Friendica/friendica-github
1
0
Fork 0
mirror of https://github.com/friendica/friendica synced 2024-11-15 08:53:54 +00:00
Code Issues Projects Releases Packages Wiki Activity
friendica-github/library/langdet/data/unicode_blocks.php
Tobias Diekershoff c86840c4b9 updated pear LangDetect to version 1.0.0
2017-09-04 09:52:14 +02:00

874 lines
12 KiB
PHP
Raw Blame History

<?php
return array (
0 =>
array (
0 => 0x0000,
1 => 0x007F,
2 => 'Basic Latin',
),
1 =>
array (
0 => 0x0080,
1 => 0x00FF,
2 => 'Latin-1 Supplement',
),
2 =>
array (
0 => 0x0100,
1 => 0x017F,
2 => 'Latin Extended-A',
),
3 =>
array (
0 => 0x0180,
1 => 0x024F,
2 => 'Latin Extended-B',
),
4 =>
array (
0 => 0x0250,
1 => 0x02AF,
2 => 'IPA Extensions',
),
5 =>
array (
0 => 0x02B0,
1 => 0x02FF,
2 => 'Spacing Modifier Letters',
),
6 =>
array (
0 => 0x0300,
1 => 0x036F,
2 => 'Combining Diacritical Marks',
),
7 =>
array (
0 => 0x0370,
1 => 0x03FF,
2 => 'Greek and Coptic',
),
8 =>
array (
0 => 0x0400,
1 => 0x04FF,
2 => 'Cyrillic',
),
9 =>
array (
0 => 0x0500,
1 => 0x052F,
2 => 'Cyrillic Supplement',
),
10 =>
array (
0 => 0x0530,
1 => 0x058F,
2 => 'Armenian',
),
11 =>
array (
0 => 0x0590,
1 => 0x05FF,
2 => 'Hebrew',
),
12 =>
array (
0 => 0x0600,
1 => 0x06FF,
2 => 'Arabic',
),
13 =>
array (
0 => 0x0700,
1 => 0x074F,
2 => 'Syriac',
),
14 =>
array (
0 => 0x0750,
1 => 0x077F,
2 => 'Arabic Supplement',
),
15 =>
array (
0 => 0x0780,
1 => 0x07BF,
2 => 'Thaana',
),
16 =>
array (
0 => 0x0900,
1 => 0x097F,
2 => 'Devanagari',
),
17 =>
array (
0 => 0x0980,
1 => 0x09FF,
2 => 'Bengali',
),
18 =>
array (
0 => 0x0A00,
1 => 0x0A7F,
2 => 'Gurmukhi',
),
19 =>
array (
0 => 0x0A80,
1 => 0x0AFF,
2 => 'Gujarati',
),
20 =>
array (
0 => 0x0B00,
1 => 0x0B7F,
2 => 'Oriya',
),
21 =>
array (
0 => 0x0B80,
1 => 0x0BFF,
2 => 'Tamil',
),
22 =>
array (
0 => 0x0C00,
1 => 0x0C7F,
2 => 'Telugu',
),
23 =>
array (
0 => 0x0C80,
1 => 0x0CFF,
2 => 'Kannada',
),
24 =>
array (
0 => 0x0D00,
1 => 0x0D7F,
2 => 'Malayalam',
),
25 =>
array (
0 => 0x0D80,
1 => 0x0DFF,
2 => 'Sinhala',
),
26 =>
array (
0 => 0x0E00,
1 => 0x0E7F,
2 => 'Thai',
),
27 =>
array (
0 => 0x0E80,
1 => 0x0EFF,
2 => 'Lao',
),
28 =>
array (
0 => 0x0F00,
1 => 0x0FFF,
2 => 'Tibetan',
),
29 =>
array (
0 => 0x1000,
1 => 0x109F,
2 => 'Myanmar',
),
30 =>
array (
0 => 0x10A0,
1 => 0x10FF,
2 => 'Georgian',
),
31 =>
array (
0 => 0x1100,
1 => 0x11FF,
2 => 'Hangul Jamo',
),
32 =>
array (
0 => 0x1200,
1 => 0x137F,
2 => 'Ethiopic',
),
33 =>
array (
0 => 0x1380,
1 => 0x139F,
2 => 'Ethiopic Supplement',
),
34 =>
array (
0 => 0x13A0,
1 => 0x13FF,
2 => 'Cherokee',
),
35 =>
array (
0 => 0x1400,
1 => 0x167F,
2 => 'Unified Canadian Aboriginal Syllabics',
),
36 =>
array (
0 => 0x1680,
1 => 0x169F,
2 => 'Ogham',
),
37 =>
array (
0 => 0x16A0,
1 => 0x16FF,
2 => 'Runic',
),
38 =>
array (
0 => 0x1700,
1 => 0x171F,
2 => 'Tagalog',
),
39 =>
array (
0 => 0x1720,
1 => 0x173F,
2 => 'Hanunoo',
),
40 =>
array (
0 => 0x1740,
1 => 0x175F,
2 => 'Buhid',
),
41 =>
array (
0 => 0x1760,
1 => 0x177F,
2 => 'Tagbanwa',
),
42 =>
array (
0 => 0x1780,
1 => 0x17FF,
2 => 'Khmer',
),
43 =>
array (
0 => 0x1800,
1 => 0x18AF,
2 => 'Mongolian',
),
44 =>
array (
0 => 0x1900,
1 => 0x194F,
2 => 'Limbu',
),
45 =>
array (
0 => 0x1950,
1 => 0x197F,
2 => 'Tai Le',
),
46 =>
array (
0 => 0x1980,
1 => 0x19DF,
2 => 'New Tai Lue',
),
47 =>
array (
0 => 0x19E0,
1 => 0x19FF,
2 => 'Khmer Symbols',
),
48 =>
array (
0 => 0x1A00,
1 => 0x1A1F,
2 => 'Buginese',
),
49 =>
array (
0 => 0x1D00,
1 => 0x1D7F,
2 => 'Phonetic Extensions',
),
50 =>
array (
0 => 0x1D80,
1 => 0x1DBF,
2 => 'Phonetic Extensions Supplement',
),
51 =>
array (
0 => 0x1DC0,
1 => 0x1DFF,
2 => 'Combining Diacritical Marks Supplement',
),
52 =>
array (
0 => 0x1E00,
1 => 0x1EFF,
2 => 'Latin Extended Additional',
),
53 =>
array (
0 => 0x1F00,
1 => 0x1FFF,
2 => 'Greek Extended',
),
54 =>
array (
0 => 0x2000,
1 => 0x206F,
2 => 'General Punctuation',
),
55 =>
array (
0 => 0x2070,
1 => 0x209F,
2 => 'Superscripts and Subscripts',
),
56 =>
array (
0 => 0x20A0,
1 => 0x20CF,
2 => 'Currency Symbols',
),
57 =>
array (
0 => 0x20D0,
1 => 0x20FF,
2 => 'Combining Diacritical Marks for Symbols',
),
58 =>
array (
0 => 0x2100,
1 => 0x214F,
2 => 'Letterlike Symbols',
),
59 =>
array (
0 => 0x2150,
1 => 0x218F,
2 => 'Number Forms',
),
60 =>
array (
0 => 0x2190,
1 => 0x21FF,
2 => 'Arrows',
),
61 =>
array (
0 => 0x2200,
1 => 0x22FF,
2 => 'Mathematical Operators',
),
62 =>
array (
0 => 0x2300,
1 => 0x23FF,
2 => 'Miscellaneous Technical',
),
63 =>
array (
0 => 0x2400,
1 => 0x243F,
2 => 'Control Pictures',
),
64 =>
array (
0 => 0x2440,
1 => 0x245F,
2 => 'Optical Character Recognition',
),
65 =>
array (
0 => 0x2460,
1 => 0x24FF,
2 => 'Enclosed Alphanumerics',
),
66 =>
array (
0 => 0x2500,
1 => 0x257F,
2 => 'Box Drawing',
),
67 =>
array (
0 => 0x2580,
1 => 0x259F,
2 => 'Block Elements',
),
68 =>
array (
0 => 0x25A0,
1 => 0x25FF,
2 => 'Geometric Shapes',
),
69 =>
array (
0 => 0x2600,
1 => 0x26FF,
2 => 'Miscellaneous Symbols',
),
70 =>
array (
0 => 0x2700,
1 => 0x27BF,
2 => 'Dingbats',
),
71 =>
array (
0 => 0x27C0,
1 => 0x27EF,
2 => 'Miscellaneous Mathematical Symbols-A',
),
72 =>
array (
0 => 0x27F0,
1 => 0x27FF,
2 => 'Supplemental Arrows-A',
),
73 =>
array (
0 => 0x2800,
1 => 0x28FF,
2 => 'Braille Patterns',
),
74 =>
array (
0 => 0x2900,
1 => 0x297F,
2 => 'Supplemental Arrows-B',
),
75 =>
array (
0 => 0x2980,
1 => 0x29FF,
2 => 'Miscellaneous Mathematical Symbols-B',
),
76 =>
array (
0 => 0x2A00,
1 => 0x2AFF,
2 => 'Supplemental Mathematical Operators',
),
77 =>
array (
0 => 0x2B00,
1 => 0x2BFF,
2 => 'Miscellaneous Symbols and Arrows',
),
78 =>
array (
0 => 0x2C00,
1 => 0x2C5F,
2 => 'Glagolitic',
),
79 =>
array (
0 => 0x2C80,
1 => 0x2CFF,
2 => 'Coptic',
),
80 =>
array (
0 => 0x2D00,
1 => 0x2D2F,
2 => 'Georgian Supplement',
),
81 =>
array (
0 => 0x2D30,
1 => 0x2D7F,
2 => 'Tifinagh',
),
82 =>
array (
0 => 0x2D80,
1 => 0x2DDF,
2 => 'Ethiopic Extended',
),
83 =>
array (
0 => 0x2E00,
1 => 0x2E7F,
2 => 'Supplemental Punctuation',
),
84 =>
array (
0 => 0x2E80,
1 => 0x2EFF,
2 => 'CJK Radicals Supplement',
),
85 =>
array (
0 => 0x2F00,
1 => 0x2FDF,
2 => 'Kangxi Radicals',
),
86 =>
array (
0 => 0x2FF0,
1 => 0x2FFF,
2 => 'Ideographic Description Characters',
),
87 =>
array (
0 => 0x3000,
1 => 0x303F,
2 => 'CJK Symbols and Punctuation',
),
88 =>
array (
0 => 0x3040,
1 => 0x309F,
2 => 'Hiragana',
),
89 =>
array (
0 => 0x30A0,
1 => 0x30FF,
2 => 'Katakana',
),
90 =>
array (
0 => 0x3100,
1 => 0x312F,
2 => 'Bopomofo',
),
91 =>
array (
0 => 0x3130,
1 => 0x318F,
2 => 'Hangul Compatibility Jamo',
),
92 =>
array (
0 => 0x3190,
1 => 0x319F,
2 => 'Kanbun',
),
93 =>
array (
0 => 0x31A0,
1 => 0x31BF,
2 => 'Bopomofo Extended',
),
94 =>
array (
0 => 0x31C0,
1 => 0x31EF,
2 => 'CJK Strokes',
),
95 =>
array (
0 => 0x31F0,
1 => 0x31FF,
2 => 'Katakana Phonetic Extensions',
),
96 =>
array (
0 => 0x3200,
1 => 0x32FF,
2 => 'Enclosed CJK Letters and Months',
),
97 =>
array (
0 => 0x3300,
1 => 0x33FF,
2 => 'CJK Compatibility',
),
98 =>
array (
0 => 0x3400,
1 => 0x4DBF,
2 => 'CJK Unified Ideographs Extension A',
),
99 =>
array (
0 => 0x4DC0,
1 => 0x4DFF,
2 => 'Yijing Hexagram Symbols',
),
100 =>
array (
0 => 0x4E00,
1 => 0x9FFF,
2 => 'CJK Unified Ideographs',
),
101 =>
array (
0 => 0xA000,
1 => 0xA48F,
2 => 'Yi Syllables',
),
102 =>
array (
0 => 0xA490,
1 => 0xA4CF,
2 => 'Yi Radicals',
),
103 =>
array (
0 => 0xA700,
1 => 0xA71F,
2 => 'Modifier Tone Letters',
),
104 =>
array (
0 => 0xA800,
1 => 0xA82F,
2 => 'Syloti Nagri',
),
105 =>
array (
0 => 0xAC00,
1 => 0xD7AF,
2 => 'Hangul Syllables',
),
106 =>
array (
0 => 0xD800,
1 => 0xDB7F,
2 => 'High Surrogates',
),
107 =>
array (
0 => 0xDB80,
1 => 0xDBFF,
2 => 'High Private Use Surrogates',
),
108 =>
array (
0 => 0xDC00,
1 => 0xDFFF,
2 => 'Low Surrogates',
),
109 =>
array (
0 => 0xE000,
1 => 0xF8FF,
2 => 'Private Use Area',
),
110 =>
array (
0 => 0xF900,
1 => 0xFAFF,
2 => 'CJK Compatibility Ideographs',
),
111 =>
array (
0 => 0xFB00,
1 => 0xFB4F,
2 => 'Alphabetic Presentation Forms',
),
112 =>
array (
0 => 0xFB50,
1 => 0xFDFF,
2 => 'Arabic Presentation Forms-A',
),
113 =>
array (
0 => 0xFE00,
1 => 0xFE0F,
2 => 'Variation Selectors',
),
114 =>
array (
0 => 0xFE10,
1 => 0xFE1F,
2 => 'Vertical Forms',
),
115 =>
array (
0 => 0xFE20,
1 => 0xFE2F,
2 => 'Combining Half Marks',
),
116 =>
array (
0 => 0xFE30,
1 => 0xFE4F,
2 => 'CJK Compatibility Forms',
),
117 =>
array (
0 => 0xFE50,
1 => 0xFE6F,
2 => 'Small Form Variants',
),
118 =>
array (
0 => 0xFE70,
1 => 0xFEFF,
2 => 'Arabic Presentation Forms-B',
),
119 =>
array (
0 => 0xFF00,
1 => 0xFFEF,
2 => 'Halfwidth and Fullwidth Forms',
),
120 =>
array (
0 => 0xFFF0,
1 => 0xFFFF,
2 => 'Specials',
),
121 =>
array (
0 => 0x10000,
1 => 0x1007F,
2 => 'Linear B Syllabary',
),
122 =>
array (
0 => 0x10080,
1 => 0x100FF,
2 => 'Linear B Ideograms',
),
123 =>
array (
0 => 0x10100,
1 => 0x1013F,
2 => 'Aegean Numbers',
),
124 =>
array (
0 => 0x10140,
1 => 0x1018F,
2 => 'Ancient Greek Numbers',
),
125 =>
array (
0 => 0x10300,
1 => 0x1032F,
2 => 'Old Italic',
),
126 =>
array (
0 => 0x10330,
1 => 0x1034F,
2 => 'Gothic',
),
127 =>
array (
0 => 0x10380,
1 => 0x1039F,
2 => 'Ugaritic',
),
128 =>
array (
0 => 0x103A0,
1 => 0x103DF,
2 => 'Old Persian',
),
129 =>
array (
0 => 0x10400,
1 => 0x1044F,
2 => 'Deseret',
),
130 =>
array (
0 => 0x10450,
1 => 0x1047F,
2 => 'Shavian',
),
131 =>
array (
0 => 0x10480,
1 => 0x104AF,
2 => 'Osmanya',
),
132 =>
array (
0 => 0x10800,
1 => 0x1083F,
2 => 'Cypriot Syllabary',
),
133 =>
array (
0 => 0x10A00,
1 => 0x10A5F,
2 => 'Kharoshthi',
),
134 =>
array (
0 => 0x1D000,
1 => 0x1D0FF,
2 => 'Byzantine Musical Symbols',
),
135 =>
array (
0 => 0x1D100,
1 => 0x1D1FF,
2 => 'Musical Symbols',
),
136 =>
array (
0 => 0x1D200,
1 => 0x1D24F,
2 => 'Ancient Greek Musical Notation',
),
137 =>
array (
0 => 0x1D300,
1 => 0x1D35F,
2 => 'Tai Xuan Jing Symbols',
),
138 =>
array (
0 => 0x1D400,
1 => 0x1D7FF,
2 => 'Mathematical Alphanumeric Symbols',
),
139 =>
array (
0 => 0x20000,
1 => 0x2A6DF,
2 => 'CJK Unified Ideographs Extension B',
),
140 =>
array (
0 => 0x2F800,
1 => 0x2FA1F,
2 => 'CJK Compatibility Ideographs Supplement',
),
141 =>
array (
0 => 0xE0000,
1 => 0xE007F,
2 => 'Tags',
),
142 =>
array (
0 => 0xE0100,
1 => 0xE01EF,
2 => 'Variation Selectors Supplement',
),
143 =>
array (
0 => 0xF0000,
1 => 0xFFFFF,
2 => 'Supplementary Private Use Area-A',
),
144 =>
array (
0 => 0x100000,
1 => 0x10FFFF,
2 => 'Supplementary Private Use Area-B',
),
);
?>
Reference in a new issue View git blame Copy permalink