UNPKG

json-to-ast

Version:
790 lines (752 loc) 140 kB
(function (global, factory) { typeof exports === 'object' && typeof module !== 'undefined' ? module.exports = factory() : typeof define === 'function' && define.amd ? define(factory) : (global.jsonToAst = factory()); }(this, (function () { 'use strict'; var commonjsGlobal = typeof window !== 'undefined' ? window : typeof global !== 'undefined' ? global : typeof self !== 'undefined' ? self : {}; function createCommonjsModule(fn, module) { return module = { exports: {} }, fn(module, module.exports), module.exports; } var graphemeSplitter = createCommonjsModule(function (module) { /* Breaks a Javascript string into individual user-perceived "characters" called extended grapheme clusters by implementing the Unicode UAX-29 standard, version 10.0.0 Usage: var splitter = new GraphemeSplitter(); //returns an array of strings, one string for each grapheme cluster var graphemes = splitter.splitGraphemes(string); */ function GraphemeSplitter() { var CR = 0, LF = 1, Control = 2, Extend = 3, Regional_Indicator = 4, SpacingMark = 5, L = 6, V = 7, T = 8, LV = 9, LVT = 10, Other = 11, Prepend = 12, E_Base = 13, E_Modifier = 14, ZWJ = 15, Glue_After_Zwj = 16, E_Base_GAZ = 17; // BreakTypes var NotBreak = 0, BreakStart = 1, Break = 2, BreakLastRegional = 3, BreakPenultimateRegional = 4; function isSurrogate(str, pos) { return 0xd800 <= str.charCodeAt(pos) && str.charCodeAt(pos) <= 0xdbff && 0xdc00 <= str.charCodeAt(pos + 1) && str.charCodeAt(pos + 1) <= 0xdfff; } // Private function, gets a Unicode code point from a JavaScript UTF-16 string // handling surrogate pairs appropriately function codePointAt(str, idx) { if (idx === undefined) { idx = 0; } var code = str.charCodeAt(idx); // if a high surrogate if (0xD800 <= code && code <= 0xDBFF && idx < str.length - 1) { var hi = code; var low = str.charCodeAt(idx + 1); if (0xDC00 <= low && low <= 0xDFFF) { return (hi - 0xD800) * 0x400 + (low - 0xDC00) + 0x10000; } return hi; } // if a low surrogate if (0xDC00 <= code && code <= 0xDFFF && idx >= 1) { var hi = str.charCodeAt(idx - 1); var low = code; if (0xD800 <= hi && hi <= 0xDBFF) { return (hi - 0xD800) * 0x400 + (low - 0xDC00) + 0x10000; } return low; } //just return the char if an unmatched surrogate half or a //single-char codepoint return code; } // Private function, returns whether a break is allowed between the // two given grapheme breaking classes function shouldBreak(start, mid, end) { var all = [start].concat(mid).concat([end]); var previous = all[all.length - 2]; var next = end; // Lookahead termintor for: // GB10. (E_Base | EBG) Extend* ? E_Modifier var eModifierIndex = all.lastIndexOf(E_Modifier); if (eModifierIndex > 1 && all.slice(1, eModifierIndex).every(function (c) { return c == Extend; }) && [Extend, E_Base, E_Base_GAZ].indexOf(start) == -1) { return Break; } // Lookahead termintor for: // GB12. ^ (RI RI)* RI ? RI // GB13. [^RI] (RI RI)* RI ? RI var rIIndex = all.lastIndexOf(Regional_Indicator); if (rIIndex > 0 && all.slice(1, rIIndex).every(function (c) { return c == Regional_Indicator; }) && [Prepend, Regional_Indicator].indexOf(previous) == -1) { if (all.filter(function (c) { return c == Regional_Indicator; }).length % 2 == 1) { return BreakLastRegional; } else { return BreakPenultimateRegional; } } // GB3. CR X LF if (previous == CR && next == LF) { return NotBreak; } // GB4. (Control|CR|LF) ÷ else if (previous == Control || previous == CR || previous == LF) { if (next == E_Modifier && mid.every(function (c) { return c == Extend; })) { return Break; } else { return BreakStart; } } // GB5. ÷ (Control|CR|LF) else if (next == Control || next == CR || next == LF) { return BreakStart; } // GB6. L X (L|V|LV|LVT) else if (previous == L && (next == L || next == V || next == LV || next == LVT)) { return NotBreak; } // GB7. (LV|V) X (V|T) else if ((previous == LV || previous == V) && (next == V || next == T)) { return NotBreak; } // GB8. (LVT|T) X (T) else if ((previous == LVT || previous == T) && next == T) { return NotBreak; } // GB9. X (Extend|ZWJ) else if (next == Extend || next == ZWJ) { return NotBreak; } // GB9a. X SpacingMark else if (next == SpacingMark) { return NotBreak; } // GB9b. Prepend X else if (previous == Prepend) { return NotBreak; } // GB10. (E_Base | EBG) Extend* ? E_Modifier var previousNonExtendIndex = all.indexOf(Extend) != -1 ? all.lastIndexOf(Extend) - 1 : all.length - 2; if ([E_Base, E_Base_GAZ].indexOf(all[previousNonExtendIndex]) != -1 && all.slice(previousNonExtendIndex + 1, -1).every(function (c) { return c == Extend; }) && next == E_Modifier) { return NotBreak; } // GB11. ZWJ ? (Glue_After_Zwj | EBG) if (previous == ZWJ && [Glue_After_Zwj, E_Base_GAZ].indexOf(next) != -1) { return NotBreak; } // GB12. ^ (RI RI)* RI ? RI // GB13. [^RI] (RI RI)* RI ? RI if (mid.indexOf(Regional_Indicator) != -1) { return Break; } if (previous == Regional_Indicator && next == Regional_Indicator) { return NotBreak; } // GB999. Any ? Any return BreakStart; } // Returns the next grapheme break in the string after the given index this.nextBreak = function (string, index) { if (index === undefined) { index = 0; } if (index < 0) { return 0; } if (index >= string.length - 1) { return string.length; } var prev = getGraphemeBreakProperty(codePointAt(string, index)); var mid = []; for (var i = index + 1; i < string.length; i++) { // check for already processed low surrogates if (isSurrogate(string, i - 1)) { continue; } var next = getGraphemeBreakProperty(codePointAt(string, i)); if (shouldBreak(prev, mid, next)) { return i; } mid.push(next); } return string.length; }; // Breaks the given string into an array of grapheme cluster strings this.splitGraphemes = function (str) { var res = []; var index = 0; var brk; while ((brk = this.nextBreak(str, index)) < str.length) { res.push(str.slice(index, brk)); index = brk; } if (index < str.length) { res.push(str.slice(index)); } return res; }; // Returns the iterator of grapheme clusters there are in the given string this.iterateGraphemes = function (str) { var index = 0; var res = { next: function () { var value; var brk; if ((brk = this.nextBreak(str, index)) < str.length) { value = str.slice(index, brk); index = brk; return { value: value, done: false }; } if (index < str.length) { value = str.slice(index); index = str.length; return { value: value, done: false }; } return { value: undefined, done: true }; }.bind(this) }; // ES2015 @@iterator method (iterable) for spread syntax and for...of statement if (typeof Symbol !== 'undefined' && Symbol.iterator) { res[Symbol.iterator] = function () { return res; }; } return res; }; // Returns the number of grapheme clusters there are in the given string this.countGraphemes = function (str) { var count = 0; var index = 0; var brk; while ((brk = this.nextBreak(str, index)) < str.length) { index = brk; count++; } if (index < str.length) { count++; } return count; }; //given a Unicode code point, determines this symbol's grapheme break property function getGraphemeBreakProperty(code) { //grapheme break property for Unicode 10.0.0, //taken from http://www.unicode.org/Public/10.0.0/ucd/auxiliary/GraphemeBreakProperty.txt //and adapted to JavaScript rules if (0x0600 <= code && code <= 0x0605 || // Cf [6] ARABIC NUMBER SIGN..ARABIC NUMBER MARK ABOVE 0x06DD == code || // Cf ARABIC END OF AYAH 0x070F == code || // Cf SYRIAC ABBREVIATION MARK 0x08E2 == code || // Cf ARABIC DISPUTED END OF AYAH 0x0D4E == code || // Lo MALAYALAM LETTER DOT REPH 0x110BD == code || // Cf KAITHI NUMBER SIGN 0x111C2 <= code && code <= 0x111C3 || // Lo [2] SHARADA SIGN JIHVAMULIYA..SHARADA SIGN UPADHMANIYA 0x11A3A == code || // Lo ZANABAZAR SQUARE CLUSTER-INITIAL LETTER RA 0x11A86 <= code && code <= 0x11A89 || // Lo [4] SOYOMBO CLUSTER-INITIAL LETTER RA..SOYOMBO CLUSTER-INITIAL LETTER SA 0x11D46 == code // Lo MASARAM GONDI REPHA ) { return Prepend; } if (0x000D == code // Cc <control-000D> ) { return CR; } if (0x000A == code // Cc <control-000A> ) { return LF; } if (0x0000 <= code && code <= 0x0009 || // Cc [10] <control-0000>..<control-0009> 0x000B <= code && code <= 0x000C || // Cc [2] <control-000B>..<control-000C> 0x000E <= code && code <= 0x001F || // Cc [18] <control-000E>..<control-001F> 0x007F <= code && code <= 0x009F || // Cc [33] <control-007F>..<control-009F> 0x00AD == code || // Cf SOFT HYPHEN 0x061C == code || // Cf ARABIC LETTER MARK 0x180E == code || // Cf MONGOLIAN VOWEL SEPARATOR 0x200B == code || // Cf ZERO WIDTH SPACE 0x200E <= code && code <= 0x200F || // Cf [2] LEFT-TO-RIGHT MARK..RIGHT-TO-LEFT MARK 0x2028 == code || // Zl LINE SEPARATOR 0x2029 == code || // Zp PARAGRAPH SEPARATOR 0x202A <= code && code <= 0x202E || // Cf [5] LEFT-TO-RIGHT EMBEDDING..RIGHT-TO-LEFT OVERRIDE 0x2060 <= code && code <= 0x2064 || // Cf [5] WORD JOINER..INVISIBLE PLUS 0x2065 == code || // Cn <reserved-2065> 0x2066 <= code && code <= 0x206F || // Cf [10] LEFT-TO-RIGHT ISOLATE..NOMINAL DIGIT SHAPES 0xD800 <= code && code <= 0xDFFF || // Cs [2048] <surrogate-D800>..<surrogate-DFFF> 0xFEFF == code || // Cf ZERO WIDTH NO-BREAK SPACE 0xFFF0 <= code && code <= 0xFFF8 || // Cn [9] <reserved-FFF0>..<reserved-FFF8> 0xFFF9 <= code && code <= 0xFFFB || // Cf [3] INTERLINEAR ANNOTATION ANCHOR..INTERLINEAR ANNOTATION TERMINATOR 0x1BCA0 <= code && code <= 0x1BCA3 || // Cf [4] SHORTHAND FORMAT LETTER OVERLAP..SHORTHAND FORMAT UP STEP 0x1D173 <= code && code <= 0x1D17A || // Cf [8] MUSICAL SYMBOL BEGIN BEAM..MUSICAL SYMBOL END PHRASE 0xE0000 == code || // Cn <reserved-E0000> 0xE0001 == code || // Cf LANGUAGE TAG 0xE0002 <= code && code <= 0xE001F || // Cn [30] <reserved-E0002>..<reserved-E001F> 0xE0080 <= code && code <= 0xE00FF || // Cn [128] <reserved-E0080>..<reserved-E00FF> 0xE01F0 <= code && code <= 0xE0FFF // Cn [3600] <reserved-E01F0>..<reserved-E0FFF> ) { return Control; } if (0x0300 <= code && code <= 0x036F || // Mn [112] COMBINING GRAVE ACCENT..COMBINING LATIN SMALL LETTER X 0x0483 <= code && code <= 0x0487 || // Mn [5] COMBINING CYRILLIC TITLO..COMBINING CYRILLIC POKRYTIE 0x0488 <= code && code <= 0x0489 || // Me [2] COMBINING CYRILLIC HUNDRED THOUSANDS SIGN..COMBINING CYRILLIC MILLIONS SIGN 0x0591 <= code && code <= 0x05BD || // Mn [45] HEBREW ACCENT ETNAHTA..HEBREW POINT METEG 0x05BF == code || // Mn HEBREW POINT RAFE 0x05C1 <= code && code <= 0x05C2 || // Mn [2] HEBREW POINT SHIN DOT..HEBREW POINT SIN DOT 0x05C4 <= code && code <= 0x05C5 || // Mn [2] HEBREW MARK UPPER DOT..HEBREW MARK LOWER DOT 0x05C7 == code || // Mn HEBREW POINT QAMATS QATAN 0x0610 <= code && code <= 0x061A || // Mn [11] ARABIC SIGN SALLALLAHOU ALAYHE WASSALLAM..ARABIC SMALL KASRA 0x064B <= code && code <= 0x065F || // Mn [21] ARABIC FATHATAN..ARABIC WAVY HAMZA BELOW 0x0670 == code || // Mn ARABIC LETTER SUPERSCRIPT ALEF 0x06D6 <= code && code <= 0x06DC || // Mn [7] ARABIC SMALL HIGH LIGATURE SAD WITH LAM WITH ALEF MAKSURA..ARABIC SMALL HIGH SEEN 0x06DF <= code && code <= 0x06E4 || // Mn [6] ARABIC SMALL HIGH ROUNDED ZERO..ARABIC SMALL HIGH MADDA 0x06E7 <= code && code <= 0x06E8 || // Mn [2] ARABIC SMALL HIGH YEH..ARABIC SMALL HIGH NOON 0x06EA <= code && code <= 0x06ED || // Mn [4] ARABIC EMPTY CENTRE LOW STOP..ARABIC SMALL LOW MEEM 0x0711 == code || // Mn SYRIAC LETTER SUPERSCRIPT ALAPH 0x0730 <= code && code <= 0x074A || // Mn [27] SYRIAC PTHAHA ABOVE..SYRIAC BARREKH 0x07A6 <= code && code <= 0x07B0 || // Mn [11] THAANA ABAFILI..THAANA SUKUN 0x07EB <= code && code <= 0x07F3 || // Mn [9] NKO COMBINING SHORT HIGH TONE..NKO COMBINING DOUBLE DOT ABOVE 0x0816 <= code && code <= 0x0819 || // Mn [4] SAMARITAN MARK IN..SAMARITAN MARK DAGESH 0x081B <= code && code <= 0x0823 || // Mn [9] SAMARITAN MARK EPENTHETIC YUT..SAMARITAN VOWEL SIGN A 0x0825 <= code && code <= 0x0827 || // Mn [3] SAMARITAN VOWEL SIGN SHORT A..SAMARITAN VOWEL SIGN U 0x0829 <= code && code <= 0x082D || // Mn [5] SAMARITAN VOWEL SIGN LONG I..SAMARITAN MARK NEQUDAA 0x0859 <= code && code <= 0x085B || // Mn [3] MANDAIC AFFRICATION MARK..MANDAIC GEMINATION MARK 0x08D4 <= code && code <= 0x08E1 || // Mn [14] ARABIC SMALL HIGH WORD AR-RUB..ARABIC SMALL HIGH SIGN SAFHA 0x08E3 <= code && code <= 0x0902 || // Mn [32] ARABIC TURNED DAMMA BELOW..DEVANAGARI SIGN ANUSVARA 0x093A == code || // Mn DEVANAGARI VOWEL SIGN OE 0x093C == code || // Mn DEVANAGARI SIGN NUKTA 0x0941 <= code && code <= 0x0948 || // Mn [8] DEVANAGARI VOWEL SIGN U..DEVANAGARI VOWEL SIGN AI 0x094D == code || // Mn DEVANAGARI SIGN VIRAMA 0x0951 <= code && code <= 0x0957 || // Mn [7] DEVANAGARI STRESS SIGN UDATTA..DEVANAGARI VOWEL SIGN UUE 0x0962 <= code && code <= 0x0963 || // Mn [2] DEVANAGARI VOWEL SIGN VOCALIC L..DEVANAGARI VOWEL SIGN VOCALIC LL 0x0981 == code || // Mn BENGALI SIGN CANDRABINDU 0x09BC == code || // Mn BENGALI SIGN NUKTA 0x09BE == code || // Mc BENGALI VOWEL SIGN AA 0x09C1 <= code && code <= 0x09C4 || // Mn [4] BENGALI VOWEL SIGN U..BENGALI VOWEL SIGN VOCALIC RR 0x09CD == code || // Mn BENGALI SIGN VIRAMA 0x09D7 == code || // Mc BENGALI AU LENGTH MARK 0x09E2 <= code && code <= 0x09E3 || // Mn [2] BENGALI VOWEL SIGN VOCALIC L..BENGALI VOWEL SIGN VOCALIC LL 0x0A01 <= code && code <= 0x0A02 || // Mn [2] GURMUKHI SIGN ADAK BINDI..GURMUKHI SIGN BINDI 0x0A3C == code || // Mn GURMUKHI SIGN NUKTA 0x0A41 <= code && code <= 0x0A42 || // Mn [2] GURMUKHI VOWEL SIGN U..GURMUKHI VOWEL SIGN UU 0x0A47 <= code && code <= 0x0A48 || // Mn [2] GURMUKHI VOWEL SIGN EE..GURMUKHI VOWEL SIGN AI 0x0A4B <= code && code <= 0x0A4D || // Mn [3] GURMUKHI VOWEL SIGN OO..GURMUKHI SIGN VIRAMA 0x0A51 == code || // Mn GURMUKHI SIGN UDAAT 0x0A70 <= code && code <= 0x0A71 || // Mn [2] GURMUKHI TIPPI..GURMUKHI ADDAK 0x0A75 == code || // Mn GURMUKHI SIGN YAKASH 0x0A81 <= code && code <= 0x0A82 || // Mn [2] GUJARATI SIGN CANDRABINDU..GUJARATI SIGN ANUSVARA 0x0ABC == code || // Mn GUJARATI SIGN NUKTA 0x0AC1 <= code && code <= 0x0AC5 || // Mn [5] GUJARATI VOWEL SIGN U..GUJARATI VOWEL SIGN CANDRA E 0x0AC7 <= code && code <= 0x0AC8 || // Mn [2] GUJARATI VOWEL SIGN E..GUJARATI VOWEL SIGN AI 0x0ACD == code || // Mn GUJARATI SIGN VIRAMA 0x0AE2 <= code && code <= 0x0AE3 || // Mn [2] GUJARATI VOWEL SIGN VOCALIC L..GUJARATI VOWEL SIGN VOCALIC LL 0x0AFA <= code && code <= 0x0AFF || // Mn [6] GUJARATI SIGN SUKUN..GUJARATI SIGN TWO-CIRCLE NUKTA ABOVE 0x0B01 == code || // Mn ORIYA SIGN CANDRABINDU 0x0B3C == code || // Mn ORIYA SIGN NUKTA 0x0B3E == code || // Mc ORIYA VOWEL SIGN AA 0x0B3F == code || // Mn ORIYA VOWEL SIGN I 0x0B41 <= code && code <= 0x0B44 || // Mn [4] ORIYA VOWEL SIGN U..ORIYA VOWEL SIGN VOCALIC RR 0x0B4D == code || // Mn ORIYA SIGN VIRAMA 0x0B56 == code || // Mn ORIYA AI LENGTH MARK 0x0B57 == code || // Mc ORIYA AU LENGTH MARK 0x0B62 <= code && code <= 0x0B63 || // Mn [2] ORIYA VOWEL SIGN VOCALIC L..ORIYA VOWEL SIGN VOCALIC LL 0x0B82 == code || // Mn TAMIL SIGN ANUSVARA 0x0BBE == code || // Mc TAMIL VOWEL SIGN AA 0x0BC0 == code || // Mn TAMIL VOWEL SIGN II 0x0BCD == code || // Mn TAMIL SIGN VIRAMA 0x0BD7 == code || // Mc TAMIL AU LENGTH MARK 0x0C00 == code || // Mn TELUGU SIGN COMBINING CANDRABINDU ABOVE 0x0C3E <= code && code <= 0x0C40 || // Mn [3] TELUGU VOWEL SIGN AA..TELUGU VOWEL SIGN II 0x0C46 <= code && code <= 0x0C48 || // Mn [3] TELUGU VOWEL SIGN E..TELUGU VOWEL SIGN AI 0x0C4A <= code && code <= 0x0C4D || // Mn [4] TELUGU VOWEL SIGN O..TELUGU SIGN VIRAMA 0x0C55 <= code && code <= 0x0C56 || // Mn [2] TELUGU LENGTH MARK..TELUGU AI LENGTH MARK 0x0C62 <= code && code <= 0x0C63 || // Mn [2] TELUGU VOWEL SIGN VOCALIC L..TELUGU VOWEL SIGN VOCALIC LL 0x0C81 == code || // Mn KANNADA SIGN CANDRABINDU 0x0CBC == code || // Mn KANNADA SIGN NUKTA 0x0CBF == code || // Mn KANNADA VOWEL SIGN I 0x0CC2 == code || // Mc KANNADA VOWEL SIGN UU 0x0CC6 == code || // Mn KANNADA VOWEL SIGN E 0x0CCC <= code && code <= 0x0CCD || // Mn [2] KANNADA VOWEL SIGN AU..KANNADA SIGN VIRAMA 0x0CD5 <= code && code <= 0x0CD6 || // Mc [2] KANNADA LENGTH MARK..KANNADA AI LENGTH MARK 0x0CE2 <= code && code <= 0x0CE3 || // Mn [2] KANNADA VOWEL SIGN VOCALIC L..KANNADA VOWEL SIGN VOCALIC LL 0x0D00 <= code && code <= 0x0D01 || // Mn [2] MALAYALAM SIGN COMBINING ANUSVARA ABOVE..MALAYALAM SIGN CANDRABINDU 0x0D3B <= code && code <= 0x0D3C || // Mn [2] MALAYALAM SIGN VERTICAL BAR VIRAMA..MALAYALAM SIGN CIRCULAR VIRAMA 0x0D3E == code || // Mc MALAYALAM VOWEL SIGN AA 0x0D41 <= code && code <= 0x0D44 || // Mn [4] MALAYALAM VOWEL SIGN U..MALAYALAM VOWEL SIGN VOCALIC RR 0x0D4D == code || // Mn MALAYALAM SIGN VIRAMA 0x0D57 == code || // Mc MALAYALAM AU LENGTH MARK 0x0D62 <= code && code <= 0x0D63 || // Mn [2] MALAYALAM VOWEL SIGN VOCALIC L..MALAYALAM VOWEL SIGN VOCALIC LL 0x0DCA == code || // Mn SINHALA SIGN AL-LAKUNA 0x0DCF == code || // Mc SINHALA VOWEL SIGN AELA-PILLA 0x0DD2 <= code && code <= 0x0DD4 || // Mn [3] SINHALA VOWEL SIGN KETTI IS-PILLA..SINHALA VOWEL SIGN KETTI PAA-PILLA 0x0DD6 == code || // Mn SINHALA VOWEL SIGN DIGA PAA-PILLA 0x0DDF == code || // Mc SINHALA VOWEL SIGN GAYANUKITTA 0x0E31 == code || // Mn THAI CHARACTER MAI HAN-AKAT 0x0E34 <= code && code <= 0x0E3A || // Mn [7] THAI CHARACTER SARA I..THAI CHARACTER PHINTHU 0x0E47 <= code && code <= 0x0E4E || // Mn [8] THAI CHARACTER MAITAIKHU..THAI CHARACTER YAMAKKAN 0x0EB1 == code || // Mn LAO VOWEL SIGN MAI KAN 0x0EB4 <= code && code <= 0x0EB9 || // Mn [6] LAO VOWEL SIGN I..LAO VOWEL SIGN UU 0x0EBB <= code && code <= 0x0EBC || // Mn [2] LAO VOWEL SIGN MAI KON..LAO SEMIVOWEL SIGN LO 0x0EC8 <= code && code <= 0x0ECD || // Mn [6] LAO TONE MAI EK..LAO NIGGAHITA 0x0F18 <= code && code <= 0x0F19 || // Mn [2] TIBETAN ASTROLOGICAL SIGN -KHYUD PA..TIBETAN ASTROLOGICAL SIGN SDONG TSHUGS 0x0F35 == code || // Mn TIBETAN MARK NGAS BZUNG NYI ZLA 0x0F37 == code || // Mn TIBETAN MARK NGAS BZUNG SGOR RTAGS 0x0F39 == code || // Mn TIBETAN MARK TSA -PHRU 0x0F71 <= code && code <= 0x0F7E || // Mn [14] TIBETAN VOWEL SIGN AA..TIBETAN SIGN RJES SU NGA RO 0x0F80 <= code && code <= 0x0F84 || // Mn [5] TIBETAN VOWEL SIGN REVERSED I..TIBETAN MARK HALANTA 0x0F86 <= code && code <= 0x0F87 || // Mn [2] TIBETAN SIGN LCI RTAGS..TIBETAN SIGN YANG RTAGS 0x0F8D <= code && code <= 0x0F97 || // Mn [11] TIBETAN SUBJOINED SIGN LCE TSA CAN..TIBETAN SUBJOINED LETTER JA 0x0F99 <= code && code <= 0x0FBC || // Mn [36] TIBETAN SUBJOINED LETTER NYA..TIBETAN SUBJOINED LETTER FIXED-FORM RA 0x0FC6 == code || // Mn TIBETAN SYMBOL PADMA GDAN 0x102D <= code && code <= 0x1030 || // Mn [4] MYANMAR VOWEL SIGN I..MYANMAR VOWEL SIGN UU 0x1032 <= code && code <= 0x1037 || // Mn [6] MYANMAR VOWEL SIGN AI..MYANMAR SIGN DOT BELOW 0x1039 <= code && code <= 0x103A || // Mn [2] MYANMAR SIGN VIRAMA..MYANMAR SIGN ASAT 0x103D <= code && code <= 0x103E || // Mn [2] MYANMAR CONSONANT SIGN MEDIAL WA..MYANMAR CONSONANT SIGN MEDIAL HA 0x1058 <= code && code <= 0x1059 || // Mn [2] MYANMAR VOWEL SIGN VOCALIC L..MYANMAR VOWEL SIGN VOCALIC LL 0x105E <= code && code <= 0x1060 || // Mn [3] MYANMAR CONSONANT SIGN MON MEDIAL NA..MYANMAR CONSONANT SIGN MON MEDIAL LA 0x1071 <= code && code <= 0x1074 || // Mn [4] MYANMAR VOWEL SIGN GEBA KAREN I..MYANMAR VOWEL SIGN KAYAH EE 0x1082 == code || // Mn MYANMAR CONSONANT SIGN SHAN MEDIAL WA 0x1085 <= code && code <= 0x1086 || // Mn [2] MYANMAR VOWEL SIGN SHAN E ABOVE..MYANMAR VOWEL SIGN SHAN FINAL Y 0x108D == code || // Mn MYANMAR SIGN SHAN COUNCIL EMPHATIC TONE 0x109D == code || // Mn MYANMAR VOWEL SIGN AITON AI 0x135D <= code && code <= 0x135F || // Mn [3] ETHIOPIC COMBINING GEMINATION AND VOWEL LENGTH MARK..ETHIOPIC COMBINING GEMINATION MARK 0x1712 <= code && code <= 0x1714 || // Mn [3] TAGALOG VOWEL SIGN I..TAGALOG SIGN VIRAMA 0x1732 <= code && code <= 0x1734 || // Mn [3] HANUNOO VOWEL SIGN I..HANUNOO SIGN PAMUDPOD 0x1752 <= code && code <= 0x1753 || // Mn [2] BUHID VOWEL SIGN I..BUHID VOWEL SIGN U 0x1772 <= code && code <= 0x1773 || // Mn [2] TAGBANWA VOWEL SIGN I..TAGBANWA VOWEL SIGN U 0x17B4 <= code && code <= 0x17B5 || // Mn [2] KHMER VOWEL INHERENT AQ..KHMER VOWEL INHERENT AA 0x17B7 <= code && code <= 0x17BD || // Mn [7] KHMER VOWEL SIGN I..KHMER VOWEL SIGN UA 0x17C6 == code || // Mn KHMER SIGN NIKAHIT 0x17C9 <= code && code <= 0x17D3 || // Mn [11] KHMER SIGN MUUSIKATOAN..KHMER SIGN BATHAMASAT 0x17DD == code || // Mn KHMER SIGN ATTHACAN 0x180B <= code && code <= 0x180D || // Mn [3] MONGOLIAN FREE VARIATION SELECTOR ONE..MONGOLIAN FREE VARIATION SELECTOR THREE 0x1885 <= code && code <= 0x1886 || // Mn [2] MONGOLIAN LETTER ALI GALI BALUDA..MONGOLIAN LETTER ALI GALI THREE BALUDA 0x18A9 == code || // Mn MONGOLIAN LETTER ALI GALI DAGALGA 0x1920 <= code && code <= 0x1922 || // Mn [3] LIMBU VOWEL SIGN A..LIMBU VOWEL SIGN U 0x1927 <= code && code <= 0x1928 || // Mn [2] LIMBU VOWEL SIGN E..LIMBU VOWEL SIGN O 0x1932 == code || // Mn LIMBU SMALL LETTER ANUSVARA 0x1939 <= code && code <= 0x193B || // Mn [3] LIMBU SIGN MUKPHRENG..LIMBU SIGN SA-I 0x1A17 <= code && code <= 0x1A18 || // Mn [2] BUGINESE VOWEL SIGN I..BUGINESE VOWEL SIGN U 0x1A1B == code || // Mn BUGINESE VOWEL SIGN AE 0x1A56 == code || // Mn TAI THAM CONSONANT SIGN MEDIAL LA 0x1A58 <= code && code <= 0x1A5E || // Mn [7] TAI THAM SIGN MAI KANG LAI..TAI THAM CONSONANT SIGN SA 0x1A60 == code || // Mn TAI THAM SIGN SAKOT 0x1A62 == code || // Mn TAI THAM VOWEL SIGN MAI SAT 0x1A65 <= code && code <= 0x1A6C || // Mn [8] TAI THAM VOWEL SIGN I..TAI THAM VOWEL SIGN OA BELOW 0x1A73 <= code && code <= 0x1A7C || // Mn [10] TAI THAM VOWEL SIGN OA ABOVE..TAI THAM SIGN KHUEN-LUE KARAN 0x1A7F == code || // Mn TAI THAM COMBINING CRYPTOGRAMMIC DOT 0x1AB0 <= code && code <= 0x1ABD || // Mn [14] COMBINING DOUBLED CIRCUMFLEX ACCENT..COMBINING PARENTHESES BELOW 0x1ABE == code || // Me COMBINING PARENTHESES OVERLAY 0x1B00 <= code && code <= 0x1B03 || // Mn [4] BALINESE SIGN ULU RICEM..BALINESE SIGN SURANG 0x1B34 == code || // Mn BALINESE SIGN REREKAN 0x1B36 <= code && code <= 0x1B3A || // Mn [5] BALINESE VOWEL SIGN ULU..BALINESE VOWEL SIGN RA REPA 0x1B3C == code || // Mn BALINESE VOWEL SIGN LA LENGA 0x1B42 == code || // Mn BALINESE VOWEL SIGN PEPET 0x1B6B <= code && code <= 0x1B73 || // Mn [9] BALINESE MUSICAL SYMBOL COMBINING TEGEH..BALINESE MUSICAL SYMBOL COMBINING GONG 0x1B80 <= code && code <= 0x1B81 || // Mn [2] SUNDANESE SIGN PANYECEK..SUNDANESE SIGN PANGLAYAR 0x1BA2 <= code && code <= 0x1BA5 || // Mn [4] SUNDANESE CONSONANT SIGN PANYAKRA..SUNDANESE VOWEL SIGN PANYUKU 0x1BA8 <= code && code <= 0x1BA9 || // Mn [2] SUNDANESE VOWEL SIGN PAMEPET..SUNDANESE VOWEL SIGN PANEULEUNG 0x1BAB <= code && code <= 0x1BAD || // Mn [3] SUNDANESE SIGN VIRAMA..SUNDANESE CONSONANT SIGN PASANGAN WA 0x1BE6 == code || // Mn BATAK SIGN TOMPI 0x1BE8 <= code && code <= 0x1BE9 || // Mn [2] BATAK VOWEL SIGN PAKPAK E..BATAK VOWEL SIGN EE 0x1BED == code || // Mn BATAK VOWEL SIGN KARO O 0x1BEF <= code && code <= 0x1BF1 || // Mn [3] BATAK VOWEL SIGN U FOR SIMALUNGUN SA..BATAK CONSONANT SIGN H 0x1C2C <= code && code <= 0x1C33 || // Mn [8] LEPCHA VOWEL SIGN E..LEPCHA CONSONANT SIGN T 0x1C36 <= code && code <= 0x1C37 || // Mn [2] LEPCHA SIGN RAN..LEPCHA SIGN NUKTA 0x1CD0 <= code && code <= 0x1CD2 || // Mn [3] VEDIC TONE KARSHANA..VEDIC TONE PRENKHA 0x1CD4 <= code && code <= 0x1CE0 || // Mn [13] VEDIC SIGN YAJURVEDIC MIDLINE SVARITA..VEDIC TONE RIGVEDIC KASHMIRI INDEPENDENT SVARITA 0x1CE2 <= code && code <= 0x1CE8 || // Mn [7] VEDIC SIGN VISARGA SVARITA..VEDIC SIGN VISARGA ANUDATTA WITH TAIL 0x1CED == code || // Mn VEDIC SIGN TIRYAK 0x1CF4 == code || // Mn VEDIC TONE CANDRA ABOVE 0x1CF8 <= code && code <= 0x1CF9 || // Mn [2] VEDIC TONE RING ABOVE..VEDIC TONE DOUBLE RING ABOVE 0x1DC0 <= code && code <= 0x1DF9 || // Mn [58] COMBINING DOTTED GRAVE ACCENT..COMBINING WIDE INVERTED BRIDGE BELOW 0x1DFB <= code && code <= 0x1DFF || // Mn [5] COMBINING DELETION MARK..COMBINING RIGHT ARROWHEAD AND DOWN ARROWHEAD BELOW 0x200C == code || // Cf ZERO WIDTH NON-JOINER 0x20D0 <= code && code <= 0x20DC || // Mn [13] COMBINING LEFT HARPOON ABOVE..COMBINING FOUR DOTS ABOVE 0x20DD <= code && code <= 0x20E0 || // Me [4] COMBINING ENCLOSING CIRCLE..COMBINING ENCLOSING CIRCLE BACKSLASH 0x20E1 == code || // Mn COMBINING LEFT RIGHT ARROW ABOVE 0x20E2 <= code && code <= 0x20E4 || // Me [3] COMBINING ENCLOSING SCREEN..COMBINING ENCLOSING UPWARD POINTING TRIANGLE 0x20E5 <= code && code <= 0x20F0 || // Mn [12] COMBINING REVERSE SOLIDUS OVERLAY..COMBINING ASTERISK ABOVE 0x2CEF <= code && code <= 0x2CF1 || // Mn [3] COPTIC COMBINING NI ABOVE..COPTIC COMBINING SPIRITUS LENIS 0x2D7F == code || // Mn TIFINAGH CONSONANT JOINER 0x2DE0 <= code && code <= 0x2DFF || // Mn [32] COMBINING CYRILLIC LETTER BE..COMBINING CYRILLIC LETTER IOTIFIED BIG YUS 0x302A <= code && code <= 0x302D || // Mn [4] IDEOGRAPHIC LEVEL TONE MARK..IDEOGRAPHIC ENTERING TONE MARK 0x302E <= code && code <= 0x302F || // Mc [2] HANGUL SINGLE DOT TONE MARK..HANGUL DOUBLE DOT TONE MARK 0x3099 <= code && code <= 0x309A || // Mn [2] COMBINING KATAKANA-HIRAGANA VOICED SOUND MARK..COMBINING KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK 0xA66F == code || // Mn COMBINING CYRILLIC VZMET 0xA670 <= code && code <= 0xA672 || // Me [3] COMBINING CYRILLIC TEN MILLIONS SIGN..COMBINING CYRILLIC THOUSAND MILLIONS SIGN 0xA674 <= code && code <= 0xA67D || // Mn [10] COMBINING CYRILLIC LETTER UKRAINIAN IE..COMBINING CYRILLIC PAYEROK 0xA69E <= code && code <= 0xA69F || // Mn [2] COMBINING CYRILLIC LETTER EF..COMBINING CYRILLIC LETTER IOTIFIED E 0xA6F0 <= code && code <= 0xA6F1 || // Mn [2] BAMUM COMBINING MARK KOQNDON..BAMUM COMBINING MARK TUKWENTIS 0xA802 == code || // Mn SYLOTI NAGRI SIGN DVISVARA 0xA806 == code || // Mn SYLOTI NAGRI SIGN HASANTA 0xA80B == code || // Mn SYLOTI NAGRI SIGN ANUSVARA 0xA825 <= code && code <= 0xA826 || // Mn [2] SYLOTI NAGRI VOWEL SIGN U..SYLOTI NAGRI VOWEL SIGN E 0xA8C4 <= code && code <= 0xA8C5 || // Mn [2] SAURASHTRA SIGN VIRAMA..SAURASHTRA SIGN CANDRABINDU 0xA8E0 <= code && code <= 0xA8F1 || // Mn [18] COMBINING DEVANAGARI DIGIT ZERO..COMBINING DEVANAGARI SIGN AVAGRAHA 0xA926 <= code && code <= 0xA92D || // Mn [8] KAYAH LI VOWEL UE..KAYAH LI TONE CALYA PLOPHU 0xA947 <= code && code <= 0xA951 || // Mn [11] REJANG VOWEL SIGN I..REJANG CONSONANT SIGN R 0xA980 <= code && code <= 0xA982 || // Mn [3] JAVANESE SIGN PANYANGGA..JAVANESE SIGN LAYAR 0xA9B3 == code || // Mn JAVANESE SIGN CECAK TELU 0xA9B6 <= code && code <= 0xA9B9 || // Mn [4] JAVANESE VOWEL SIGN WULU..JAVANESE VOWEL SIGN SUKU MENDUT 0xA9BC == code || // Mn JAVANESE VOWEL SIGN PEPET 0xA9E5 == code || // Mn MYANMAR SIGN SHAN SAW 0xAA29 <= code && code <= 0xAA2E || // Mn [6] CHAM VOWEL SIGN AA..CHAM VOWEL SIGN OE 0xAA31 <= code && code <= 0xAA32 || // Mn [2] CHAM VOWEL SIGN AU..CHAM VOWEL SIGN UE 0xAA35 <= code && code <= 0xAA36 || // Mn [2] CHAM CONSONANT SIGN LA..CHAM CONSONANT SIGN WA 0xAA43 == code || // Mn CHAM CONSONANT SIGN FINAL NG 0xAA4C == code || // Mn CHAM CONSONANT SIGN FINAL M 0xAA7C == code || // Mn MYANMAR SIGN TAI LAING TONE-2 0xAAB0 == code || // Mn TAI VIET MAI KANG 0xAAB2 <= code && code <= 0xAAB4 || // Mn [3] TAI VIET VOWEL I..TAI VIET VOWEL U 0xAAB7 <= code && code <= 0xAAB8 || // Mn [2] TAI VIET MAI KHIT..TAI VIET VOWEL IA 0xAABE <= code && code <= 0xAABF || // Mn [2] TAI VIET VOWEL AM..TAI VIET TONE MAI EK 0xAAC1 == code || // Mn TAI VIET TONE MAI THO 0xAAEC <= code && code <= 0xAAED || // Mn [2] MEETEI MAYEK VOWEL SIGN UU..MEETEI MAYEK VOWEL SIGN AAI 0xAAF6 == code || // Mn MEETEI MAYEK VIRAMA 0xABE5 == code || // Mn MEETEI MAYEK VOWEL SIGN ANAP 0xABE8 == code || // Mn MEETEI MAYEK VOWEL SIGN UNAP 0xABED == code || // Mn MEETEI MAYEK APUN IYEK 0xFB1E == code || // Mn HEBREW POINT JUDEO-SPANISH VARIKA 0xFE00 <= code && code <= 0xFE0F || // Mn [16] VARIATION SELECTOR-1..VARIATION SELECTOR-16 0xFE20 <= code && code <= 0xFE2F || // Mn [16] COMBINING LIGATURE LEFT HALF..COMBINING CYRILLIC TITLO RIGHT HALF 0xFF9E <= code && code <= 0xFF9F || // Lm [2] HALFWIDTH KATAKANA VOICED SOUND MARK..HALFWIDTH KATAKANA SEMI-VOICED SOUND MARK 0x101FD == code || // Mn PHAISTOS DISC SIGN COMBINING OBLIQUE STROKE 0x102E0 == code || // Mn COPTIC EPACT THOUSANDS MARK 0x10376 <= code && code <= 0x1037A || // Mn [5] COMBINING OLD PERMIC LETTER AN..COMBINING OLD PERMIC LETTER SII 0x10A01 <= code && code <= 0x10A03 || // Mn [3] KHAROSHTHI VOWEL SIGN I..KHAROSHTHI VOWEL SIGN VOCALIC R 0x10A05 <= code && code <= 0x10A06 || // Mn [2] KHAROSHTHI VOWEL SIGN E..KHAROSHTHI VOWEL SIGN O 0x10A0C <= code && code <= 0x10A0F || // Mn [4] KHAROSHTHI VOWEL LENGTH MARK..KHAROSHTHI SIGN VISARGA 0x10A38 <= code && code <= 0x10A3A || // Mn [3] KHAROSHTHI SIGN BAR ABOVE..KHAROSHTHI SIGN DOT BELOW 0x10A3F == code || // Mn KHAROSHTHI VIRAMA 0x10AE5 <= code && code <= 0x10AE6 || // Mn [2] MANICHAEAN ABBREVIATION MARK ABOVE..MANICHAEAN ABBREVIATION MARK BELOW 0x11001 == code || // Mn BRAHMI SIGN ANUSVARA 0x11038 <= code && code <= 0x11046 || // Mn [15] BRAHMI VOWEL SIGN AA..BRAHMI VIRAMA 0x1107F <= code && code <= 0x11081 || // Mn [3] BRAHMI NUMBER JOINER..KAITHI SIGN ANUSVARA 0x110B3 <= code && code <= 0x110B6 || // Mn [4] KAITHI VOWEL SIGN U..KAITHI VOWEL SIGN AI 0x110B9 <= code && code <= 0x110BA || // Mn [2] KAITHI SIGN VIRAMA..KAITHI SIGN NUKTA 0x11100 <= code && code <= 0x11102 || // Mn [3] CHAKMA SIGN CANDRABINDU..CHAKMA SIGN VISARGA 0x11127 <= code && code <= 0x1112B || // Mn [5] CHAKMA VOWEL SIGN A..CHAKMA VOWEL SIGN UU 0x1112D <= code && code <= 0x11134 || // Mn [8] CHAKMA VOWEL SIGN AI..CHAKMA MAAYYAA 0x11173 == code || // Mn MAHAJANI SIGN NUKTA 0x11180 <= code && code <= 0x11181 || // Mn [2] SHARADA SIGN CANDRABINDU..SHARADA SIGN ANUSVARA 0x111B6 <= code && code <= 0x111BE || // Mn [9] SHARADA VOWEL SIGN U..SHARADA VOWEL SIGN O 0x111CA <= code && code <= 0x111CC || // Mn [3] SHARADA SIGN NUKTA..SHARADA EXTRA SHORT VOWEL MARK 0x1122F <= code && code <= 0x11231 || // Mn [3] KHOJKI VOWEL SIGN U..KHOJKI VOWEL SIGN AI 0x11234 == code || // Mn KHOJKI SIGN ANUSVARA 0x11236 <= code && code <= 0x11237 || // Mn [2] KHOJKI SIGN NUKTA..KHOJKI SIGN SHADDA 0x1123E == code || // Mn KHOJKI SIGN SUKUN 0x112DF == code || // Mn KHUDAWADI SIGN ANUSVARA 0x112E3 <= code && code <= 0x112EA || // Mn [8] KHUDAWADI VOWEL SIGN U..KHUDAWADI SIGN VIRAMA 0x11300 <= code && code <= 0x11301 || // Mn [2] GRANTHA SIGN COMBINING ANUSVARA ABOVE..GRANTHA SIGN CANDRABINDU 0x1133C == code || // Mn GRANTHA SIGN NUKTA 0x1133E == code || // Mc GRANTHA VOWEL SIGN AA 0x11340 == code || // Mn GRANTHA VOWEL SIGN II 0x11357 == code || // Mc GRANTHA AU LENGTH MARK 0x11366 <= code && code <= 0x1136C || // Mn [7] COMBINING GRANTHA DIGIT ZERO..COMBINING GRANTHA DIGIT SIX 0x11370 <= code && code <= 0x11374 || // Mn [5] COMBINING GRANTHA LETTER A..COMBINING GRANTHA LETTER PA 0x11438 <= code && code <= 0x1143F || // Mn [8] NEWA VOWEL SIGN U..NEWA VOWEL SIGN AI 0x11442 <= code && code <= 0x11444 || // Mn [3] NEWA SIGN VIRAMA..NEWA SIGN ANUSVARA 0x11446 == code || // Mn NEWA SIGN NUKTA 0x114B0 == code || // Mc TIRHUTA VOWEL SIGN AA 0x114B3 <= code && code <= 0x114B8 || // Mn [6] TIRHUTA VOWEL SIGN U..TIRHUTA VOWEL SIGN VOCALIC LL 0x114BA == code || // Mn TIRHUTA VOWEL SIGN SHORT E 0x114BD == code || // Mc TIRHUTA VOWEL SIGN SHORT O 0x114BF <= code && code <= 0x114C0 || // Mn [2] TIRHUTA SIGN CANDRABINDU..TIRHUTA SIGN ANUSVARA 0x114C2 <= code && code <= 0x114C3 || // Mn [2] TIRHUTA SIGN VIRAMA..TIRHUTA SIGN NUKTA 0x115AF == code || // Mc SIDDHAM VOWEL SIGN AA 0x115B2 <= code && code <= 0x115B5 || // Mn [4] SIDDHAM VOWEL SIGN U..SIDDHAM VOWEL SIGN VOCALIC RR 0x115BC <= code && code <= 0x115BD || // Mn [2] SIDDHAM SIGN CANDRABINDU..SIDDHAM SIGN ANUSVARA 0x115BF <= code && code <= 0x115C0 || // Mn [2] SIDDHAM SIGN VIRAMA..SIDDHAM SIGN NUKTA 0x115DC <= code && code <= 0x115DD || // Mn [2] SIDDHAM VOWEL SIGN ALTERNATE U..SIDDHAM VOWEL SIGN ALTERNATE UU 0x11633 <= code && code <= 0x1163A || // Mn [8] MODI VOWEL SIGN U..MODI VOWEL SIGN AI 0x1163D == code || // Mn MODI SIGN ANUSVARA 0x1163F <= code && code <= 0x11640 || // Mn [2] MODI SIGN VIRAMA..MODI SIGN ARDHACANDRA 0x116AB == code || // Mn TAKRI SIGN ANUSVARA 0x116AD == code || // Mn TAKRI VOWEL SIGN AA 0x116B0 <= code && code <= 0x116B5 || // Mn [6] TAKRI VOWEL SIGN U..TAKRI VOWEL SIGN AU 0x116B7 == code || // Mn TAKRI SIGN NUKTA 0x1171D <= code && code <= 0x1171F || // Mn [3] AHOM CONSONANT SIGN MEDIAL LA..AHOM CONSONANT SIGN MEDIAL LIGATING RA 0x11722 <= code && code <= 0x11725 || // Mn [4] AHOM VOWEL SIGN I..AHOM VOWEL SIGN UU 0x11727 <= code && code <= 0x1172B || // Mn [5] AHOM VOWEL SIGN AW..AHOM SIGN KILLER 0x11A01 <= code && code <= 0x11A06 || // Mn [6] ZANABAZAR SQUARE VOWEL SIGN I..ZANABAZAR SQUARE VOWEL SIGN O 0x11A09 <= code && code <= 0x11A0A || // Mn [2] ZANABAZAR SQUARE VOWEL SIGN REVERSED I..ZANABAZAR SQUARE VOWEL LENGTH MARK 0x11A33 <= code && code <= 0x11A38 || // Mn [6] ZANABAZAR SQUARE FINAL CONSONANT MARK..ZANABAZAR SQUARE SIGN ANUSVARA 0x11A3B <= code && code <= 0x11A3E || // Mn [4] ZANABAZAR SQUARE CLUSTER-FINAL LETTER YA..ZANABAZAR SQUARE CLUSTER-FINAL LETTER VA 0x11A47 == code || // Mn ZANABAZAR SQUARE SUBJOINER 0x11A51 <= code && code <= 0x11A56 || // Mn [6] SOYOMBO VOWEL SIGN I..SOYOMBO VOWEL SIGN OE 0x11A59 <= code && code <= 0x11A5B || // Mn [3] SOYOMBO VOWEL SIGN VOCALIC R..SOYOMBO VOWEL LENGTH MARK 0x11A8A <= code && code <= 0x11A96 || // Mn [13] SOYOMBO FINAL CONSONANT SIGN G..SOYOMBO SIGN ANUSVARA 0x11A98 <= code && code <= 0x11A99 || // Mn [2] SOYOMBO GEMINATION MARK..SOYOMBO SUBJOINER 0x11C30 <= code && code <= 0x11C36 || // Mn [7] BHAIKSUKI VOWEL SIGN I..BHAIKSUKI VOWEL SIGN VOCALIC L 0x11C38 <= code && code <= 0x11C3D || // Mn [6] BHAIKSUKI VOWEL SIGN E..BHAIKSUKI SIGN ANUSVARA 0x11C3F == code || // Mn BHAIKSUKI SIGN VIRAMA 0x11C92 <= code && code <= 0x11CA7 || // Mn [22] MARCHEN SUBJOINED LETTER KA..MARCHEN SUBJOINED LETTER ZA 0x11CAA <= code && code <= 0x11CB0 || // Mn [7] MARCHEN SUBJOINED LETTER RA..MARCHEN VOWEL SIGN AA 0x11CB2 <= code && code <= 0x11CB3 || // Mn [2] MARCHEN VOWEL SIGN U..MARCHEN VOWEL SIGN E 0x11CB5 <= code && code <= 0x11CB6 || // Mn [2] MARCHEN SIGN ANUSVARA..MARCHEN SIGN CANDRABINDU 0x11D31 <= code && code <= 0x11D36 || // Mn [6] MASARAM GONDI VOWEL SIGN AA..MASARAM GONDI VOWEL SIGN VOCALIC R 0x11D3A == code || // Mn MASARAM GONDI VOWEL SIGN E 0x11D3C <= code && code <= 0x11D3D || // Mn [2] MASARAM GONDI VOWEL SIGN AI..MASARAM GONDI VOWEL SIGN O 0x11D3F <= code && code <= 0x11D45 || // Mn [7] MASARAM GONDI VOWEL SIGN AU..MASARAM GONDI VIRAMA 0x11D47 == code || // Mn MASARAM GONDI RA-KARA 0x16AF0 <= code && code <= 0x16AF4 || // Mn [5] BASSA VAH COMBINING HIGH TONE..BASSA VAH COMBINING HIGH-LOW TONE 0x16B30 <= code && code <= 0x16B36 || // Mn [7] PAHAWH HMONG MARK CIM TUB..PAHAWH HMONG MARK CIM TAUM 0x16F8F <= code && code <= 0x16F92 || // Mn [4] MIAO TONE RIGHT..MIAO TONE BELOW 0x1BC9D <= code && code <= 0x1BC9E || // Mn [2] DUPLOYAN THICK LETTER SELECTOR..DUPLOYAN DOUBLE MARK 0x1D165 == code || // Mc MUSICAL SYMBOL COMBINING STEM 0x1D167 <= code && code <= 0x1D169 || // Mn [3] MUSICAL SYMBOL COMBINING TREMOLO-1..MUSICAL SYMBOL COMBINING TREMOLO-3 0x1D16E <= code && code <= 0x1D172 || // Mc [5] MUSICAL SYMBOL COMBINING FLAG-1..MUSICAL SYMBOL COMBINING FLAG-5 0x1D17B <= code && code <= 0x1D182 || // Mn [8] MUSICAL SYMBOL COMBINING ACCENT..MUSICAL SYMBOL COMBINING LOURE 0x1D185 <= code && code <= 0x1D18B || // Mn [7] MUSICAL SYMBOL COMBINING DOIT..MUSICAL SYMBOL COMBINING TRIPLE TONGUE 0x1D1AA <= code && code <= 0x1D1AD || // Mn [4] MUSICAL SYMBOL COMBINING DOWN BOW..MUSICAL SYMBOL COMBINING SNAP PIZZICATO 0x1D242 <= code && code <= 0x1D244 || // Mn [3] COMBINING GREEK MUSICAL TRISEME..COMBINING GREEK MUSICAL PENTASEME 0x1DA00 <= code && code <= 0x1DA36 || // Mn [55] SIGNWRITING HEAD RIM..SIGNWRITING AIR SUCKING IN 0x1DA3B <= code && code <= 0x1DA6C || // Mn [50] SIGNWRITING MOUTH CLOSED NEUTRAL..SIGNWRITING EXCITEMENT 0x1DA75 == code || // Mn SIGNWRITING UPPER BODY TILTING FROM HIP JOINTS 0x1DA84 == code || // Mn SIGNWRITING LOCATION HEAD NECK 0x1DA9B <= code && code <= 0x1DA9F || // Mn [5] SIGNWRITING FILL MODIFIER-2..SIGNWRITING FILL MODIFIER-6 0x1DAA1 <= code && code <= 0x1DAAF || // Mn [15] SIGNWRITING ROTATION MODIFIER-2..SIGNWRITING ROTATION MODIFIER-16 0x1E000 <= code && code <= 0x1E006 || // Mn [7] COMBINING GLAGOLITIC LETTER AZU..COMBINING GLAGOLITIC LETTER ZHIVETE 0x1E008 <= code && code <= 0x1E018 || // Mn [17] COMBINING GLAGOLITIC LETTER ZEMLJA..COMBINING GLAGOLITIC LETTER HERU 0x1E01B <= code && code <= 0x1E021 || // Mn [7] COMBINING GLAGOLITIC LETTER SHTA..COMBINING GLAGOLITIC LETTER YATI 0x1E023 <= code && code <= 0x1E024 || // Mn [2] COMBINING GLAGOLITIC LETTER YU..COMBINING GLAGOLITIC LETTER SMALL YUS 0x1E026 <= code && code <= 0x1E02A || // Mn [5] COMBINING GLAGOLITIC LETTER YO..COMBINING GLAGOLITIC LETTER FITA 0x1E8D0 <= code && code <= 0x1E8D6 || // Mn [7] MENDE KIKAKUI COMBINING NUMBER TEENS..MENDE KIKAKUI COMBINING NUMBER MILLIONS 0x1E944 <= code && code <= 0x1E94A || // Mn [7] ADLAM ALIF LENGTHENER..ADLAM NUKTA 0xE0020 <= code && code <= 0xE007F || // Cf [96] TAG SPACE..CANCEL TAG 0xE0100 <= code && code <= 0xE01EF // Mn [240] VARIATION SELECTOR-17..VARIATION SELECTOR-256 ) { return Extend; } if (0x1F1E6 <= code && code <= 0x1F1FF) // So [26] REGIONAL INDICATOR SYMBOL LETTER A..REGIONAL INDICATOR SYMBOL LETTER Z { return Regional_Indicator; } if (0x0903 == code || // Mc DEVANAGARI SIGN VISARGA 0x093B == code || // Mc DEVANAGARI VOWEL SIGN OOE 0x093E <= code && code <= 0x0940 || // Mc [3] DEVANAGARI VOWEL SIGN AA..DEVANAGARI VOWEL SIGN II 0x0949 <= code && code <= 0x094C || // Mc [4] DEVANAGARI VOWEL SIGN CANDRA O..DEVANAGARI VOWEL SIGN AU 0x094E <= code && code <= 0x094F || // Mc [2] DEVANAGARI VOWEL SIGN PRISHTHAMATRA E..DEVANAGARI VOWEL SIGN AW 0x0982 <= code && code <= 0x0983 || // Mc [2] BENGALI SIGN ANUSVARA..BENGALI SIGN VISARGA 0x09BF <= code && code <= 0x09C0 || // Mc [2] BENGALI VOWEL SIGN I..BENGALI VOWEL SIGN II 0x09C7 <= code && code <= 0x09C8 || // Mc [2] BENGALI VOWEL SIGN E..BENGALI VOWEL SIGN AI 0x09CB <= code && code <= 0x09CC || // Mc [2] BENGALI VOWEL SIGN O..BENGALI VOWEL SIGN AU 0x0A03 == code || // Mc GURMUKHI SIGN VISARGA 0x0A3E <= code && code <= 0x0A40 || // Mc [3] GURMUKHI VOWEL SIGN AA..GURMUKHI VOWEL SIGN II 0x0A83 == code || // Mc GUJARATI SIGN VISARGA 0x0ABE <= code && code <= 0x0AC0 || // Mc [3] GUJARATI VOWEL SIGN AA..GUJARATI VOWEL SIGN II 0x0AC9 == code || // Mc GUJARATI VOWEL SIGN CANDRA O 0x0ACB <= code && code <= 0x0ACC || // Mc [2] GUJARATI VOWEL SIGN O..GUJARATI VOWEL SIGN AU 0x0B02 <= code && code <= 0x0B03 || // Mc [2] ORIYA SIGN ANUSVARA..ORIYA SIGN VISARGA 0x0B40 == code || // Mc ORIYA VOWEL SIGN II 0x0B47 <= code && code <= 0x0B48 || // Mc [2] ORIYA VOWEL SIGN E..ORIYA VOWEL SIGN AI 0x0B4B <= code && code <= 0x0B4C || // Mc [2] ORIYA VOWEL SIGN O..ORIYA VOWEL SIGN AU 0x0BBF == code || // Mc TAMIL VOWEL SIGN I 0x0BC1 <= code && code <= 0x0BC2 || // Mc [2] TAMIL VOWEL SIGN U..TAMIL VOWEL SIGN UU 0x0BC6 <= code && code <= 0x0BC8 || // Mc [3] TAMIL VOWEL SIGN E..TAMIL VOWEL SIGN AI 0x0BCA <= code && code <= 0x0BCC || // Mc [3] TAMIL VOWEL SIGN O..TAMIL VOWEL SIGN AU 0x0C01 <= code && code <= 0x0C03 || // Mc [3] TELUGU SIGN CANDRABINDU..TELUGU SIGN VISARGA 0x0C41 <= code && code <= 0x0C44 || // Mc [4] TELUGU VOWEL SIGN U..TELUGU VOWEL SIGN VOCALIC RR 0x0C82 <= code && code <= 0x0C83 || // Mc [2] KANNADA SIGN ANUSVARA..KANNADA SIGN VISARGA 0x0CBE == code || // Mc KANNADA VOWEL SIGN AA 0x0CC0 <= code && code <= 0x0CC1 || // Mc [2] KANNADA VOWEL SIGN II..KANNADA VOWEL SIGN U 0x0CC3 <= code && code <= 0x0CC4 || // Mc [2] KANNADA VOWEL SIGN VOCALIC R..KANNADA VOWEL SIGN VOCALIC RR 0x0CC7 <= code && code <= 0x0CC8 || // Mc [2] KANNADA VOWEL SIGN EE..KANNADA VOWEL SIGN AI 0x0CCA <= code && code <= 0x0CCB || // Mc [2] KANNADA VOWEL SIGN O..KANNADA VOWEL SIGN OO 0x0D02 <= code && code <= 0x0D03 || // Mc [2] MALAYALAM SIGN ANUSVARA..MALAYALAM SIGN VISARGA 0x0D3F <= code && code <= 0x0D40 || // Mc [2] MALAYALAM VOWEL SIGN I..MALAYALAM VOWEL SIGN II 0x0D46 <= code && code <= 0x0D48 || // Mc [3] MALAYALAM VOWEL SIGN E..MALAYALAM VOWEL SIGN AI 0x0D4A <= code && code <= 0x0D4C || // Mc [3] MALAYALAM VOWEL SIGN O..MALAYALAM VOWEL SIGN AU 0x0D82 <= code && code <= 0x0D83 || // Mc [2] SINHALA SIGN ANUSVARAYA..SINHALA SIGN VISARGAYA 0x0DD0 <= code && code <= 0x0DD1 || // Mc [2] SINHALA VOWEL SIGN KETTI AEDA-PILLA..SINHALA VOWEL SIGN DIGA AEDA-PILLA 0x0DD8 <= code && code <= 0x0DDE || // Mc [7] SINHALA VOWEL SIGN GAETTA-PILLA..SINHALA VOWEL SIGN KOMBUVA HAA GAYANUKITTA 0x0DF2 <= code && code <= 0x0DF3 || // Mc [2] SINHALA VOWEL SIGN DIGA GAETTA-PILLA..SINHALA VOWEL SIGN DIGA GAYANUKITTA 0x0E33 == code || // Lo THAI CHARACTER SARA AM 0x0EB3 == code || // Lo LAO VOWEL SIGN AM 0x0F3E <= code && code <= 0x0F3F || // Mc [2] TIBETAN SIGN YAR TSHES..TIBETAN SIGN MAR TSHES 0x0F7F == code || // Mc TIBETAN SIGN RNAM BCAD 0x1031 == code || // Mc MYANMAR VOWEL SIGN E 0x103B <= code && code <= 0x103C || // Mc [2] MYANMAR CONSONANT SIGN MEDIAL YA..MYANMAR CONSONANT SIGN MEDIAL RA 0x1056 <= code && code <= 0x1057 || // Mc [2] MYANMAR VOWEL SIGN VOCALIC R..MYANMAR VOWEL SIGN VOCALIC RR 0x1084 == code || // Mc MYANMAR VOWEL SIGN SHAN E 0x17B6 == code || // Mc KHMER VOWEL SIGN AA 0x17BE <= code && code <= 0x17C5 || // Mc [8] KHMER VOWEL SIGN OE..KHMER VOWEL SIGN AU 0x17C7 <= code && code <= 0x17C8 || // Mc [2] KHMER SIGN REAHMUK..KHMER SIGN YUUKALEAPINTU 0x1923 <= code && code <= 0x1926 || // Mc [4] LIMBU VOWEL SIGN EE..LIMBU VOWEL SIGN AU 0x1929 <= code && code <= 0x192B || // Mc [3] LIMBU SUBJOINED LETTER YA..LIMBU SUBJOINED LETTER WA 0x1930 <= code && code <= 0x1931 || // Mc [2] LIMBU SMALL LETTER KA..LIMBU SMALL LETTER NGA 0x1933 <= code && code <= 0x1938 || // Mc [6] LIMBU SMALL LETTER TA..LIMBU SMALL LETTER LA 0x1A19 <= code && code <= 0x1A1A || // Mc [2] BUGINESE VOWEL SIGN E..BUGINESE VOWEL SIGN O 0x1A55 == code || // Mc TAI THAM CONSONANT SIGN MEDIAL RA 0x1A57 == code || // Mc TAI THAM CONSONANT SIGN LA TANG LAI 0x1A6D <= code && code <= 0x1A72 || // Mc [6] TAI THAM VOWEL SIGN OY..TAI THAM VOWEL SIGN THAM AI 0x1B04 == code || // Mc BALINESE SIGN BISAH 0x1B35 == code || // Mc BALINESE VOWEL SIGN TEDUNG 0x1B3B == code || // Mc BALINESE VOWEL SIGN RA REPA TEDUNG 0x1B3D <= code && code <= 0x1B41 || // Mc [5] BALINESE VOWEL SIGN LA LENGA TEDUNG..BALINESE VOWEL SIGN TALING REPA TEDUNG 0x1B43 <= code && code <= 0x1B44 || // Mc [2] BALINESE VOWEL SIGN PEPET TEDUNG..BALINESE ADEG ADEG 0x1B82 == code || // Mc SUNDANESE SIGN PANGWISAD 0x1BA1 == code || // Mc SUNDANESE CONSONANT SIGN PAMINGKAL 0x1BA6 <= code && code <= 0x1BA7 || // Mc [2] SUNDANESE VOWEL SIGN PANAELAENG..SUNDANESE VOWEL SIGN PANOLONG 0x1BAA == code || // Mc SUNDANESE SIGN PAMAAEH 0x1BE7 == code || // Mc BATAK VOWEL SIGN E 0x1BEA <= code && code <= 0x1BEC || // Mc [3] BATAK VOWEL SIGN I..BATAK VOWEL SIGN O 0x1BEE == code || // Mc BATAK VOWEL SIGN U 0x1BF2 <= code && code <= 0x1BF3 || // Mc [2] BATAK PANGOLAT..BATAK PANONGONAN 0x1C24 <= code && code <= 0x1C2B || // Mc [8] LEPCHA SUBJOINED LETTER YA..LEPCHA VOWEL SIGN UU 0x1C34 <= code && code <= 0x1C35 || // Mc [2] LEPCHA CONSONANT SIGN NYIN-DO..LEPCHA CONSONANT SIGN KANG 0x1CE1 == code || // Mc VEDIC TONE ATHARVAVEDIC INDEPENDENT SVARITA 0x1CF2 <= code && code <= 0x1CF3 || // Mc [2] VEDIC SIGN ARDHAVISARGA..VEDIC SIGN ROTATED ARDHAVISARGA 0x1CF7 == code || // Mc VEDIC SIGN ATIKRAMA 0xA823 <= code && code <= 0xA824 || // Mc [2] SYLOTI NAGRI VOWEL SIGN A..SYLOTI NAGRI VOWEL SIGN I 0xA827 == code || // Mc SYLOTI NAGRI VOWEL SIGN OO 0xA880 <= code && code <= 0xA881 || // Mc [2] SAURASHTRA SIGN ANUSVARA..SAURASHTRA SIGN VISARGA 0xA8B4 <= code && code <= 0xA8C3 || // Mc [16] SAURASHTRA CONSONANT SIGN HAARU..SAURASHTRA VOWEL SIGN AU 0xA952 <= code && code <= 0xA953 || // Mc [2] REJANG CONSONANT SIGN H..REJANG VIRAMA 0xA983 == code || // Mc JAVANESE SIGN WIGNYAN 0xA9B4 <= code && code <= 0xA9B5 || // Mc [2] JAVANESE VOWEL SIGN TARUNG..JAVANESE VOWEL SIGN TOLONG 0xA9BA <= code && code <= 0xA9BB || // Mc [2] JAVANESE VOWEL SIGN TALING..JAVANESE VOWEL SIGN DIRGA MURE 0xA9BD <= code && code <= 0xA9C0 || // Mc [4] JAVANESE CONSONANT SIGN KERET..JAVANESE PANGKON 0xAA2F <= code && code <= 0xAA30 || // Mc [2] CHAM VOWEL SIGN O..CHAM VOWEL SIGN AI 0xAA33 <= code && code <= 0xAA34 || // Mc [2] CHAM CONSONANT SIGN YA..CHAM CONSONANT SIGN RA 0xAA4D == code || // Mc CHAM CONSONANT SIGN FINAL H 0xAAEB == code || // Mc MEETEI MAYEK VOWEL SIGN II 0xAAEE <= code && code <= 0xAAEF || // Mc [2] MEETEI MAYEK VOWEL SIGN AU..MEETEI MAYEK VOWEL SIGN AAU 0xAAF5 == code || // Mc MEETEI MAYEK VOWEL SIGN VISARGA 0xABE3 <= code && code <= 0xABE4 || // Mc [2] MEETEI MAYEK VOWEL SIGN ONAP..MEETEI MAYEK VOWEL SIGN INAP 0xABE6 <= code && code <= 0xABE7 || // Mc [2] MEETEI MAYEK VOWEL SIGN YENAP..MEETEI MAYEK VOWEL SIGN SOUNAP 0xABE9 <= code && code <= 0xABEA || // Mc [2] MEETEI MAYEK VOWEL SIGN CHEINAP..MEETEI MAYEK VOWEL SIGN NUNG 0xABEC == code || // Mc MEETEI MAYEK LUM IYEK 0x11000 == code || // Mc BRAHMI SIGN CANDRABINDU 0x11002 == code || // Mc BRAHMI SIGN VISARGA 0x11082 == code || // Mc KAITHI SIGN VISARGA 0x110B0 <= code && code <= 0x110B2 || // Mc [3] KAITHI VOWEL SIGN AA..KAITHI VOWEL SIGN II 0x110B7 <= code && code <= 0x110B8 || // Mc [2] KAITHI VOWEL SIGN O..KAITHI VOWEL SIGN AU 0x1112C == code || // Mc CHAKMA VOWEL SIGN E 0x11182 == code || // Mc SHARADA SIGN VISARGA 0x111B3 <= code && code <= 0x111B5 || // Mc [3] SHARADA VOWEL SIGN AA..SHARADA VOWEL SIGN II 0x111BF <= code && code <= 0x111C0 || // Mc [2] SHARADA VOWEL SIGN AU..SHARADA SIGN VIRAMA 0x1122C <= code && code <= 0x1122E || // Mc [3] KHOJKI VOWEL SIGN AA..KHOJKI VOWEL SIGN II 0x11232 <= code && code <= 0x11233 || // Mc [2] KHOJKI VOWEL SIGN O..KHOJKI VOWEL SIGN AU 0x11235 == code || // Mc KHOJKI SIGN VIRAMA 0x112E0 <= code && code <= 0x112E2 || // Mc [3] KHUDAWADI VOWEL SIGN AA..KHUDAWADI VOWEL SIGN II 0x11302 <= code && code <= 0x11303 || // Mc [2] GRANTHA SIGN ANUSVARA..GRANTHA SIGN VISARGA 0x1133F == code || // Mc GRANTHA VOWEL SIGN I 0x