pub const UNICODE_BASIC_LATIN: u16 = 1;
pub const UNICODE_LATIN_1_SUPPLEMENT: u16 = 2;
pub const UNICODE_LATIN_EXTENDED_A: u16 = 3;
pub const UNICODE_LATIN_EXTENDED_B: u16 = 4;
pub const UNICODE_IPA_EXTENSIONS: u16 = 5;
pub const UNICODE_SPACING_MODIFIER_LETTERS: u16 = 6;
pub const UNICODE_COMBINING_DIACRITICAL_MARKS: u16 = 7;
pub const UNICODE_GREEK_AND_COPTIC: u16 = 8;
pub const UNICODE_CYRILLIC: u16 = 9;
pub const UNICODE_CYRILLIC_SUPPLEMENT: u16 = 10;
pub const UNICODE_ARMENIAN: u16 = 11;
pub const UNICODE_HEBREW: u16 = 12;
pub const UNICODE_ARABIC: u16 = 13;
pub const UNICODE_SYRIAC: u16 = 14;
pub const UNICODE_ARABIC_SUPPLEMENT: u16 = 15;
pub const UNICODE_THAANA: u16 = 16;
pub const UNICODE_NKO: u16 = 17;
pub const UNICODE_SAMARITAN: u16 = 18;
pub const UNICODE_MANDAIC: u16 = 19;
pub const UNICODE_ARABIC_EXTENDED_A: u16 = 20;
pub const UNICODE_DEVANAGARI: u16 = 21;
pub const UNICODE_BENGALI: u16 = 22;
pub const UNICODE_GURMUKHI: u16 = 23;
pub const UNICODE_GUJARATI: u16 = 24;
pub const UNICODE_ORIYA: u16 = 25;
pub const UNICODE_TAMIL: u16 = 26;
pub const UNICODE_TELUGU: u16 = 27;
pub const UNICODE_KANNADA: u16 = 28;
pub const UNICODE_MALAYALAM: u16 = 29;
pub const UNICODE_SINHALA: u16 = 30;
pub const UNICODE_THAI: u16 = 31;
pub const UNICODE_LAO: u16 = 32;
pub const UNICODE_TIBETAN: u16 = 33;
pub const UNICODE_MYANMAR: u16 = 34;
pub const UNICODE_GEORGIAN: u16 = 35;
pub const UNICODE_HANGUL_JAMO: u16 = 36;
pub const UNICODE_ETHIOPIC: u16 = 37;
pub const UNICODE_ETHIOPIC_SUPPLEMENT: u16 = 38;
pub const UNICODE_CHEROKEE: u16 = 39;
pub const UNICODE_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS: u16 = 40;
pub const UNICODE_OGHAM: u16 = 41;
pub const UNICODE_RUNIC: u16 = 42;
pub const UNICODE_TAGALOG: u16 = 43;
pub const UNICODE_HANUNOO: u16 = 44;
pub const UNICODE_BUHID: u16 = 45;
pub const UNICODE_TAGBANWA: u16 = 46;
pub const UNICODE_KHMER: u16 = 47;
pub const UNICODE_MONGOLIAN: u16 = 48;
pub const UNICODE_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED: u16 = 49;
pub const UNICODE_LIMBU: u16 = 50;
pub const UNICODE_TAI_LE: u16 = 51;
pub const UNICODE_NEW_TAI_LUE: u16 = 52;
pub const UNICODE_KHMER_SYMBOLS: u16 = 53;
pub const UNICODE_BUGINESE: u16 = 54;
pub const UNICODE_TAI_THAM: u16 = 55;
pub const UNICODE_BALINESE: u16 = 56;
pub const UNICODE_SUNDANESE: u16 = 57;
pub const UNICODE_BATAK: u16 = 58;
pub const UNICODE_LEPCHA: u16 = 59;
pub const UNICODE_OL_CHIKI: u16 = 60;
pub const UNICODE_SUNDANESE_SUPPLEMENT: u16 = 61;
pub const UNICODE_VEDIC_EXTENSIONS: u16 = 62;
pub const UNICODE_PHONETIC_EXTENSIONS: u16 = 63;
pub const UNICODE_PHONETIC_EXTENSIONS_SUPPLEMENT: u16 = 64;
pub const UNICODE_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT: u16 = 65;
pub const UNICODE_LATIN_EXTENDED_ADDITIONAL: u16 = 66;
pub const UNICODE_GREEK_EXTENDED: u16 = 67;
pub const UNICODE_GENERAL_PUNCTUATION: u16 = 68;
pub const UNICODE_SUPERSCRIPTS_AND_SUBSCRIPTS: u16 = 69;
pub const UNICODE_CURRENCY_SYMBOLS: u16 = 70;
pub const UNICODE_COMBINING_DIACRITICAL_MARKS_FOR_SYMBOLS: u16 = 71;
pub const UNICODE_LETTERLIKE_SYMBOLS: u16 = 72;
pub const UNICODE_NUMBER_FORMS: u16 = 73;
pub const UNICODE_ARROWS: u16 = 74;
pub const UNICODE_MATHEMATICAL_OPERATORS: u16 = 75;
pub const UNICODE_MISCELLANEOUS_TECHNICAL: u16 = 76;
pub const UNICODE_CONTROL_PICTURES: u16 = 77;
pub const UNICODE_OPTICAL_CHARACTER_RECOGNITION: u16 = 78;
pub const UNICODE_ENCLOSED_ALPHANUMERICS: u16 = 79;
pub const UNICODE_BOX_DRAWING: u16 = 80;
pub const UNICODE_BLOCK_ELEMENTS: u16 = 81;
pub const UNICODE_GEOMETRIC_SHAPES: u16 = 82;
pub const UNICODE_MISCELLANEOUS_SYMBOLS: u16 = 83;
pub const UNICODE_DINGBATS: u16 = 84;
pub const UNICODE_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A: u16 = 85;
pub const UNICODE_SUPPLEMENTAL_ARROWS_A: u16 = 86;
pub const UNICODE_BRAILLE_PATTERNS: u16 = 87;
pub const UNICODE_SUPPLEMENTAL_ARROWS_B: u16 = 88;
pub const UNICODE_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B: u16 = 89;
pub const UNICODE_SUPPLEMENTAL_MATHEMATICAL_OPERATORS: u16 = 90;
pub const UNICODE_MISCELLANEOUS_SYMBOLS_AND_ARROWS: u16 = 91;
pub const UNICODE_GLAGOLITIC: u16 = 92;
pub const UNICODE_LATIN_EXTENDED_C: u16 = 93;
pub const UNICODE_COPTIC: u16 = 94;
pub const UNICODE_GEORGIAN_SUPPLEMENT: u16 = 95;
pub const UNICODE_TIFINAGH: u16 = 96;
pub const UNICODE_ETHIOPIC_EXTENDED: u16 = 97;
pub const UNICODE_CYRILLIC_EXTENDED_A: u16 = 98;
pub const UNICODE_SUPPLEMENTAL_PUNCTUATION: u16 = 99;
pub const UNICODE_CJK_RADICALS_SUPPLEMENT: u16 = 100;
pub const UNICODE_KANGXI_RADICALS: u16 = 101;
pub const UNICODE_IDEOGRAPHIC_DESCRIPTION_CHARACTERS: u16 = 102;
pub const UNICODE_CJK_SYMBOLS_AND_PUNCTUATION: u16 = 103;
pub const UNICODE_HIRAGANA: u16 = 104;
pub const UNICODE_KATAKANA: u16 = 105;
pub const UNICODE_BOPOMOFO: u16 = 106;
pub const UNICODE_HANGUL_COMPATIBILITY_JAMO: u16 = 107;
pub const UNICODE_KANBUN: u16 = 108;
pub const UNICODE_BOPOMOFO_EXTENDED: u16 = 109;
pub const UNICODE_CJK_STROKES: u16 = 110;
pub const UNICODE_KATAKANA_PHONETIC_EXTENSIONS: u16 = 111;
pub const UNICODE_ENCLOSED_CJK_LETTERS_AND_MONTHS: u16 = 112;
pub const UNICODE_CJK_COMPATIBILITY: u16 = 113;
pub const UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A: u16 = 114;
pub const UNICODE_YIJING_HEXAGRAM_SYMBOLS: u16 = 115;
pub const UNICODE_CJK_UNIFIED_IDEOGRAPHS: u16 = 116;
pub const UNICODE_YI_SYLLABLES: u16 = 117;
pub const UNICODE_YI_RADICALS: u16 = 118;
pub const UNICODE_LISU: u16 = 119;
pub const UNICODE_VAI: u16 = 120;
pub const UNICODE_CYRILLIC_EXTENDED_B: u16 = 121;
pub const UNICODE_BAMUM: u16 = 122;
pub const UNICODE_MODIFIER_TONE_LETTERS: u16 = 123;
pub const UNICODE_LATIN_EXTENDED_D: u16 = 124;
pub const UNICODE_SYLOTI_NAGRI: u16 = 125;
pub const UNICODE_COMMON_INDIC_NUMBER_FORMS: u16 = 126;
pub const UNICODE_PHAGS_PA: u16 = 127;
pub const UNICODE_SAURASHTRA: u16 = 128;
pub const UNICODE_DEVANAGARI_EXTENDED: u16 = 129;
pub const UNICODE_KAYAH_LI: u16 = 130;
pub const UNICODE_REJANG: u16 = 131;
pub const UNICODE_HANGUL_JAMO_EXTENDED_A: u16 = 132;
pub const UNICODE_JAVANESE: u16 = 133;
pub const UNICODE_CHAM: u16 = 134;
pub const UNICODE_MYANMAR_EXTENDED_A: u16 = 135;
pub const UNICODE_TAI_VIET: u16 = 136;
pub const UNICODE_MEETEI_MAYEK_EXTENSIONS: u16 = 137;
pub const UNICODE_ETHIOPIC_EXTENDED_A: u16 = 138;
pub const UNICODE_MEETEI_MAYEK: u16 = 139;
pub const UNICODE_HANGUL_SYLLABLES: u16 = 140;
pub const UNICODE_HANGUL_JAMO_EXTENDED_B: u16 = 141;
pub const UNICODE_HIGH_SURROGATES: u16 = 142;
pub const UNICODE_HIGH_PRIVATE_USE_SURROGATES: u16 = 143;
pub const UNICODE_LOW_SURROGATES: u16 = 144;
pub const UNICODE_PRIVATE_USE_AREA: u16 = 145;
pub const UNICODE_CJK_COMPATIBILITY_IDEOGRAPHS: u16 = 146;
pub const UNICODE_ALPHABETIC_PRESENTATION_FORMS: u16 = 147;
pub const UNICODE_ARABIC_PRESENTATION_FORMS_A: u16 = 148;
pub const UNICODE_VARIATION_SELECTORS: u16 = 149;
pub const UNICODE_VERTICAL_FORMS: u16 = 150;
pub const UNICODE_COMBINING_HALF_MARKS: u16 = 151;
pub const UNICODE_CJK_COMPATIBILITY_FORMS: u16 = 152;
pub const UNICODE_SMALL_FORM_VARIANTS: u16 = 153;
pub const UNICODE_ARABIC_PRESENTATION_FORMS_B: u16 = 154;
pub const UNICODE_HALFWIDTH_AND_FULLWIDTH_FORMS: u16 = 155;
pub const UNICODE_SPECIALS: u16 = 156;
pub const UNICODE_LINEAR_B_SYLLABARY: u16 = 157;
pub const UNICODE_LINEAR_B_IDEOGRAMS: u16 = 158;
pub const UNICODE_AEGEAN_NUMBERS: u16 = 159;
pub const UNICODE_ANCIENT_GREEK_NUMBERS: u16 = 160;
pub const UNICODE_ANCIENT_SYMBOLS: u16 = 161;
pub const UNICODE_PHAISTOS_DISC: u16 = 162;
pub const UNICODE_LYCIAN: u16 = 163;
pub const UNICODE_CARIAN: u16 = 164;
pub const UNICODE_OLD_ITALIC: u16 = 165;
pub const UNICODE_GOTHIC: u16 = 166;
pub const UNICODE_UGARITIC: u16 = 167;
pub const UNICODE_OLD_PERSIAN: u16 = 168;
pub const UNICODE_DESERET: u16 = 169;
pub const UNICODE_SHAVIAN: u16 = 170;
pub const UNICODE_OSMANYA: u16 = 171;
pub const UNICODE_CYPRIOT_SYLLABARY: u16 = 172;
pub const UNICODE_IMPERIAL_ARAMAIC: u16 = 173;
pub const UNICODE_PHOENICIAN: u16 = 174;
pub const UNICODE_LYDIAN: u16 = 175;
pub const UNICODE_MEROITIC_HIEROGLYPHS: u16 = 176;
pub const UNICODE_MEROITIC_CURSIVE: u16 = 177;
pub const UNICODE_KHAROSHTHI: u16 = 178;
pub const UNICODE_OLD_SOUTH_ARABIAN: u16 = 179;
pub const UNICODE_AVESTAN: u16 = 180;
pub const UNICODE_INSCRIPTIONAL_PARTHIAN: u16 = 181;
pub const UNICODE_INSCRIPTIONAL_PAHLAVI: u16 = 182;
pub const UNICODE_OLD_TURKIC: u16 = 183;
pub const UNICODE_RUMI_NUMERAL_SYMBOLS: u16 = 184;
pub const UNICODE_BRAHMI: u16 = 185;
pub const UNICODE_KAITHI: u16 = 186;
pub const UNICODE_SORA_SOMPENG: u16 = 187;
pub const UNICODE_CHAKMA: u16 = 188;
pub const UNICODE_SHARADA: u16 = 189;
pub const UNICODE_TAKRI: u16 = 190;
pub const UNICODE_CUNEIFORM: u16 = 191;
pub const UNICODE_CUNEIFORM_NUMBERS_AND_PUNCTUATION: u16 = 192;
pub const UNICODE_EGYPTIAN_HIEROGLYPHS: u16 = 193;
pub const UNICODE_BAMUM_SUPPLEMENT: u16 = 194;
pub const UNICODE_MIAO: u16 = 195;
pub const UNICODE_KANA_SUPPLEMENT: u16 = 196;
pub const UNICODE_BYZANTINE_MUSICAL_SYMBOLS: u16 = 197;
pub const UNICODE_MUSICAL_SYMBOLS: u16 = 198;
pub const UNICODE_ANCIENT_GREEK_MUSICAL_NOTATION: u16 = 199;
pub const UNICODE_TAI_XUAN_JING_SYMBOLS: u16 = 200;
pub const UNICODE_COUNTING_ROD_NUMERALS: u16 = 201;
pub const UNICODE_MATHEMATICAL_ALPHANUMERIC_SYMBOLS: u16 = 202;
pub const UNICODE_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS: u16 = 203;
pub const UNICODE_MAHJONG_TILES: u16 = 204;
pub const UNICODE_DOMINO_TILES: u16 = 205;
pub const UNICODE_PLAYING_CARDS: u16 = 206;
pub const UNICODE_ENCLOSED_ALPHANUMERIC_SUPPLEMENT: u16 = 207;
pub const UNICODE_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT: u16 = 208;
pub const UNICODE_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS: u16 = 209;
pub const UNICODE_EMOTICONS: u16 = 210;
pub const UNICODE_TRANSPORT_AND_MAP_SYMBOLS: u16 = 211;
pub const UNICODE_ALCHEMICAL_SYMBOLS: u16 = 212;
pub const UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B: u16 = 213;
pub const UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C: u16 = 214;
pub const UNICODE_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D: u16 = 215;
pub const UNICODE_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT: u16 = 216;
pub const UNICODE_TAGS: u16 = 217;
pub const UNICODE_VARIATION_SELECTORS_SUPPLEMENT: u16 = 218;
pub const UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_A: u16 = 219;
pub const UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_B: u16 = 220;
pub const UNICODE_BLOCKS: &[(u16, u32, u32)] = &[
(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),
];
pub fn unicode_block(ch: char) -> Option<u16> {
let cp = ch as u32;
if cp <= 0x7F {
return Some(UNICODE_BASIC_LATIN);
}
let mut be = 0;
let mut en = UNICODE_BLOCKS.len() as isize - 1;
while be <= en {
let mid = (be + en) >> 1;
let (name, start, end) = UNICODE_BLOCKS[mid as usize];
if start <= cp && cp <= end {
return Some(name);
}
if cp < start {
en = mid - 1;
} else {
be = mid + 1;
}
}
None
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_unicode_block() {
assert_eq!(unicode_block('\u{0065}').unwrap(), UNICODE_BASIC_LATIN);
assert_eq!(unicode_block('\u{007F}').unwrap(), UNICODE_BASIC_LATIN);
assert_eq!(unicode_block('\u{0080}').unwrap(), UNICODE_LATIN_1_SUPPLEMENT);
assert_eq!(unicode_block('\u{21FF}').unwrap(), UNICODE_ARROWS);
assert_eq!(unicode_block('\u{2200}').unwrap(), UNICODE_MATHEMATICAL_OPERATORS);
assert_eq!(unicode_block('\u{2201}').unwrap(), UNICODE_MATHEMATICAL_OPERATORS);
assert_eq!(unicode_block('\u{22FF}').unwrap(), UNICODE_MATHEMATICAL_OPERATORS);
assert_eq!(unicode_block('\u{2300}').unwrap(), UNICODE_MISCELLANEOUS_TECHNICAL);
assert_eq!(unicode_block('\u{F0000}').unwrap(), UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_A);
assert_eq!(unicode_block('\u{FFFFF}').unwrap(), UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_A);
assert_eq!(unicode_block('\u{100000}').unwrap(), UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_B);
assert_eq!(unicode_block('\u{10FFFF}').unwrap(), UNICODE_SUPPLEMENTARY_PRIVATE_USE_AREA_B);
}
}