /* Georgian Supplement */
94 {0x2d30, 0x2d7f}, /* Tifinagh */
95 {0x2d80, 0x2ddf}, /* Ethiopic Extended */
96 {0x2de0, 0x2dff}, /* Cyrillic Extended-A */
97 {0x2e00, 0x2e7f}, /* Supplemental Punctuation */
98 {0x2e80, 0x2eff}, /* CJK Radicals Supplement */
99 {0x2f00, 0x2fdf}, /* Kangxi Radicals */
100 {0x2ff0, 0x2fff}, /* Ideographic Description Characters */
101 {0x3000, 0x303f}, /* CJK Symbols and Punctuation */
102 {0x3040, 0x309f}, /* Hiragana */
103 {0x30a0, 0x30ff}, /* Katakana */
104 {0x3100, 0x312f}, /* Bopomofo */
105 {0x3130, 0x318f}, /* Hangul Compatibility Jamo */
106 {0x3190, 0x319f}, /* Kanbun */
107 {0x31a0, 0x31bf}, /* Bopomofo Extended */
108 {0x31c0, 0x31ef}, /* CJK Strokes */
109 {0x31f0, 0x31ff}, /* Katakana Phonetic Extensions */
110 {0x3200, 0x32ff}, /* Enclosed CJK Letters and Months */
111 {0x3300, 0x33ff}, /* CJK Compatibility */
112 {0x3400, 0x9fff}, /* CJK ... */
113 {0xa000, 0xa48f}, /* Yi Syllables */
114 {0xa490, 0xa4cf}, /* Yi Radicals */
115 {0xa4d0, 0xa4ff}, /* Lisu */
116 {0xa500, 0xa63f}, /* Vai */
117 {0xa640, 0xa69f}, /* Cyrillic Extended-B */
118 {0xa6a0, 0xa6ff}, /* Bamum */
119 {0xa700, 0xa71f}, /* Modifier Tone Letters */
120 {0xa720, 0xa7ff}, /* Latin Extended-D */
121 {0xa800, 0xa82f}, /* Syloti Nagri */
122 {0xa830, 0xa83f}, /* Common Indic Number Forms */
123 {0xa840, 0xa87f}, /* Phags-pa */
124 {0xa880, 0xa8df}, /* Saurashtra */
125 {0xa8e0, 0xa8ff}, /* Devanagari Extended */
126 {0xa900, 0xa92f}, /* Kayah Li */
127 {0xa930, 0xa95f}, /* Rejang */
128 {0xa960, 0xa97f}, /* Hangul Jamo Extended-A */
129 {0xa980, 0xa9df}, /* Javanese */
130 {0xaa00, 0xaa5f}, /* Cham */
131 {0xaa60, 0xaa7f}, /* Myanmar Extended-A */
132 {0xac00, 0xd7ff}, /* Hangul ... */
133 {0xd800, 0xdb7f}, /* High Surrogates */
134 {0xdc00, 0xdfff}, /* Low Surrogates */
135 {0xf900, 0xfaff}, /* CJK Compatibility Ideographs */
136 {0xfb00, 0xfb4f}, /* Alphabetic Presentation Forms */
137 {0xfb50, 0xfdff}, /* Arabic Presentation Forms-A */
138 {0xfe00, 0xfe0f}, /* Variation Selectors */
139 {0xfe10, 0xfe1f}, /* Vertical Forms */
140 {0xfe20, 0xfe2f}, /* Combining Half Marks */
141 {0xfe30, 0xfe4f}, /* CJK Compatibility Forms */
142 {0xfe50, 0xfe6f}, /* Small Form Variants */
143 {0xfe70, 0xfeff}, /* Arabic Presentation Forms-B */
144 {0xff00, 0xffef}, /* Halfwidth and Fullwidth Forms */
145 {0xfff0, 0xffff}, /* Specials */
