You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
bazarr/libs/langdetect/utils/unicode_block.py

466 lines
18 KiB

UNICODE_BASIC_LATIN = 1
UNICODE_LATIN_1_SUPPLEMENT = 2
UNICODE_LATIN_EXTENDED_A = 3
UNICODE_LATIN_EXTENDED_B = 4
UNICODE_IPA_EXTENSIONS = 5
UNICODE_SPACING_MODIFIER_LETTERS = 6
UNICODE_COMBINING_DIACRITICAL_MARKS = 7
UNICODE_GREEK_AND_COPTIC = 8
UNICODE_CYRILLIC = 9
UNICODE_CYRILLIC_SUPPLEMENT = 10
UNICODE_ARMENIAN = 11
UNICODE_HEBREW = 12
UNICODE_ARABIC = 13
UNICODE_SYRIAC = 14
UNICODE_ARABIC_SUPPLEMENT = 15
UNICODE_THAANA = 16
UNICODE_NKO = 17
UNICODE_SAMARITAN = 18
UNICODE_MANDAIC = 19
UNICODE_ARABIC_EXTENDED_A = 20
UNICODE_DEVANAGARI = 21
UNICODE_BENGALI = 22
UNICODE_GURMUKHI = 23
UNICODE_GUJARATI = 24
UNICODE_ORIYA = 25
UNICODE_TAMIL = 26
UNICODE_TELUGU = 27
UNICODE_KANNADA = 28
UNICODE_MALAYALAM = 29
UNICODE_SINHALA = 30
UNICODE_THAI = 31
UNICODE_LAO = 32
UNICODE_TIBETAN = 33
UNICODE_MYANMAR = 34
UNICODE_GEORGIAN = 35
UNICODE_HANGUL_JAMO = 36
UNICODE_ETHIOPIC = 37
UNICODE_ETHIOPIC_SUPPLEMENT = 38
UNICODE_CHEROKEE = 39
UNICODE_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS = 40
UNICODE_OGHAM = 41
UNICODE_RUNIC = 42
UNICODE_TAGALOG = 43
UNICODE_HANUNOO = 44
UNICODE_BUHID = 45
UNICODE_TAGBANWA = 46
UNICODE_KHMER = 47
UNICODE_MONGOLIAN = 48
UNICODE_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED = 49
UNICODE_LIMBU = 50
UNICODE_TAI_LE = 51
UNICODE_NEW_TAI_LUE = 52
UNICODE_KHMER_SYMBOLS = 53
UNICODE_BUGINESE = 54
UNICODE_TAI_THAM = 55
UNICODE_BALINESE = 56
UNICODE_SUNDANESE = 57
UNICODE_BATAK = 58
UNICODE_LEPCHA = 59
UNICODE_OL_CHIKI = 60
UNICODE_SUNDANESE_SUPPLEMENT = 61
UNICODE_VEDIC_EXTENSIONS = 62
UNICODE_PHONETIC_EXTENSIONS = 63
UNICODE_PHONETIC_EXTENSIONS_SUPPLEMENT = 64
UNICODE_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT = 65
UNICODE_LATIN_EXTENDED_ADDITIONAL = 66
UNICODE_GREEK_EXTENDED = 67
UNICODE_GENERAL_PUNCTUATION = 68
UNICODE_SUPERSCRIPTS_AND_SUBSCRIPTS = 69
UNICODE_CURRENCY_SYMBOLS = 70
UNICODE_COMBINING_DIACRITICAL_MARKS_FOR_SYMBOLS = 71
UNICODE_LETTERLIKE_SYMBOLS = 72
UNICODE_NUMBER_FORMS = 73
UNICODE_ARROWS = 74
UNICODE_MATHEMATICAL_OPERATORS = 75
UNICODE_MISCELLANEOUS_TECHNICAL = 76
UNICODE_CONTROL_PICTURES = 77
UNICODE_OPTICAL_CHARACTER_RECOGNITION = 78
UNICODE_ENCLOSED_ALPHANUMERICS = 79
UNICODE_BOX_DRAWING = 80
UNICODE_BLOCK_ELEMENTS = 81
UNICODE_GEOMETRIC_SHAPES = 82
UNICODE_MISCELLANEOUS_SYMBOLS = 83
UNICODE_DINGBATS = 84
UNICODE_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A = 85
UNICODE_SUPPLEMENTAL_ARROWS_A = 86
UNICODE_BRAILLE_PATTERNS = 87
UNICODE_SUPPLEMENTAL_ARROWS_B = 88
UNICODE_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B = 89
UNICODE_SUPPLEMENTAL_MATHEMATICAL_OPERATORS = 90
UNICODE_MISCELLANEOUS_SYMBOLS_AND_ARROWS = 91
UNICODE_GLAGOLITIC = 92
UNICODE_LATIN_EXTENDED_C = 93
UNICODE_COPTIC = 94
UNICODE_GEORGIAN_SUPPLEMENT = 95
UNICODE_TIFINAGH = 96
UNICODE_ETHIOPIC_EXTENDED = 97
UNICODE_CYRILLIC_EXTENDED_A = 98
UNICODE_SUPPLEMENTAL_PUNCTUATION = 99
UNICODE_CJK_RADICALS_SUPPLEMENT = 100
UNICODE_KANGXI_RADICALS = 101
UNICODE_IDEOGRAPHIC_DESCRIPTION_CHARACTERS = 102
UNICODE_CJK_SYMBOLS_AND_PUNCTUATION = 103
UNICODE_HIRAGANA = 104
UNICODE_KATAKANA = 105
UNICODE_BOPOMOFO = 106
UNICODE_HANGUL_COMPATIBILITY_JAMO = 107
UNICODE_KANBUN = 108
UNICODE_BOPOMOFO_EXTENDED = 109
UNICODE_CJK_STROKES = 110
UNICODE_KATAKANA_PHONETIC_EXTENSIONS = 111
UNICODE_ENCLOSED_CJK_LETTERS_AND_MONTHS = 112
UNICODE_CJK_COMPATIBILITY = 113
UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A = 114
UNICODE_YIJING_HEXAGRAM_SYMBOLS = 115
UNICODE_CJK_UNIFIED_IDEOGRAPHS = 116
UNICODE_YI_SYLLABLES = 117
UNICODE_YI_RADICALS = 118
UNICODE_LISU = 119
UNICODE_VAI = 120
UNICODE_CYRILLIC_EXTENDED_B = 121
UNICODE_BAMUM = 122
UNICODE_MODIFIER_TONE_LETTERS = 123
UNICODE_LATIN_EXTENDED_D = 124
UNICODE_SYLOTI_NAGRI = 125
UNICODE_COMMON_INDIC_NUMBER_FORMS = 126
UNICODE_PHAGS_PA = 127
UNICODE_SAURASHTRA = 128
UNICODE_DEVANAGARI_EXTENDED = 129
UNICODE_KAYAH_LI = 130
UNICODE_REJANG = 131
UNICODE_HANGUL_JAMO_EXTENDED_A = 132
UNICODE_JAVANESE = 133
UNICODE_CHAM = 134
UNICODE_MYANMAR_EXTENDED_A = 135
UNICODE_TAI_VIET = 136
UNICODE_MEETEI_MAYEK_EXTENSIONS = 137
UNICODE_ETHIOPIC_EXTENDED_A = 138
UNICODE_MEETEI_MAYEK = 139
UNICODE_HANGUL_SYLLABLES = 140
UNICODE_HANGUL_JAMO_EXTENDED_B = 141
UNICODE_HIGH_SURROGATES = 142
UNICODE_HIGH_PRIVATE_USE_SURROGATES = 143
UNICODE_LOW_SURROGATES = 144
UNICODE_PRIVATE_USE_AREA = 145
UNICODE_CJK_COMPATIBILITY_IDEOGRAPHS = 146
UNICODE_ALPHABETIC_PRESENTATION_FORMS = 147
UNICODE_ARABIC_PRESENTATION_FORMS_A = 148
UNICODE_VARIATION_SELECTORS = 149
UNICODE_VERTICAL_FORMS = 150
UNICODE_COMBINING_HALF_MARKS = 151
UNICODE_CJK_COMPATIBILITY_FORMS = 152
UNICODE_SMALL_FORM_VARIANTS = 153
UNICODE_ARABIC_PRESENTATION_FORMS_B = 154
UNICODE_HALFWIDTH_AND_FULLWIDTH_FORMS = 155
UNICODE_SPECIALS = 156
UNICODE_LINEAR_B_SYLLABARY = 157
UNICODE_LINEAR_B_IDEOGRAMS = 158
UNICODE_AEGEAN_NUMBERS = 159
UNICODE_ANCIENT_GREEK_NUMBERS = 160
UNICODE_ANCIENT_SYMBOLS = 161
UNICODE_PHAISTOS_DISC = 162
UNICODE_LYCIAN = 163
UNICODE_CARIAN = 164
UNICODE_OLD_ITALIC = 165
UNICODE_GOTHIC = 166
UNICODE_UGARITIC = 167
UNICODE_OLD_PERSIAN = 168
UNICODE_DESERET = 169
UNICODE_SHAVIAN = 170
UNICODE_OSMANYA = 171
UNICODE_CYPRIOT_SYLLABARY = 172
UNICODE_IMPERIAL_ARAMAIC = 173
UNICODE_PHOENICIAN = 174
UNICODE_LYDIAN = 175
UNICODE_MEROITIC_HIEROGLYPHS = 176
UNICODE_MEROITIC_CURSIVE = 177
UNICODE_KHAROSHTHI = 178
UNICODE_OLD_SOUTH_ARABIAN = 179
UNICODE_AVESTAN = 180
UNICODE_INSCRIPTIONAL_PARTHIAN = 181
UNICODE_INSCRIPTIONAL_PAHLAVI = 182
UNICODE_OLD_TURKIC = 183
UNICODE_RUMI_NUMERAL_SYMBOLS = 184
UNICODE_BRAHMI = 185
UNICODE_KAITHI = 186
UNICODE_SORA_SOMPENG = 187
UNICODE_CHAKMA = 188
UNICODE_SHARADA = 189
UNICODE_TAKRI = 190
UNICODE_CUNEIFORM = 191
UNICODE_CUNEIFORM_NUMBERS_AND_PUNCTUATION = 192
UNICODE_EGYPTIAN_HIEROGLYPHS = 193
UNICODE_BAMUM_SUPPLEMENT = 194
UNICODE_MIAO = 195
UNICODE_KANA_SUPPLEMENT = 196
UNICODE_BYZANTINE_MUSICAL_SYMBOLS = 197
UNICODE_MUSICAL_SYMBOLS = 198
UNICODE_ANCIENT_GREEK_MUSICAL_NOTATION = 199
UNICODE_TAI_XUAN_JING_SYMBOLS = 200
UNICODE_COUNTING_ROD_NUMERALS = 201
UNICODE_MATHEMATICAL_ALPHANUMERIC_SYMBOLS = 202
UNICODE_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS = 203
UNICODE_MAHJONG_TILES = 204
UNICODE_DOMINO_TILES = 205
UNICODE_PLAYING_CARDS = 206
UNICODE_ENCLOSED_ALPHANUMERIC_SUPPLEMENT = 207
UNICODE_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT = 208
UNICODE_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS = 209
UNICODE_EMOTICONS = 210
UNICODE_TRANSPORT_AND_MAP_SYMBOLS = 211
UNICODE_ALCHEMICAL_SYMBOLS = 212
UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B = 213
UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C = 214
UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D = 215
UNICODE_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT = 216
UNICODE_TAGS = 217
UNICODE_VARIATION_SELECTORS_SUPPLEMENT = 218
UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_A = 219
UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_B = 220
_unicode_blocks = [
(UNICODE_BASIC_LATIN, 0x0000, 0x007F),
(UNICODE_LATIN_1_SUPPLEMENT, 0x0080, 0x00FF),
(UNICODE_LATIN_EXTENDED_A, 0x0100, 0x017F),
(UNICODE_LATIN_EXTENDED_B, 0x0180, 0x024F),
(UNICODE_IPA_EXTENSIONS, 0x0250, 0x02AF),
(UNICODE_SPACING_MODIFIER_LETTERS, 0x02B0, 0x02FF),
(UNICODE_COMBINING_DIACRITICAL_MARKS, 0x0300, 0x036F),
(UNICODE_GREEK_AND_COPTIC, 0x0370, 0x03FF),
(UNICODE_CYRILLIC, 0x0400, 0x04FF),
(UNICODE_CYRILLIC_SUPPLEMENT, 0x0500, 0x052F),
(UNICODE_ARMENIAN, 0x0530, 0x058F),
(UNICODE_HEBREW, 0x0590, 0x05FF),
(UNICODE_ARABIC, 0x0600, 0x06FF),
(UNICODE_SYRIAC, 0x0700, 0x074F),
(UNICODE_ARABIC_SUPPLEMENT, 0x0750, 0x077F),
(UNICODE_THAANA, 0x0780, 0x07BF),
(UNICODE_NKO, 0x07C0, 0x07FF),
(UNICODE_SAMARITAN, 0x0800, 0x083F),
(UNICODE_MANDAIC, 0x0840, 0x085F),
(UNICODE_ARABIC_EXTENDED_A, 0x08A0, 0x08FF),
(UNICODE_DEVANAGARI, 0x0900, 0x097F),
(UNICODE_BENGALI, 0x0980, 0x09FF),
(UNICODE_GURMUKHI, 0x0A00, 0x0A7F),
(UNICODE_GUJARATI, 0x0A80, 0x0AFF),
(UNICODE_ORIYA, 0x0B00, 0x0B7F),
(UNICODE_TAMIL, 0x0B80, 0x0BFF),
(UNICODE_TELUGU, 0x0C00, 0x0C7F),
(UNICODE_KANNADA, 0x0C80, 0x0CFF),
(UNICODE_MALAYALAM, 0x0D00, 0x0D7F),
(UNICODE_SINHALA, 0x0D80, 0x0DFF),
(UNICODE_THAI, 0x0E00, 0x0E7F),
(UNICODE_LAO, 0x0E80, 0x0EFF),
(UNICODE_TIBETAN, 0x0F00, 0x0FFF),
(UNICODE_MYANMAR, 0x1000, 0x109F),
(UNICODE_GEORGIAN, 0x10A0, 0x10FF),
(UNICODE_HANGUL_JAMO, 0x1100, 0x11FF),
(UNICODE_ETHIOPIC, 0x1200, 0x137F),
(UNICODE_ETHIOPIC_SUPPLEMENT, 0x1380, 0x139F),
(UNICODE_CHEROKEE, 0x13A0, 0x13FF),
(UNICODE_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS, 0x1400, 0x167F),
(UNICODE_OGHAM, 0x1680, 0x169F),
(UNICODE_RUNIC, 0x16A0, 0x16FF),
(UNICODE_TAGALOG, 0x1700, 0x171F),
(UNICODE_HANUNOO, 0x1720, 0x173F),
(UNICODE_BUHID, 0x1740, 0x175F),
(UNICODE_TAGBANWA, 0x1760, 0x177F),
(UNICODE_KHMER, 0x1780, 0x17FF),
(UNICODE_MONGOLIAN, 0x1800, 0x18AF),
(UNICODE_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED, 0x18B0, 0x18FF),
(UNICODE_LIMBU, 0x1900, 0x194F),
(UNICODE_TAI_LE, 0x1950, 0x197F),
(UNICODE_NEW_TAI_LUE, 0x1980, 0x19DF),
(UNICODE_KHMER_SYMBOLS, 0x19E0, 0x19FF),
(UNICODE_BUGINESE, 0x1A00, 0x1A1F),
(UNICODE_TAI_THAM, 0x1A20, 0x1AAF),
(UNICODE_BALINESE, 0x1B00, 0x1B7F),
(UNICODE_SUNDANESE, 0x1B80, 0x1BBF),
(UNICODE_BATAK, 0x1BC0, 0x1BFF),
(UNICODE_LEPCHA, 0x1C00, 0x1C4F),
(UNICODE_OL_CHIKI, 0x1C50, 0x1C7F),
(UNICODE_SUNDANESE_SUPPLEMENT, 0x1CC0, 0x1CCF),
(UNICODE_VEDIC_EXTENSIONS, 0x1CD0, 0x1CFF),
(UNICODE_PHONETIC_EXTENSIONS, 0x1D00, 0x1D7F),
(UNICODE_PHONETIC_EXTENSIONS_SUPPLEMENT, 0x1D80, 0x1DBF),
(UNICODE_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT, 0x1DC0, 0x1DFF),
(UNICODE_LATIN_EXTENDED_ADDITIONAL, 0x1E00, 0x1EFF),
(UNICODE_GREEK_EXTENDED, 0x1F00, 0x1FFF),
(UNICODE_GENERAL_PUNCTUATION, 0x2000, 0x206F),
(UNICODE_SUPERSCRIPTS_AND_SUBSCRIPTS, 0x2070, 0x209F),
(UNICODE_CURRENCY_SYMBOLS, 0x20A0, 0x20CF),
(UNICODE_COMBINING_DIACRITICAL_MARKS_FOR_SYMBOLS, 0x20D0, 0x20FF),
(UNICODE_LETTERLIKE_SYMBOLS, 0x2100, 0x214F),
(UNICODE_NUMBER_FORMS, 0x2150, 0x218F),
(UNICODE_ARROWS, 0x2190, 0x21FF),
(UNICODE_MATHEMATICAL_OPERATORS, 0x2200, 0x22FF),
(UNICODE_MISCELLANEOUS_TECHNICAL, 0x2300, 0x23FF),
(UNICODE_CONTROL_PICTURES, 0x2400, 0x243F),
(UNICODE_OPTICAL_CHARACTER_RECOGNITION, 0x2440, 0x245F),
(UNICODE_ENCLOSED_ALPHANUMERICS, 0x2460, 0x24FF),
(UNICODE_BOX_DRAWING, 0x2500, 0x257F),
(UNICODE_BLOCK_ELEMENTS, 0x2580, 0x259F),
(UNICODE_GEOMETRIC_SHAPES, 0x25A0, 0x25FF),
(UNICODE_MISCELLANEOUS_SYMBOLS, 0x2600, 0x26FF),
(UNICODE_DINGBATS, 0x2700, 0x27BF),
(UNICODE_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A, 0x27C0, 0x27EF),
(UNICODE_SUPPLEMENTAL_ARROWS_A, 0x27F0, 0x27FF),
(UNICODE_BRAILLE_PATTERNS, 0x2800, 0x28FF),
(UNICODE_SUPPLEMENTAL_ARROWS_B, 0x2900, 0x297F),
(UNICODE_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B, 0x2980, 0x29FF),
(UNICODE_SUPPLEMENTAL_MATHEMATICAL_OPERATORS, 0x2A00, 0x2AFF),
(UNICODE_MISCELLANEOUS_SYMBOLS_AND_ARROWS, 0x2B00, 0x2BFF),
(UNICODE_GLAGOLITIC, 0x2C00, 0x2C5F),
(UNICODE_LATIN_EXTENDED_C, 0x2C60, 0x2C7F),
(UNICODE_COPTIC, 0x2C80, 0x2CFF),
(UNICODE_GEORGIAN_SUPPLEMENT, 0x2D00, 0x2D2F),
(UNICODE_TIFINAGH, 0x2D30, 0x2D7F),
(UNICODE_ETHIOPIC_EXTENDED, 0x2D80, 0x2DDF),
(UNICODE_CYRILLIC_EXTENDED_A, 0x2DE0, 0x2DFF),
(UNICODE_SUPPLEMENTAL_PUNCTUATION, 0x2E00, 0x2E7F),
(UNICODE_CJK_RADICALS_SUPPLEMENT, 0x2E80, 0x2EFF),
(UNICODE_KANGXI_RADICALS, 0x2F00, 0x2FDF),
(UNICODE_IDEOGRAPHIC_DESCRIPTION_CHARACTERS, 0x2FF0, 0x2FFF),
(UNICODE_CJK_SYMBOLS_AND_PUNCTUATION, 0x3000, 0x303F),
(UNICODE_HIRAGANA, 0x3040, 0x309F),
(UNICODE_KATAKANA, 0x30A0, 0x30FF),
(UNICODE_BOPOMOFO, 0x3100, 0x312F),
(UNICODE_HANGUL_COMPATIBILITY_JAMO, 0x3130, 0x318F),
(UNICODE_KANBUN, 0x3190, 0x319F),
(UNICODE_BOPOMOFO_EXTENDED, 0x31A0, 0x31BF),
(UNICODE_CJK_STROKES, 0x31C0, 0x31EF),
(UNICODE_KATAKANA_PHONETIC_EXTENSIONS, 0x31F0, 0x31FF),
(UNICODE_ENCLOSED_CJK_LETTERS_AND_MONTHS, 0x3200, 0x32FF),
(UNICODE_CJK_COMPATIBILITY, 0x3300, 0x33FF),
(UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A, 0x3400, 0x4DBF),
(UNICODE_YIJING_HEXAGRAM_SYMBOLS, 0x4DC0, 0x4DFF),
(UNICODE_CJK_UNIFIED_IDEOGRAPHS, 0x4E00, 0x9FFF),
(UNICODE_YI_SYLLABLES, 0xA000, 0xA48F),
(UNICODE_YI_RADICALS, 0xA490, 0xA4CF),
(UNICODE_LISU, 0xA4D0, 0xA4FF),
(UNICODE_VAI, 0xA500, 0xA63F),
(UNICODE_CYRILLIC_EXTENDED_B, 0xA640, 0xA69F),
(UNICODE_BAMUM, 0xA6A0, 0xA6FF),
(UNICODE_MODIFIER_TONE_LETTERS, 0xA700, 0xA71F),
(UNICODE_LATIN_EXTENDED_D, 0xA720, 0xA7FF),
(UNICODE_SYLOTI_NAGRI, 0xA800, 0xA82F),
(UNICODE_COMMON_INDIC_NUMBER_FORMS, 0xA830, 0xA83F),
(UNICODE_PHAGS_PA, 0xA840, 0xA87F),
(UNICODE_SAURASHTRA, 0xA880, 0xA8DF),
(UNICODE_DEVANAGARI_EXTENDED, 0xA8E0, 0xA8FF),
(UNICODE_KAYAH_LI, 0xA900, 0xA92F),
(UNICODE_REJANG, 0xA930, 0xA95F),
(UNICODE_HANGUL_JAMO_EXTENDED_A, 0xA960, 0xA97F),
(UNICODE_JAVANESE, 0xA980, 0xA9DF),
(UNICODE_CHAM, 0xAA00, 0xAA5F),
(UNICODE_MYANMAR_EXTENDED_A, 0xAA60, 0xAA7F),
(UNICODE_TAI_VIET, 0xAA80, 0xAADF),
(UNICODE_MEETEI_MAYEK_EXTENSIONS, 0xAAE0, 0xAAFF),
(UNICODE_ETHIOPIC_EXTENDED_A, 0xAB00, 0xAB2F),
(UNICODE_MEETEI_MAYEK, 0xABC0, 0xABFF),
(UNICODE_HANGUL_SYLLABLES, 0xAC00, 0xD7AF),
(UNICODE_HANGUL_JAMO_EXTENDED_B, 0xD7B0, 0xD7FF),
(UNICODE_HIGH_SURROGATES, 0xD800, 0xDB7F),
(UNICODE_HIGH_PRIVATE_USE_SURROGATES, 0xDB80, 0xDBFF),
(UNICODE_LOW_SURROGATES, 0xDC00, 0xDFFF),
(UNICODE_PRIVATE_USE_AREA, 0xE000, 0xF8FF),
(UNICODE_CJK_COMPATIBILITY_IDEOGRAPHS, 0xF900, 0xFAFF),
(UNICODE_ALPHABETIC_PRESENTATION_FORMS, 0xFB00, 0xFB4F),
(UNICODE_ARABIC_PRESENTATION_FORMS_A, 0xFB50, 0xFDFF),
(UNICODE_VARIATION_SELECTORS, 0xFE00, 0xFE0F),
(UNICODE_VERTICAL_FORMS, 0xFE10, 0xFE1F),
(UNICODE_COMBINING_HALF_MARKS, 0xFE20, 0xFE2F),
(UNICODE_CJK_COMPATIBILITY_FORMS, 0xFE30, 0xFE4F),
(UNICODE_SMALL_FORM_VARIANTS, 0xFE50, 0xFE6F),
(UNICODE_ARABIC_PRESENTATION_FORMS_B, 0xFE70, 0xFEFF),
(UNICODE_HALFWIDTH_AND_FULLWIDTH_FORMS, 0xFF00, 0xFFEF),
(UNICODE_SPECIALS, 0xFFF0, 0xFFFF),
(UNICODE_LINEAR_B_SYLLABARY, 0x10000, 0x1007F),
(UNICODE_LINEAR_B_IDEOGRAMS, 0x10080, 0x100FF),
(UNICODE_AEGEAN_NUMBERS, 0x10100, 0x1013F),
(UNICODE_ANCIENT_GREEK_NUMBERS, 0x10140, 0x1018F),
(UNICODE_ANCIENT_SYMBOLS, 0x10190, 0x101CF),
(UNICODE_PHAISTOS_DISC, 0x101D0, 0x101FF),
(UNICODE_LYCIAN, 0x10280, 0x1029F),
(UNICODE_CARIAN, 0x102A0, 0x102DF),
(UNICODE_OLD_ITALIC, 0x10300, 0x1032F),
(UNICODE_GOTHIC, 0x10330, 0x1034F),
(UNICODE_UGARITIC, 0x10380, 0x1039F),
(UNICODE_OLD_PERSIAN, 0x103A0, 0x103DF),
(UNICODE_DESERET, 0x10400, 0x1044F),
(UNICODE_SHAVIAN, 0x10450, 0x1047F),
(UNICODE_OSMANYA, 0x10480, 0x104AF),
(UNICODE_CYPRIOT_SYLLABARY, 0x10800, 0x1083F),
(UNICODE_IMPERIAL_ARAMAIC, 0x10840, 0x1085F),
(UNICODE_PHOENICIAN, 0x10900, 0x1091F),
(UNICODE_LYDIAN, 0x10920, 0x1093F),
(UNICODE_MEROITIC_HIEROGLYPHS, 0x10980, 0x1099F),
(UNICODE_MEROITIC_CURSIVE, 0x109A0, 0x109FF),
(UNICODE_KHAROSHTHI, 0x10A00, 0x10A5F),
(UNICODE_OLD_SOUTH_ARABIAN, 0x10A60, 0x10A7F),
(UNICODE_AVESTAN, 0x10B00, 0x10B3F),
(UNICODE_INSCRIPTIONAL_PARTHIAN, 0x10B40, 0x10B5F),
(UNICODE_INSCRIPTIONAL_PAHLAVI, 0x10B60, 0x10B7F),
(UNICODE_OLD_TURKIC, 0x10C00, 0x10C4F),
(UNICODE_RUMI_NUMERAL_SYMBOLS, 0x10E60, 0x10E7F),
(UNICODE_BRAHMI, 0x11000, 0x1107F),
(UNICODE_KAITHI, 0x11080, 0x110CF),
(UNICODE_SORA_SOMPENG, 0x110D0, 0x110FF),
(UNICODE_CHAKMA, 0x11100, 0x1114F),
(UNICODE_SHARADA, 0x11180, 0x111DF),
(UNICODE_TAKRI, 0x11680, 0x116CF),
(UNICODE_CUNEIFORM, 0x12000, 0x123FF),
(UNICODE_CUNEIFORM_NUMBERS_AND_PUNCTUATION, 0x12400, 0x1247F),
(UNICODE_EGYPTIAN_HIEROGLYPHS, 0x13000, 0x1342F),
(UNICODE_BAMUM_SUPPLEMENT, 0x16800, 0x16A3F),
(UNICODE_MIAO, 0x16F00, 0x16F9F),
(UNICODE_KANA_SUPPLEMENT, 0x1B000, 0x1B0FF),
(UNICODE_BYZANTINE_MUSICAL_SYMBOLS, 0x1D000, 0x1D0FF),
(UNICODE_MUSICAL_SYMBOLS, 0x1D100, 0x1D1FF),
(UNICODE_ANCIENT_GREEK_MUSICAL_NOTATION, 0x1D200, 0x1D24F),
(UNICODE_TAI_XUAN_JING_SYMBOLS, 0x1D300, 0x1D35F),
(UNICODE_COUNTING_ROD_NUMERALS, 0x1D360, 0x1D37F),
(UNICODE_MATHEMATICAL_ALPHANUMERIC_SYMBOLS, 0x1D400, 0x1D7FF),
(UNICODE_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS, 0x1EE00, 0x1EEFF),
(UNICODE_MAHJONG_TILES, 0x1F000, 0x1F02F),
(UNICODE_DOMINO_TILES, 0x1F030, 0x1F09F),
(UNICODE_PLAYING_CARDS, 0x1F0A0, 0x1F0FF),
(UNICODE_ENCLOSED_ALPHANUMERIC_SUPPLEMENT, 0x1F100, 0x1F1FF),
(UNICODE_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT, 0x1F200, 0x1F2FF),
(UNICODE_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS, 0x1F300, 0x1F5FF),
(UNICODE_EMOTICONS, 0x1F600, 0x1F64F),
(UNICODE_TRANSPORT_AND_MAP_SYMBOLS, 0x1F680, 0x1F6FF),
(UNICODE_ALCHEMICAL_SYMBOLS, 0x1F700, 0x1F77F),
(UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B, 0x20000, 0x2A6DF),
(UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C, 0x2A700, 0x2B73F),
(UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D, 0x2B740, 0x2B81F),
(UNICODE_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT, 0x2F800, 0x2FA1F),
(UNICODE_TAGS, 0xE0000, 0xE007F),
(UNICODE_VARIATION_SELECTORS_SUPPLEMENT, 0xE0100, 0xE01EF),
(UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_A, 0xF0000, 0xFFFFF),
(UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_B, 0x100000, 0x10FFFF),
]
NUM_BLOCKS = len(_unicode_blocks)
def unicode_block(ch):
'''Return the Unicode block name for ch, or None if ch has no block.'''
cp = ord(ch)
# special case basic latin
if cp <= 0x7F:
return UNICODE_BASIC_LATIN
# binary search for the correct block
be, en = 0, NUM_BLOCKS - 1
while be <= en:
mid = (be+en) >> 1
name, start, end = _unicode_blocks[mid]
if start <= cp <= end:
return name
if cp < start:
en = mid-1
else:
be = mid+1