charfold.cc #1

#include <stdhdrs.h>
#include <strbuf.h>
#include "i18napi.h"
#include "charcvt.h"

static const CharSetCvt::MapEnt caseFold[] = {
0x00B5, 0x03BC, // MICRO SIGN
0x00C0, 0x00E0, // LATIN CAPITAL LETTER A WITH GRAVE
0x00C1, 0x00E1, // LATIN CAPITAL LETTER A WITH ACUTE
0x00C2, 0x00E2, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX
0x00C3, 0x00E3, // LATIN CAPITAL LETTER A WITH TILDE
0x00C4, 0x00E4, // LATIN CAPITAL LETTER A WITH DIAERESIS
0x00C5, 0x00E5, // LATIN CAPITAL LETTER A WITH RING ABOVE
0x00C6, 0x00E6, // LATIN CAPITAL LETTER AE
0x00C7, 0x00E7, // LATIN CAPITAL LETTER C WITH CEDILLA
0x00C8, 0x00E8, // LATIN CAPITAL LETTER E WITH GRAVE
0x00C9, 0x00E9, // LATIN CAPITAL LETTER E WITH ACUTE
0x00CA, 0x00EA, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX
0x00CB, 0x00EB, // LATIN CAPITAL LETTER E WITH DIAERESIS
0x00CC, 0x00EC, // LATIN CAPITAL LETTER I WITH GRAVE
0x00CD, 0x00ED, // LATIN CAPITAL LETTER I WITH ACUTE
0x00CE, 0x00EE, // LATIN CAPITAL LETTER I WITH CIRCUMFLEX
0x00CF, 0x00EF, // LATIN CAPITAL LETTER I WITH DIAERESIS
0x00D0, 0x00F0, // LATIN CAPITAL LETTER ETH
0x00D1, 0x00F1, // LATIN CAPITAL LETTER N WITH TILDE
0x00D2, 0x00F2, // LATIN CAPITAL LETTER O WITH GRAVE
0x00D3, 0x00F3, // LATIN CAPITAL LETTER O WITH ACUTE
0x00D4, 0x00F4, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX
0x00D5, 0x00F5, // LATIN CAPITAL LETTER O WITH TILDE
0x00D6, 0x00F6, // LATIN CAPITAL LETTER O WITH DIAERESIS
0x00D8, 0x00F8, // LATIN CAPITAL LETTER O WITH STROKE
0x00D9, 0x00F9, // LATIN CAPITAL LETTER U WITH GRAVE
0x00DA, 0x00FA, // LATIN CAPITAL LETTER U WITH ACUTE
0x00DB, 0x00FB, // LATIN CAPITAL LETTER U WITH CIRCUMFLEX
0x00DC, 0x00FC, // LATIN CAPITAL LETTER U WITH DIAERESIS
0x00DD, 0x00FD, // LATIN CAPITAL LETTER Y WITH ACUTE
0x00DE, 0x00FE, // LATIN CAPITAL LETTER THORN
0x0100, 0x0101, // LATIN CAPITAL LETTER A WITH MACRON
0x0102, 0x0103, // LATIN CAPITAL LETTER A WITH BREVE
0x0104, 0x0105, // LATIN CAPITAL LETTER A WITH OGONEK
0x0106, 0x0107, // LATIN CAPITAL LETTER C WITH ACUTE
0x0108, 0x0109, // LATIN CAPITAL LETTER C WITH CIRCUMFLEX
0x010A, 0x010B, // LATIN CAPITAL LETTER C WITH DOT ABOVE
0x010C, 0x010D, // LATIN CAPITAL LETTER C WITH CARON
0x010E, 0x010F, // LATIN CAPITAL LETTER D WITH CARON
0x0110, 0x0111, // LATIN CAPITAL LETTER D WITH STROKE
0x0112, 0x0113, // LATIN CAPITAL LETTER E WITH MACRON
0x0114, 0x0115, // LATIN CAPITAL LETTER E WITH BREVE
0x0116, 0x0117, // LATIN CAPITAL LETTER E WITH DOT ABOVE
0x0118, 0x0119, // LATIN CAPITAL LETTER E WITH OGONEK
0x011A, 0x011B, // LATIN CAPITAL LETTER E WITH CARON
0x011C, 0x011D, // LATIN CAPITAL LETTER G WITH CIRCUMFLEX
0x011E, 0x011F, // LATIN CAPITAL LETTER G WITH BREVE
0x0120, 0x0121, // LATIN CAPITAL LETTER G WITH DOT ABOVE
0x0122, 0x0123, // LATIN CAPITAL LETTER G WITH CEDILLA
0x0124, 0x0125, // LATIN CAPITAL LETTER H WITH CIRCUMFLEX
0x0126, 0x0127, // LATIN CAPITAL LETTER H WITH STROKE
0x0128, 0x0129, // LATIN CAPITAL LETTER I WITH TILDE
0x012A, 0x012B, // LATIN CAPITAL LETTER I WITH MACRON
0x012C, 0x012D, // LATIN CAPITAL LETTER I WITH BREVE
0x012E, 0x012F, // LATIN CAPITAL LETTER I WITH OGONEK
0x0132, 0x0133, // LATIN CAPITAL LIGATURE IJ
0x0134, 0x0135, // LATIN CAPITAL LETTER J WITH CIRCUMFLEX
0x0136, 0x0137, // LATIN CAPITAL LETTER K WITH CEDILLA
0x0139, 0x013A, // LATIN CAPITAL LETTER L WITH ACUTE
0x013B, 0x013C, // LATIN CAPITAL LETTER L WITH CEDILLA
0x013D, 0x013E, // LATIN CAPITAL LETTER L WITH CARON
0x013F, 0x0140, // LATIN CAPITAL LETTER L WITH MIDDLE DOT
0x0141, 0x0142, // LATIN CAPITAL LETTER L WITH STROKE
0x0143, 0x0144, // LATIN CAPITAL LETTER N WITH ACUTE
0x0145, 0x0146, // LATIN CAPITAL LETTER N WITH CEDILLA
0x0147, 0x0148, // LATIN CAPITAL LETTER N WITH CARON
0x014A, 0x014B, // LATIN CAPITAL LETTER ENG
0x014C, 0x014D, // LATIN CAPITAL LETTER O WITH MACRON
0x014E, 0x014F, // LATIN CAPITAL LETTER O WITH BREVE
0x0150, 0x0151, // LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
0x0152, 0x0153, // LATIN CAPITAL LIGATURE OE
0x0154, 0x0155, // LATIN CAPITAL LETTER R WITH ACUTE
0x0156, 0x0157, // LATIN CAPITAL LETTER R WITH CEDILLA
0x0158, 0x0159, // LATIN CAPITAL LETTER R WITH CARON
0x015A, 0x015B, // LATIN CAPITAL LETTER S WITH ACUTE
0x015C, 0x015D, // LATIN CAPITAL LETTER S WITH CIRCUMFLEX
0x015E, 0x015F, // LATIN CAPITAL LETTER S WITH CEDILLA
0x0160, 0x0161, // LATIN CAPITAL LETTER S WITH CARON
0x0162, 0x0163, // LATIN CAPITAL LETTER T WITH CEDILLA
0x0164, 0x0165, // LATIN CAPITAL LETTER T WITH CARON
0x0166, 0x0167, // LATIN CAPITAL LETTER T WITH STROKE
0x0168, 0x0169, // LATIN CAPITAL LETTER U WITH TILDE
0x016A, 0x016B, // LATIN CAPITAL LETTER U WITH MACRON
0x016C, 0x016D, // LATIN CAPITAL LETTER U WITH BREVE
0x016E, 0x016F, // LATIN CAPITAL LETTER U WITH RING ABOVE
0x0170, 0x0171, // LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
0x0172, 0x0173, // LATIN CAPITAL LETTER U WITH OGONEK
0x0174, 0x0175, // LATIN CAPITAL LETTER W WITH CIRCUMFLEX
0x0176, 0x0177, // LATIN CAPITAL LETTER Y WITH CIRCUMFLEX
0x0178, 0x00FF, // LATIN CAPITAL LETTER Y WITH DIAERESIS
0x0179, 0x017A, // LATIN CAPITAL LETTER Z WITH ACUTE
0x017B, 0x017C, // LATIN CAPITAL LETTER Z WITH DOT ABOVE
0x017D, 0x017E, // LATIN CAPITAL LETTER Z WITH CARON
0x017F, 0x0073, // LATIN SMALL LETTER LONG S
0x0181, 0x0253, // LATIN CAPITAL LETTER B WITH HOOK
0x0182, 0x0183, // LATIN CAPITAL LETTER B WITH TOPBAR
0x0184, 0x0185, // LATIN CAPITAL LETTER TONE SIX
0x0186, 0x0254, // LATIN CAPITAL LETTER OPEN O
0x0187, 0x0188, // LATIN CAPITAL LETTER C WITH HOOK
0x0189, 0x0256, // LATIN CAPITAL LETTER AFRICAN D
0x018A, 0x0257, // LATIN CAPITAL LETTER D WITH HOOK
0x018B, 0x018C, // LATIN CAPITAL LETTER D WITH TOPBAR
0x018E, 0x01DD, // LATIN CAPITAL LETTER REVERSED E
0x018F, 0x0259, // LATIN CAPITAL LETTER SCHWA
0x0190, 0x025B, // LATIN CAPITAL LETTER OPEN E
0x0191, 0x0192, // LATIN CAPITAL LETTER F WITH HOOK
0x0193, 0x0260, // LATIN CAPITAL LETTER G WITH HOOK
0x0194, 0x0263, // LATIN CAPITAL LETTER GAMMA
0x0196, 0x0269, // LATIN CAPITAL LETTER IOTA
0x0197, 0x0268, // LATIN CAPITAL LETTER I WITH STROKE
0x0198, 0x0199, // LATIN CAPITAL LETTER K WITH HOOK
0x019C, 0x026F, // LATIN CAPITAL LETTER TURNED M
0x019D, 0x0272, // LATIN CAPITAL LETTER N WITH LEFT HOOK
0x019F, 0x0275, // LATIN CAPITAL LETTER O WITH MIDDLE TILDE
0x01A0, 0x01A1, // LATIN CAPITAL LETTER O WITH HORN
0x01A2, 0x01A3, // LATIN CAPITAL LETTER OI
0x01A4, 0x01A5, // LATIN CAPITAL LETTER P WITH HOOK
0x01A6, 0x0280, // LATIN LETTER YR
0x01A7, 0x01A8, // LATIN CAPITAL LETTER TONE TWO
0x01A9, 0x0283, // LATIN CAPITAL LETTER ESH
0x01AC, 0x01AD, // LATIN CAPITAL LETTER T WITH HOOK
0x01AE, 0x0288, // LATIN CAPITAL LETTER T WITH RETROFLEX HOOK
0x01AF, 0x01B0, // LATIN CAPITAL LETTER U WITH HORN
0x01B1, 0x028A, // LATIN CAPITAL LETTER UPSILON
0x01B2, 0x028B, // LATIN CAPITAL LETTER V WITH HOOK
0x01B3, 0x01B4, // LATIN CAPITAL LETTER Y WITH HOOK
0x01B5, 0x01B6, // LATIN CAPITAL LETTER Z WITH STROKE
0x01B7, 0x0292, // LATIN CAPITAL LETTER EZH
0x01B8, 0x01B9, // LATIN CAPITAL LETTER EZH REVERSED
0x01BC, 0x01BD, // LATIN CAPITAL LETTER TONE FIVE
0x01C4, 0x01C6, // LATIN CAPITAL LETTER DZ WITH CARON
0x01C5, 0x01C6, // LATIN CAPITAL LETTER D WITH SMALL LETTER Z WITH CARON
0x01C7, 0x01C9, // LATIN CAPITAL LETTER LJ
0x01C8, 0x01C9, // LATIN CAPITAL LETTER L WITH SMALL LETTER J
0x01CA, 0x01CC, // LATIN CAPITAL LETTER NJ
0x01CB, 0x01CC, // LATIN CAPITAL LETTER N WITH SMALL LETTER J
0x01CD, 0x01CE, // LATIN CAPITAL LETTER A WITH CARON
0x01CF, 0x01D0, // LATIN CAPITAL LETTER I WITH CARON
0x01D1, 0x01D2, // LATIN CAPITAL LETTER O WITH CARON
0x01D3, 0x01D4, // LATIN CAPITAL LETTER U WITH CARON
0x01D5, 0x01D6, // LATIN CAPITAL LETTER U WITH DIAERESIS AND MACRON
0x01D7, 0x01D8, // LATIN CAPITAL LETTER U WITH DIAERESIS AND ACUTE
0x01D9, 0x01DA, // LATIN CAPITAL LETTER U WITH DIAERESIS AND CARON
0x01DB, 0x01DC, // LATIN CAPITAL LETTER U WITH DIAERESIS AND GRAVE
0x01DE, 0x01DF, // LATIN CAPITAL LETTER A WITH DIAERESIS AND MACRON
0x01E0, 0x01E1, // LATIN CAPITAL LETTER A WITH DOT ABOVE AND MACRON
0x01E2, 0x01E3, // LATIN CAPITAL LETTER AE WITH MACRON
0x01E4, 0x01E5, // LATIN CAPITAL LETTER G WITH STROKE
0x01E6, 0x01E7, // LATIN CAPITAL LETTER G WITH CARON
0x01E8, 0x01E9, // LATIN CAPITAL LETTER K WITH CARON
0x01EA, 0x01EB, // LATIN CAPITAL LETTER O WITH OGONEK
0x01EC, 0x01ED, // LATIN CAPITAL LETTER O WITH OGONEK AND MACRON
0x01EE, 0x01EF, // LATIN CAPITAL LETTER EZH WITH CARON
0x01F1, 0x01F3, // LATIN CAPITAL LETTER DZ
0x01F2, 0x01F3, // LATIN CAPITAL LETTER D WITH SMALL LETTER Z
0x01F4, 0x01F5, // LATIN CAPITAL LETTER G WITH ACUTE
0x01F6, 0x0195, // LATIN CAPITAL LETTER HWAIR
0x01F7, 0x01BF, // LATIN CAPITAL LETTER WYNN
0x01F8, 0x01F9, // LATIN CAPITAL LETTER N WITH GRAVE
0x01FA, 0x01FB, // LATIN CAPITAL LETTER A WITH RING ABOVE AND ACUTE
0x01FC, 0x01FD, // LATIN CAPITAL LETTER AE WITH ACUTE
0x01FE, 0x01FF, // LATIN CAPITAL LETTER O WITH STROKE AND ACUTE
0x0200, 0x0201, // LATIN CAPITAL LETTER A WITH DOUBLE GRAVE
0x0202, 0x0203, // LATIN CAPITAL LETTER A WITH INVERTED BREVE
0x0204, 0x0205, // LATIN CAPITAL LETTER E WITH DOUBLE GRAVE
0x0206, 0x0207, // LATIN CAPITAL LETTER E WITH INVERTED BREVE
0x0208, 0x0209, // LATIN CAPITAL LETTER I WITH DOUBLE GRAVE
0x020A, 0x020B, // LATIN CAPITAL LETTER I WITH INVERTED BREVE
0x020C, 0x020D, // LATIN CAPITAL LETTER O WITH DOUBLE GRAVE
0x020E, 0x020F, // LATIN CAPITAL LETTER O WITH INVERTED BREVE
0x0210, 0x0211, // LATIN CAPITAL LETTER R WITH DOUBLE GRAVE
0x0212, 0x0213, // LATIN CAPITAL LETTER R WITH INVERTED BREVE
0x0214, 0x0215, // LATIN CAPITAL LETTER U WITH DOUBLE GRAVE
0x0216, 0x0217, // LATIN CAPITAL LETTER U WITH INVERTED BREVE
0x0218, 0x0219, // LATIN CAPITAL LETTER S WITH COMMA BELOW
0x021A, 0x021B, // LATIN CAPITAL LETTER T WITH COMMA BELOW
0x021C, 0x021D, // LATIN CAPITAL LETTER YOGH
0x021E, 0x021F, // LATIN CAPITAL LETTER H WITH CARON
0x0220, 0x019E, // LATIN CAPITAL LETTER N WITH LONG RIGHT LEG
0x0222, 0x0223, // LATIN CAPITAL LETTER OU
0x0224, 0x0225, // LATIN CAPITAL LETTER Z WITH HOOK
0x0226, 0x0227, // LATIN CAPITAL LETTER A WITH DOT ABOVE
0x0228, 0x0229, // LATIN CAPITAL LETTER E WITH CEDILLA
0x022A, 0x022B, // LATIN CAPITAL LETTER O WITH DIAERESIS AND MACRON
0x022C, 0x022D, // LATIN CAPITAL LETTER O WITH TILDE AND MACRON
0x022E, 0x022F, // LATIN CAPITAL LETTER O WITH DOT ABOVE
0x0230, 0x0231, // LATIN CAPITAL LETTER O WITH DOT ABOVE AND MACRON
0x0232, 0x0233, // LATIN CAPITAL LETTER Y WITH MACRON
0x023A, 0x2C65, // LATIN CAPITAL LETTER A WITH STROKE
0x023B, 0x023C, // LATIN CAPITAL LETTER C WITH STROKE
0x023D, 0x019A, // LATIN CAPITAL LETTER L WITH BAR
0x023E, 0x2C66, // LATIN CAPITAL LETTER T WITH DIAGONAL STROKE
0x0241, 0x0242, // LATIN CAPITAL LETTER GLOTTAL STOP
0x0243, 0x0180, // LATIN CAPITAL LETTER B WITH STROKE
0x0244, 0x0289, // LATIN CAPITAL LETTER U BAR
0x0245, 0x028C, // LATIN CAPITAL LETTER TURNED V
0x0246, 0x0247, // LATIN CAPITAL LETTER E WITH STROKE
0x0248, 0x0249, // LATIN CAPITAL LETTER J WITH STROKE
0x024A, 0x024B, // LATIN CAPITAL LETTER SMALL Q WITH HOOK TAIL
0x024C, 0x024D, // LATIN CAPITAL LETTER R WITH STROKE
0x024E, 0x024F, // LATIN CAPITAL LETTER Y WITH STROKE
0x0345, 0x03B9, // COMBINING GREEK YPOGEGRAMMENI
0x0370, 0x0371, // GREEK CAPITAL LETTER HETA
0x0372, 0x0373, // GREEK CAPITAL LETTER ARCHAIC SAMPI
0x0376, 0x0377, // GREEK CAPITAL LETTER PAMPHYLIAN DIGAMMA
0x0386, 0x03AC, // GREEK CAPITAL LETTER ALPHA WITH TONOS
0x0388, 0x03AD, // GREEK CAPITAL LETTER EPSILON WITH TONOS
0x0389, 0x03AE, // GREEK CAPITAL LETTER ETA WITH TONOS
0x038A, 0x03AF, // GREEK CAPITAL LETTER IOTA WITH TONOS
0x038C, 0x03CC, // GREEK CAPITAL LETTER OMICRON WITH TONOS
0x038E, 0x03CD, // GREEK CAPITAL LETTER UPSILON WITH TONOS
0x038F, 0x03CE, // GREEK CAPITAL LETTER OMEGA WITH TONOS
0x0391, 0x03B1, // GREEK CAPITAL LETTER ALPHA
0x0392, 0x03B2, // GREEK CAPITAL LETTER BETA
0x0393, 0x03B3, // GREEK CAPITAL LETTER GAMMA
0x0394, 0x03B4, // GREEK CAPITAL LETTER DELTA
0x0395, 0x03B5, // GREEK CAPITAL LETTER EPSILON
0x0396, 0x03B6, // GREEK CAPITAL LETTER ZETA
0x0397, 0x03B7, // GREEK CAPITAL LETTER ETA
0x0398, 0x03B8, // GREEK CAPITAL LETTER THETA
0x0399, 0x03B9, // GREEK CAPITAL LETTER IOTA
0x039A, 0x03BA, // GREEK CAPITAL LETTER KAPPA
0x039B, 0x03BB, // GREEK CAPITAL LETTER LAMDA
0x039C, 0x03BC, // GREEK CAPITAL LETTER MU
0x039D, 0x03BD, // GREEK CAPITAL LETTER NU
0x039E, 0x03BE, // GREEK CAPITAL LETTER XI
0x039F, 0x03BF, // GREEK CAPITAL LETTER OMICRON
0x03A0, 0x03C0, // GREEK CAPITAL LETTER PI
0x03A1, 0x03C1, // GREEK CAPITAL LETTER RHO
0x03A3, 0x03C3, // GREEK CAPITAL LETTER SIGMA
0x03A4, 0x03C4, // GREEK CAPITAL LETTER TAU
0x03A5, 0x03C5, // GREEK CAPITAL LETTER UPSILON
0x03A6, 0x03C6, // GREEK CAPITAL LETTER PHI
0x03A7, 0x03C7, // GREEK CAPITAL LETTER CHI
0x03A8, 0x03C8, // GREEK CAPITAL LETTER PSI
0x03A9, 0x03C9, // GREEK CAPITAL LETTER OMEGA
0x03AA, 0x03CA, // GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
0x03AB, 0x03CB, // GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
0x03C2, 0x03C3, // GREEK SMALL LETTER FINAL SIGMA
0x03CF, 0x03D7, // GREEK CAPITAL KAI SYMBOL
0x03D0, 0x03B2, // GREEK BETA SYMBOL
0x03D1, 0x03B8, // GREEK THETA SYMBOL
0x03D5, 0x03C6, // GREEK PHI SYMBOL
0x03D6, 0x03C0, // GREEK PI SYMBOL
0x03D8, 0x03D9, // GREEK LETTER ARCHAIC KOPPA
0x03DA, 0x03DB, // GREEK LETTER STIGMA
0x03DC, 0x03DD, // GREEK LETTER DIGAMMA
0x03DE, 0x03DF, // GREEK LETTER KOPPA
0x03E0, 0x03E1, // GREEK LETTER SAMPI
0x03E2, 0x03E3, // COPTIC CAPITAL LETTER SHEI
0x03E4, 0x03E5, // COPTIC CAPITAL LETTER FEI
0x03E6, 0x03E7, // COPTIC CAPITAL LETTER KHEI
0x03E8, 0x03E9, // COPTIC CAPITAL LETTER HORI
0x03EA, 0x03EB, // COPTIC CAPITAL LETTER GANGIA
0x03EC, 0x03ED, // COPTIC CAPITAL LETTER SHIMA
0x03EE, 0x03EF, // COPTIC CAPITAL LETTER DEI
0x03F0, 0x03BA, // GREEK KAPPA SYMBOL
0x03F1, 0x03C1, // GREEK RHO SYMBOL
0x03F4, 0x03B8, // GREEK CAPITAL THETA SYMBOL
0x03F5, 0x03B5, // GREEK LUNATE EPSILON SYMBOL
0x03F7, 0x03F8, // GREEK CAPITAL LETTER SHO
0x03F9, 0x03F2, // GREEK CAPITAL LUNATE SIGMA SYMBOL
0x03FA, 0x03FB, // GREEK CAPITAL LETTER SAN
0x03FD, 0x037B, // GREEK CAPITAL REVERSED LUNATE SIGMA SYMBOL
0x03FE, 0x037C, // GREEK CAPITAL DOTTED LUNATE SIGMA SYMBOL
0x03FF, 0x037D, // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL
0x0400, 0x0450, // CYRILLIC CAPITAL LETTER IE WITH GRAVE
0x0401, 0x0451, // CYRILLIC CAPITAL LETTER IO
0x0402, 0x0452, // CYRILLIC CAPITAL LETTER DJE
0x0403, 0x0453, // CYRILLIC CAPITAL LETTER GJE
0x0404, 0x0454, // CYRILLIC CAPITAL LETTER UKRAINIAN IE
0x0405, 0x0455, // CYRILLIC CAPITAL LETTER DZE
0x0406, 0x0456, // CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
0x0407, 0x0457, // CYRILLIC CAPITAL LETTER YI
0x0408, 0x0458, // CYRILLIC CAPITAL LETTER JE
0x0409, 0x0459, // CYRILLIC CAPITAL LETTER LJE
0x040A, 0x045A, // CYRILLIC CAPITAL LETTER NJE
0x040B, 0x045B, // CYRILLIC CAPITAL LETTER TSHE
0x040C, 0x045C, // CYRILLIC CAPITAL LETTER KJE
0x040D, 0x045D, // CYRILLIC CAPITAL LETTER I WITH GRAVE
0x040E, 0x045E, // CYRILLIC CAPITAL LETTER SHORT U
0x040F, 0x045F, // CYRILLIC CAPITAL LETTER DZHE
0x0410, 0x0430, // CYRILLIC CAPITAL LETTER A
0x0411, 0x0431, // CYRILLIC CAPITAL LETTER BE
0x0412, 0x0432, // CYRILLIC CAPITAL LETTER VE
0x0413, 0x0433, // CYRILLIC CAPITAL LETTER GHE
0x0414, 0x0434, // CYRILLIC CAPITAL LETTER DE
0x0415, 0x0435, // CYRILLIC CAPITAL LETTER IE
0x0416, 0x0436, // CYRILLIC CAPITAL LETTER ZHE
0x0417, 0x0437, // CYRILLIC CAPITAL LETTER ZE
0x0418, 0x0438, // CYRILLIC CAPITAL LETTER I
0x0419, 0x0439, // CYRILLIC CAPITAL LETTER SHORT I
0x041A, 0x043A, // CYRILLIC CAPITAL LETTER KA
0x041B, 0x043B, // CYRILLIC CAPITAL LETTER EL
0x041C, 0x043C, // CYRILLIC CAPITAL LETTER EM
0x041D, 0x043D, // CYRILLIC CAPITAL LETTER EN
0x041E, 0x043E, // CYRILLIC CAPITAL LETTER O
0x041F, 0x043F, // CYRILLIC CAPITAL LETTER PE
0x0420, 0x0440, // CYRILLIC CAPITAL LETTER ER
0x0421, 0x0441, // CYRILLIC CAPITAL LETTER ES
0x0422, 0x0442, // CYRILLIC CAPITAL LETTER TE
0x0423, 0x0443, // CYRILLIC CAPITAL LETTER U
0x0424, 0x0444, // CYRILLIC CAPITAL LETTER EF
0x0425, 0x0445, // CYRILLIC CAPITAL LETTER HA
0x0426, 0x0446, // CYRILLIC CAPITAL LETTER TSE
0x0427, 0x0447, // CYRILLIC CAPITAL LETTER CHE
0x0428, 0x0448, // CYRILLIC CAPITAL LETTER SHA
0x0429, 0x0449, // CYRILLIC CAPITAL LETTER SHCHA
0x042A, 0x044A, // CYRILLIC CAPITAL LETTER HARD SIGN
0x042B, 0x044B, // CYRILLIC CAPITAL LETTER YERU
0x042C, 0x044C, // CYRILLIC CAPITAL LETTER SOFT SIGN
0x042D, 0x044D, // CYRILLIC CAPITAL LETTER E
0x042E, 0x044E, // CYRILLIC CAPITAL LETTER YU
0x042F, 0x044F, // CYRILLIC CAPITAL LETTER YA
0x0460, 0x0461, // CYRILLIC CAPITAL LETTER OMEGA
0x0462, 0x0463, // CYRILLIC CAPITAL LETTER YAT
0x0464, 0x0465, // CYRILLIC CAPITAL LETTER IOTIFIED E
0x0466, 0x0467, // CYRILLIC CAPITAL LETTER LITTLE YUS
0x0468, 0x0469, // CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS
0x046A, 0x046B, // CYRILLIC CAPITAL LETTER BIG YUS
0x046C, 0x046D, // CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS
0x046E, 0x046F, // CYRILLIC CAPITAL LETTER KSI
0x0470, 0x0471, // CYRILLIC CAPITAL LETTER PSI
0x0472, 0x0473, // CYRILLIC CAPITAL LETTER FITA
0x0474, 0x0475, // CYRILLIC CAPITAL LETTER IZHITSA
0x0476, 0x0477, // CYRILLIC CAPITAL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT
0x0478, 0x0479, // CYRILLIC CAPITAL LETTER UK
0x047A, 0x047B, // CYRILLIC CAPITAL LETTER ROUND OMEGA
0x047C, 0x047D, // CYRILLIC CAPITAL LETTER OMEGA WITH TITLO
0x047E, 0x047F, // CYRILLIC CAPITAL LETTER OT
0x0480, 0x0481, // CYRILLIC CAPITAL LETTER KOPPA
0x048A, 0x048B, // CYRILLIC CAPITAL LETTER SHORT I WITH TAIL
0x048C, 0x048D, // CYRILLIC CAPITAL LETTER SEMISOFT SIGN
0x048E, 0x048F, // CYRILLIC CAPITAL LETTER ER WITH TICK
0x0490, 0x0491, // CYRILLIC CAPITAL LETTER GHE WITH UPTURN
0x0492, 0x0493, // CYRILLIC CAPITAL LETTER GHE WITH STROKE
0x0494, 0x0495, // CYRILLIC CAPITAL LETTER GHE WITH MIDDLE HOOK
0x0496, 0x0497, // CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER
0x0498, 0x0499, // CYRILLIC CAPITAL LETTER ZE WITH DESCENDER
0x049A, 0x049B, // CYRILLIC CAPITAL LETTER KA WITH DESCENDER
0x049C, 0x049D, // CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE
0x049E, 0x049F, // CYRILLIC CAPITAL LETTER KA WITH STROKE
0x04A0, 0x04A1, // CYRILLIC CAPITAL LETTER BASHKIR KA
0x04A2, 0x04A3, // CYRILLIC CAPITAL LETTER EN WITH DESCENDER
0x04A4, 0x04A5, // CYRILLIC CAPITAL LIGATURE EN GHE
0x04A6, 0x04A7, // CYRILLIC CAPITAL LETTER PE WITH MIDDLE HOOK
0x04A8, 0x04A9, // CYRILLIC CAPITAL LETTER ABKHASIAN HA
0x04AA, 0x04AB, // CYRILLIC CAPITAL LETTER ES WITH DESCENDER
0x04AC, 0x04AD, // CYRILLIC CAPITAL LETTER TE WITH DESCENDER
0x04AE, 0x04AF, // CYRILLIC CAPITAL LETTER STRAIGHT U
0x04B0, 0x04B1, // CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE
0x04B2, 0x04B3, // CYRILLIC CAPITAL LETTER HA WITH DESCENDER
0x04B4, 0x04B5, // CYRILLIC CAPITAL LIGATURE TE TSE
0x04B6, 0x04B7, // CYRILLIC CAPITAL LETTER CHE WITH DESCENDER
0x04B8, 0x04B9, // CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE
0x04BA, 0x04BB, // CYRILLIC CAPITAL LETTER SHHA
0x04BC, 0x04BD, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE
0x04BE, 0x04BF, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER
0x04C0, 0x04CF, // CYRILLIC LETTER PALOCHKA
0x04C1, 0x04C2, // CYRILLIC CAPITAL LETTER ZHE WITH BREVE
0x04C3, 0x04C4, // CYRILLIC CAPITAL LETTER KA WITH HOOK
0x04C5, 0x04C6, // CYRILLIC CAPITAL LETTER EL WITH TAIL
0x04C7, 0x04C8, // CYRILLIC CAPITAL LETTER EN WITH HOOK
0x04C9, 0x04CA, // CYRILLIC CAPITAL LETTER EN WITH TAIL
0x04CB, 0x04CC, // CYRILLIC CAPITAL LETTER KHAKASSIAN CHE
0x04CD, 0x04CE, // CYRILLIC CAPITAL LETTER EM WITH TAIL
0x04D0, 0x04D1, // CYRILLIC CAPITAL LETTER A WITH BREVE
0x04D2, 0x04D3, // CYRILLIC CAPITAL LETTER A WITH DIAERESIS
0x04D4, 0x04D5, // CYRILLIC CAPITAL LIGATURE A IE
0x04D6, 0x04D7, // CYRILLIC CAPITAL LETTER IE WITH BREVE
0x04D8, 0x04D9, // CYRILLIC CAPITAL LETTER SCHWA
0x04DA, 0x04DB, // CYRILLIC CAPITAL LETTER SCHWA WITH DIAERESIS
0x04DC, 0x04DD, // CYRILLIC CAPITAL LETTER ZHE WITH DIAERESIS
0x04DE, 0x04DF, // CYRILLIC CAPITAL LETTER ZE WITH DIAERESIS
0x04E0, 0x04E1, // CYRILLIC CAPITAL LETTER ABKHASIAN DZE
0x04E2, 0x04E3, // CYRILLIC CAPITAL LETTER I WITH MACRON
0x04E4, 0x04E5, // CYRILLIC CAPITAL LETTER I WITH DIAERESIS
0x04E6, 0x04E7, // CYRILLIC CAPITAL LETTER O WITH DIAERESIS
0x04E8, 0x04E9, // CYRILLIC CAPITAL LETTER BARRED O
0x04EA, 0x04EB, // CYRILLIC CAPITAL LETTER BARRED O WITH DIAERESIS
0x04EC, 0x04ED, // CYRILLIC CAPITAL LETTER E WITH DIAERESIS
0x04EE, 0x04EF, // CYRILLIC CAPITAL LETTER U WITH MACRON
0x04F0, 0x04F1, // CYRILLIC CAPITAL LETTER U WITH DIAERESIS
0x04F2, 0x04F3, // CYRILLIC CAPITAL LETTER U WITH DOUBLE ACUTE
0x04F4, 0x04F5, // CYRILLIC CAPITAL LETTER CHE WITH DIAERESIS
0x04F6, 0x04F7, // CYRILLIC CAPITAL LETTER GHE WITH DESCENDER
0x04F8, 0x04F9, // CYRILLIC CAPITAL LETTER YERU WITH DIAERESIS
0x04FA, 0x04FB, // CYRILLIC CAPITAL LETTER GHE WITH STROKE AND HOOK
0x04FC, 0x04FD, // CYRILLIC CAPITAL LETTER HA WITH HOOK
0x04FE, 0x04FF, // CYRILLIC CAPITAL LETTER HA WITH STROKE
0x0500, 0x0501, // CYRILLIC CAPITAL LETTER KOMI DE
0x0502, 0x0503, // CYRILLIC CAPITAL LETTER KOMI DJE
0x0504, 0x0505, // CYRILLIC CAPITAL LETTER KOMI ZJE
0x0506, 0x0507, // CYRILLIC CAPITAL LETTER KOMI DZJE
0x0508, 0x0509, // CYRILLIC CAPITAL LETTER KOMI LJE
0x050A, 0x050B, // CYRILLIC CAPITAL LETTER KOMI NJE
0x050C, 0x050D, // CYRILLIC CAPITAL LETTER KOMI SJE
0x050E, 0x050F, // CYRILLIC CAPITAL LETTER KOMI TJE
0x0510, 0x0511, // CYRILLIC CAPITAL LETTER REVERSED ZE
0x0512, 0x0513, // CYRILLIC CAPITAL LETTER EL WITH HOOK
0x0514, 0x0515, // CYRILLIC CAPITAL LETTER LHA
0x0516, 0x0517, // CYRILLIC CAPITAL LETTER RHA
0x0518, 0x0519, // CYRILLIC CAPITAL LETTER YAE
0x051A, 0x051B, // CYRILLIC CAPITAL LETTER QA
0x051C, 0x051D, // CYRILLIC CAPITAL LETTER WE
0x051E, 0x051F, // CYRILLIC CAPITAL LETTER ALEUT KA
0x0520, 0x0521, // CYRILLIC CAPITAL LETTER EL WITH MIDDLE HOOK
0x0522, 0x0523, // CYRILLIC CAPITAL LETTER EN WITH MIDDLE HOOK
0x0524, 0x0525, // CYRILLIC CAPITAL LETTER PE WITH DESCENDER
0x0531, 0x0561, // ARMENIAN CAPITAL LETTER AYB
0x0532, 0x0562, // ARMENIAN CAPITAL LETTER BEN
0x0533, 0x0563, // ARMENIAN CAPITAL LETTER GIM
0x0534, 0x0564, // ARMENIAN CAPITAL LETTER DA
0x0535, 0x0565, // ARMENIAN CAPITAL LETTER ECH
0x0536, 0x0566, // ARMENIAN CAPITAL LETTER ZA
0x0537, 0x0567, // ARMENIAN CAPITAL LETTER EH
0x0538, 0x0568, // ARMENIAN CAPITAL LETTER ET
0x0539, 0x0569, // ARMENIAN CAPITAL LETTER TO
0x053A, 0x056A, // ARMENIAN CAPITAL LETTER ZHE
0x053B, 0x056B, // ARMENIAN CAPITAL LETTER INI
0x053C, 0x056C, // ARMENIAN CAPITAL LETTER LIWN
0x053D, 0x056D, // ARMENIAN CAPITAL LETTER XEH
0x053E, 0x056E, // ARMENIAN CAPITAL LETTER CA
0x053F, 0x056F, // ARMENIAN CAPITAL LETTER KEN
0x0540, 0x0570, // ARMENIAN CAPITAL LETTER HO
0x0541, 0x0571, // ARMENIAN CAPITAL LETTER JA
0x0542, 0x0572, // ARMENIAN CAPITAL LETTER GHAD
0x0543, 0x0573, // ARMENIAN CAPITAL LETTER CHEH
0x0544, 0x0574, // ARMENIAN CAPITAL LETTER MEN
0x0545, 0x0575, // ARMENIAN CAPITAL LETTER YI
0x0546, 0x0576, // ARMENIAN CAPITAL LETTER NOW
0x0547, 0x0577, // ARMENIAN CAPITAL LETTER SHA
0x0548, 0x0578, // ARMENIAN CAPITAL LETTER VO
0x0549, 0x0579, // ARMENIAN CAPITAL LETTER CHA
0x054A, 0x057A, // ARMENIAN CAPITAL LETTER PEH
0x054B, 0x057B, // ARMENIAN CAPITAL LETTER JHEH
0x054C, 0x057C, // ARMENIAN CAPITAL LETTER RA
0x054D, 0x057D, // ARMENIAN CAPITAL LETTER SEH
0x054E, 0x057E, // ARMENIAN CAPITAL LETTER VEW
0x054F, 0x057F, // ARMENIAN CAPITAL LETTER TIWN
0x0550, 0x0580, // ARMENIAN CAPITAL LETTER REH
0x0551, 0x0581, // ARMENIAN CAPITAL LETTER CO
0x0552, 0x0582, // ARMENIAN CAPITAL LETTER YIWN
0x0553, 0x0583, // ARMENIAN CAPITAL LETTER PIWR
0x0554, 0x0584, // ARMENIAN CAPITAL LETTER KEH
0x0555, 0x0585, // ARMENIAN CAPITAL LETTER OH
0x0556, 0x0586, // ARMENIAN CAPITAL LETTER FEH
0x10A0, 0x2D00, // GEORGIAN CAPITAL LETTER AN
0x10A1, 0x2D01, // GEORGIAN CAPITAL LETTER BAN
0x10A2, 0x2D02, // GEORGIAN CAPITAL LETTER GAN
0x10A3, 0x2D03, // GEORGIAN CAPITAL LETTER DON
0x10A4, 0x2D04, // GEORGIAN CAPITAL LETTER EN
0x10A5, 0x2D05, // GEORGIAN CAPITAL LETTER VIN
0x10A6, 0x2D06, // GEORGIAN CAPITAL LETTER ZEN
0x10A7, 0x2D07, // GEORGIAN CAPITAL LETTER TAN
0x10A8, 0x2D08, // GEORGIAN CAPITAL LETTER IN
0x10A9, 0x2D09, // GEORGIAN CAPITAL LETTER KAN
0x10AA, 0x2D0A, // GEORGIAN CAPITAL LETTER LAS
0x10AB, 0x2D0B, // GEORGIAN CAPITAL LETTER MAN
0x10AC, 0x2D0C, // GEORGIAN CAPITAL LETTER NAR
0x10AD, 0x2D0D, // GEORGIAN CAPITAL LETTER ON
0x10AE, 0x2D0E, // GEORGIAN CAPITAL LETTER PAR
0x10AF, 0x2D0F, // GEORGIAN CAPITAL LETTER ZHAR
0x10B0, 0x2D10, // GEORGIAN CAPITAL LETTER RAE
0x10B1, 0x2D11, // GEORGIAN CAPITAL LETTER SAN
0x10B2, 0x2D12, // GEORGIAN CAPITAL LETTER TAR
0x10B3, 0x2D13, // GEORGIAN CAPITAL LETTER UN
0x10B4, 0x2D14, // GEORGIAN CAPITAL LETTER PHAR
0x10B5, 0x2D15, // GEORGIAN CAPITAL LETTER KHAR
0x10B6, 0x2D16, // GEORGIAN CAPITAL LETTER GHAN
0x10B7, 0x2D17, // GEORGIAN CAPITAL LETTER QAR
0x10B8, 0x2D18, // GEORGIAN CAPITAL LETTER SHIN
0x10B9, 0x2D19, // GEORGIAN CAPITAL LETTER CHIN
0x10BA, 0x2D1A, // GEORGIAN CAPITAL LETTER CAN
0x10BB, 0x2D1B, // GEORGIAN CAPITAL LETTER JIL
0x10BC, 0x2D1C, // GEORGIAN CAPITAL LETTER CIL
0x10BD, 0x2D1D, // GEORGIAN CAPITAL LETTER CHAR
0x10BE, 0x2D1E, // GEORGIAN CAPITAL LETTER XAN
0x10BF, 0x2D1F, // GEORGIAN CAPITAL LETTER JHAN
0x10C0, 0x2D20, // GEORGIAN CAPITAL LETTER HAE
0x10C1, 0x2D21, // GEORGIAN CAPITAL LETTER HE
0x10C2, 0x2D22, // GEORGIAN CAPITAL LETTER HIE
0x10C3, 0x2D23, // GEORGIAN CAPITAL LETTER WE
0x10C4, 0x2D24, // GEORGIAN CAPITAL LETTER HAR
0x10C5, 0x2D25, // GEORGIAN CAPITAL LETTER HOE
0x1E00, 0x1E01, // LATIN CAPITAL LETTER A WITH RING BELOW
0x1E02, 0x1E03, // LATIN CAPITAL LETTER B WITH DOT ABOVE
0x1E04, 0x1E05, // LATIN CAPITAL LETTER B WITH DOT BELOW
0x1E06, 0x1E07, // LATIN CAPITAL LETTER B WITH LINE BELOW
0x1E08, 0x1E09, // LATIN CAPITAL LETTER C WITH CEDILLA AND ACUTE
0x1E0A, 0x1E0B, // LATIN CAPITAL LETTER D WITH DOT ABOVE
0x1E0C, 0x1E0D, // LATIN CAPITAL LETTER D WITH DOT BELOW
0x1E0E, 0x1E0F, // LATIN CAPITAL LETTER D WITH LINE BELOW
0x1E10, 0x1E11, // LATIN CAPITAL LETTER D WITH CEDILLA
0x1E12, 0x1E13, // LATIN CAPITAL LETTER D WITH CIRCUMFLEX BELOW
0x1E14, 0x1E15, // LATIN CAPITAL LETTER E WITH MACRON AND GRAVE
0x1E16, 0x1E17, // LATIN CAPITAL LETTER E WITH MACRON AND ACUTE
0x1E18, 0x1E19, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX BELOW
0x1E1A, 0x1E1B, // LATIN CAPITAL LETTER E WITH TILDE BELOW
0x1E1C, 0x1E1D, // LATIN CAPITAL LETTER E WITH CEDILLA AND BREVE
0x1E1E, 0x1E1F, // LATIN CAPITAL LETTER F WITH DOT ABOVE
0x1E20, 0x1E21, // LATIN CAPITAL LETTER G WITH MACRON
0x1E22, 0x1E23, // LATIN CAPITAL LETTER H WITH DOT ABOVE
0x1E24, 0x1E25, // LATIN CAPITAL LETTER H WITH DOT BELOW
0x1E26, 0x1E27, // LATIN CAPITAL LETTER H WITH DIAERESIS
0x1E28, 0x1E29, // LATIN CAPITAL LETTER H WITH CEDILLA
0x1E2A, 0x1E2B, // LATIN CAPITAL LETTER H WITH BREVE BELOW
0x1E2C, 0x1E2D, // LATIN CAPITAL LETTER I WITH TILDE BELOW
0x1E2E, 0x1E2F, // LATIN CAPITAL LETTER I WITH DIAERESIS AND ACUTE
0x1E30, 0x1E31, // LATIN CAPITAL LETTER K WITH ACUTE
0x1E32, 0x1E33, // LATIN CAPITAL LETTER K WITH DOT BELOW
0x1E34, 0x1E35, // LATIN CAPITAL LETTER K WITH LINE BELOW
0x1E36, 0x1E37, // LATIN CAPITAL LETTER L WITH DOT BELOW
0x1E38, 0x1E39, // LATIN CAPITAL LETTER L WITH DOT BELOW AND MACRON
0x1E3A, 0x1E3B, // LATIN CAPITAL LETTER L WITH LINE BELOW
0x1E3C, 0x1E3D, // LATIN CAPITAL LETTER L WITH CIRCUMFLEX BELOW
0x1E3E, 0x1E3F, // LATIN CAPITAL LETTER M WITH ACUTE
0x1E40, 0x1E41, // LATIN CAPITAL LETTER M WITH DOT ABOVE
0x1E42, 0x1E43, // LATIN CAPITAL LETTER M WITH DOT BELOW
0x1E44, 0x1E45, // LATIN CAPITAL LETTER N WITH DOT ABOVE
0x1E46, 0x1E47, // LATIN CAPITAL LETTER N WITH DOT BELOW
0x1E48, 0x1E49, // LATIN CAPITAL LETTER N WITH LINE BELOW
0x1E4A, 0x1E4B, // LATIN CAPITAL LETTER N WITH CIRCUMFLEX BELOW
0x1E4C, 0x1E4D, // LATIN CAPITAL LETTER O WITH TILDE AND ACUTE
0x1E4E, 0x1E4F, // LATIN CAPITAL LETTER O WITH TILDE AND DIAERESIS
0x1E50, 0x1E51, // LATIN CAPITAL LETTER O WITH MACRON AND GRAVE
0x1E52, 0x1E53, // LATIN CAPITAL LETTER O WITH MACRON AND ACUTE
0x1E54, 0x1E55, // LATIN CAPITAL LETTER P WITH ACUTE
0x1E56, 0x1E57, // LATIN CAPITAL LETTER P WITH DOT ABOVE
0x1E58, 0x1E59, // LATIN CAPITAL LETTER R WITH DOT ABOVE
0x1E5A, 0x1E5B, // LATIN CAPITAL LETTER R WITH DOT BELOW
0x1E5C, 0x1E5D, // LATIN CAPITAL LETTER R WITH DOT BELOW AND MACRON
0x1E5E, 0x1E5F, // LATIN CAPITAL LETTER R WITH LINE BELOW
0x1E60, 0x1E61, // LATIN CAPITAL LETTER S WITH DOT ABOVE
0x1E62, 0x1E63, // LATIN CAPITAL LETTER S WITH DOT BELOW
0x1E64, 0x1E65, // LATIN CAPITAL LETTER S WITH ACUTE AND DOT ABOVE
0x1E66, 0x1E67, // LATIN CAPITAL LETTER S WITH CARON AND DOT ABOVE
0x1E68, 0x1E69, // LATIN CAPITAL LETTER S WITH DOT BELOW AND DOT ABOVE
0x1E6A, 0x1E6B, // LATIN CAPITAL LETTER T WITH DOT ABOVE
0x1E6C, 0x1E6D, // LATIN CAPITAL LETTER T WITH DOT BELOW
0x1E6E, 0x1E6F, // LATIN CAPITAL LETTER T WITH LINE BELOW
0x1E70, 0x1E71, // LATIN CAPITAL LETTER T WITH CIRCUMFLEX BELOW
0x1E72, 0x1E73, // LATIN CAPITAL LETTER U WITH DIAERESIS BELOW
0x1E74, 0x1E75, // LATIN CAPITAL LETTER U WITH TILDE BELOW
0x1E76, 0x1E77, // LATIN CAPITAL LETTER U WITH CIRCUMFLEX BELOW
0x1E78, 0x1E79, // LATIN CAPITAL LETTER U WITH TILDE AND ACUTE
0x1E7A, 0x1E7B, // LATIN CAPITAL LETTER U WITH MACRON AND DIAERESIS
0x1E7C, 0x1E7D, // LATIN CAPITAL LETTER V WITH TILDE
0x1E7E, 0x1E7F, // LATIN CAPITAL LETTER V WITH DOT BELOW
0x1E80, 0x1E81, // LATIN CAPITAL LETTER W WITH GRAVE
0x1E82, 0x1E83, // LATIN CAPITAL LETTER W WITH ACUTE
0x1E84, 0x1E85, // LATIN CAPITAL LETTER W WITH DIAERESIS
0x1E86, 0x1E87, // LATIN CAPITAL LETTER W WITH DOT ABOVE
0x1E88, 0x1E89, // LATIN CAPITAL LETTER W WITH DOT BELOW
0x1E8A, 0x1E8B, // LATIN CAPITAL LETTER X WITH DOT ABOVE
0x1E8C, 0x1E8D, // LATIN CAPITAL LETTER X WITH DIAERESIS
0x1E8E, 0x1E8F, // LATIN CAPITAL LETTER Y WITH DOT ABOVE
0x1E90, 0x1E91, // LATIN CAPITAL LETTER Z WITH CIRCUMFLEX
0x1E92, 0x1E93, // LATIN CAPITAL LETTER Z WITH DOT BELOW
0x1E94, 0x1E95, // LATIN CAPITAL LETTER Z WITH LINE BELOW
0x1E9B, 0x1E61, // LATIN SMALL LETTER LONG S WITH DOT ABOVE
0x1E9E, 0x00DF, // LATIN CAPITAL LETTER SHARP S
0x1EA0, 0x1EA1, // LATIN CAPITAL LETTER A WITH DOT BELOW
0x1EA2, 0x1EA3, // LATIN CAPITAL LETTER A WITH HOOK ABOVE
0x1EA4, 0x1EA5, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND ACUTE
0x1EA6, 0x1EA7, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND GRAVE
0x1EA8, 0x1EA9, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND HOOK ABOVE
0x1EAA, 0x1EAB, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND TILDE
0x1EAC, 0x1EAD, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND DOT BELOW
0x1EAE, 0x1EAF, // LATIN CAPITAL LETTER A WITH BREVE AND ACUTE
0x1EB0, 0x1EB1, // LATIN CAPITAL LETTER A WITH BREVE AND GRAVE
0x1EB2, 0x1EB3, // LATIN CAPITAL LETTER A WITH BREVE AND HOOK ABOVE
0x1EB4, 0x1EB5, // LATIN CAPITAL LETTER A WITH BREVE AND TILDE
0x1EB6, 0x1EB7, // LATIN CAPITAL LETTER A WITH BREVE AND DOT BELOW
0x1EB8, 0x1EB9, // LATIN CAPITAL LETTER E WITH DOT BELOW
0x1EBA, 0x1EBB, // LATIN CAPITAL LETTER E WITH HOOK ABOVE
0x1EBC, 0x1EBD, // LATIN CAPITAL LETTER E WITH TILDE
0x1EBE, 0x1EBF, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND ACUTE
0x1EC0, 0x1EC1, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND GRAVE
0x1EC2, 0x1EC3, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND HOOK ABOVE
0x1EC4, 0x1EC5, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND TILDE
0x1EC6, 0x1EC7, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND DOT BELOW
0x1EC8, 0x1EC9, // LATIN CAPITAL LETTER I WITH HOOK ABOVE
0x1ECA, 0x1ECB, // LATIN CAPITAL LETTER I WITH DOT BELOW
0x1ECC, 0x1ECD, // LATIN CAPITAL LETTER O WITH DOT BELOW
0x1ECE, 0x1ECF, // LATIN CAPITAL LETTER O WITH HOOK ABOVE
0x1ED0, 0x1ED1, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND ACUTE
0x1ED2, 0x1ED3, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND GRAVE
0x1ED4, 0x1ED5, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND HOOK ABOVE
0x1ED6, 0x1ED7, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND TILDE
0x1ED8, 0x1ED9, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND DOT BELOW
0x1EDA, 0x1EDB, // LATIN CAPITAL LETTER O WITH HORN AND ACUTE
0x1EDC, 0x1EDD, // LATIN CAPITAL LETTER O WITH HORN AND GRAVE
0x1EDE, 0x1EDF, // LATIN CAPITAL LETTER O WITH HORN AND HOOK ABOVE
0x1EE0, 0x1EE1, // LATIN CAPITAL LETTER O WITH HORN AND TILDE
0x1EE2, 0x1EE3, // LATIN CAPITAL LETTER O WITH HORN AND DOT BELOW
0x1EE4, 0x1EE5, // LATIN CAPITAL LETTER U WITH DOT BELOW
0x1EE6, 0x1EE7, // LATIN CAPITAL LETTER U WITH HOOK ABOVE
0x1EE8, 0x1EE9, // LATIN CAPITAL LETTER U WITH HORN AND ACUTE
0x1EEA, 0x1EEB, // LATIN CAPITAL LETTER U WITH HORN AND GRAVE
0x1EEC, 0x1EED, // LATIN CAPITAL LETTER U WITH HORN AND HOOK ABOVE
0x1EEE, 0x1EEF, // LATIN CAPITAL LETTER U WITH HORN AND TILDE
0x1EF0, 0x1EF1, // LATIN CAPITAL LETTER U WITH HORN AND DOT BELOW
0x1EF2, 0x1EF3, // LATIN CAPITAL LETTER Y WITH GRAVE
0x1EF4, 0x1EF5, // LATIN CAPITAL LETTER Y WITH DOT BELOW
0x1EF6, 0x1EF7, // LATIN CAPITAL LETTER Y WITH HOOK ABOVE
0x1EF8, 0x1EF9, // LATIN CAPITAL LETTER Y WITH TILDE
0x1EFA, 0x1EFB, // LATIN CAPITAL LETTER MIDDLE-WELSH LL
0x1EFC, 0x1EFD, // LATIN CAPITAL LETTER MIDDLE-WELSH V
0x1EFE, 0x1EFF, // LATIN CAPITAL LETTER Y WITH LOOP
0x1F08, 0x1F00, // GREEK CAPITAL LETTER ALPHA WITH PSILI
0x1F09, 0x1F01, // GREEK CAPITAL LETTER ALPHA WITH DASIA
0x1F0A, 0x1F02, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA
0x1F0B, 0x1F03, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA
0x1F0C, 0x1F04, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA
0x1F0D, 0x1F05, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA
0x1F0E, 0x1F06, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI
0x1F0F, 0x1F07, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI
0x1F18, 0x1F10, // GREEK CAPITAL LETTER EPSILON WITH PSILI
0x1F19, 0x1F11, // GREEK CAPITAL LETTER EPSILON WITH DASIA
0x1F1A, 0x1F12, // GREEK CAPITAL LETTER EPSILON WITH PSILI AND VARIA
0x1F1B, 0x1F13, // GREEK CAPITAL LETTER EPSILON WITH DASIA AND VARIA
0x1F1C, 0x1F14, // GREEK CAPITAL LETTER EPSILON WITH PSILI AND OXIA
0x1F1D, 0x1F15, // GREEK CAPITAL LETTER EPSILON WITH DASIA AND OXIA
0x1F28, 0x1F20, // GREEK CAPITAL LETTER ETA WITH PSILI
0x1F29, 0x1F21, // GREEK CAPITAL LETTER ETA WITH DASIA
0x1F2A, 0x1F22, // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA
0x1F2B, 0x1F23, // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA
0x1F2C, 0x1F24, // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA
0x1F2D, 0x1F25, // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA
0x1F2E, 0x1F26, // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI
0x1F2F, 0x1F27, // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI
0x1F38, 0x1F30, // GREEK CAPITAL LETTER IOTA WITH PSILI
0x1F39, 0x1F31, // GREEK CAPITAL LETTER IOTA WITH DASIA
0x1F3A, 0x1F32, // GREEK CAPITAL LETTER IOTA WITH PSILI AND VARIA
0x1F3B, 0x1F33, // GREEK CAPITAL LETTER IOTA WITH DASIA AND VARIA
0x1F3C, 0x1F34, // GREEK CAPITAL LETTER IOTA WITH PSILI AND OXIA
0x1F3D, 0x1F35, // GREEK CAPITAL LETTER IOTA WITH DASIA AND OXIA
0x1F3E, 0x1F36, // GREEK CAPITAL LETTER IOTA WITH PSILI AND PERISPOMENI
0x1F3F, 0x1F37, // GREEK CAPITAL LETTER IOTA WITH DASIA AND PERISPOMENI
0x1F48, 0x1F40, // GREEK CAPITAL LETTER OMICRON WITH PSILI
0x1F49, 0x1F41, // GREEK CAPITAL LETTER OMICRON WITH DASIA
0x1F4A, 0x1F42, // GREEK CAPITAL LETTER OMICRON WITH PSILI AND VARIA
0x1F4B, 0x1F43, // GREEK CAPITAL LETTER OMICRON WITH DASIA AND VARIA
0x1F4C, 0x1F44, // GREEK CAPITAL LETTER OMICRON WITH PSILI AND OXIA
0x1F4D, 0x1F45, // GREEK CAPITAL LETTER OMICRON WITH DASIA AND OXIA
0x1F59, 0x1F51, // GREEK CAPITAL LETTER UPSILON WITH DASIA
0x1F5B, 0x1F53, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND VARIA
0x1F5D, 0x1F55, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND OXIA
0x1F5F, 0x1F57, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND PERISPOMENI
0x1F68, 0x1F60, // GREEK CAPITAL LETTER OMEGA WITH PSILI
0x1F69, 0x1F61, // GREEK CAPITAL LETTER OMEGA WITH DASIA
0x1F6A, 0x1F62, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA
0x1F6B, 0x1F63, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA
0x1F6C, 0x1F64, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA
0x1F6D, 0x1F65, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA
0x1F6E, 0x1F66, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI
0x1F6F, 0x1F67, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI
0x1F88, 0x1F80, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PROSGEGRAMMENI
0x1F89, 0x1F81, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PROSGEGRAMMENI
0x1F8A, 0x1F82, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA AND PROSGEGRAMMENI
0x1F8B, 0x1F83, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA AND PROSGEGRAMMENI
0x1F8C, 0x1F84, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA AND PROSGEGRAMMENI
0x1F8D, 0x1F85, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA AND PROSGEGRAMMENI
0x1F8E, 0x1F86, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
0x1F8F, 0x1F87, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
0x1F98, 0x1F90, // GREEK CAPITAL LETTER ETA WITH PSILI AND PROSGEGRAMMENI
0x1F99, 0x1F91, // GREEK CAPITAL LETTER ETA WITH DASIA AND PROSGEGRAMMENI
0x1F9A, 0x1F92, // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA AND PROSGEGRAMMENI
0x1F9B, 0x1F93, // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA AND PROSGEGRAMMENI
0x1F9C, 0x1F94, // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA AND PROSGEGRAMMENI
0x1F9D, 0x1F95, // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA AND PROSGEGRAMMENI
0x1F9E, 0x1F96, // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
0x1F9F, 0x1F97, // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
0x1FA8, 0x1FA0, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PROSGEGRAMMENI
0x1FA9, 0x1FA1, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PROSGEGRAMMENI
0x1FAA, 0x1FA2, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA AND PROSGEGRAMMENI
0x1FAB, 0x1FA3, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA AND PROSGEGRAMMENI
0x1FAC, 0x1FA4, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA AND PROSGEGRAMMENI
0x1FAD, 0x1FA5, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA AND PROSGEGRAMMENI
0x1FAE, 0x1FA6, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
0x1FAF, 0x1FA7, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
0x1FB8, 0x1FB0, // GREEK CAPITAL LETTER ALPHA WITH VRACHY
0x1FB9, 0x1FB1, // GREEK CAPITAL LETTER ALPHA WITH MACRON
0x1FBA, 0x1F70, // GREEK CAPITAL LETTER ALPHA WITH VARIA
0x1FBB, 0x1F71, // GREEK CAPITAL LETTER ALPHA WITH OXIA
0x1FBC, 0x1FB3, // GREEK CAPITAL LETTER ALPHA WITH PROSGEGRAMMENI
0x1FBE, 0x03B9, // GREEK PROSGEGRAMMENI
0x1FC8, 0x1F72, // GREEK CAPITAL LETTER EPSILON WITH VARIA
0x1FC9, 0x1F73, // GREEK CAPITAL LETTER EPSILON WITH OXIA
0x1FCA, 0x1F74, // GREEK CAPITAL LETTER ETA WITH VARIA
0x1FCB, 0x1F75, // GREEK CAPITAL LETTER ETA WITH OXIA
0x1FCC, 0x1FC3, // GREEK CAPITAL LETTER ETA WITH PROSGEGRAMMENI
0x1FD8, 0x1FD0, // GREEK CAPITAL LETTER IOTA WITH VRACHY
0x1FD9, 0x1FD1, // GREEK CAPITAL LETTER IOTA WITH MACRON
0x1FDA, 0x1F76, // GREEK CAPITAL LETTER IOTA WITH VARIA
0x1FDB, 0x1F77, // GREEK CAPITAL LETTER IOTA WITH OXIA
0x1FE8, 0x1FE0, // GREEK CAPITAL LETTER UPSILON WITH VRACHY
0x1FE9, 0x1FE1, // GREEK CAPITAL LETTER UPSILON WITH MACRON
0x1FEA, 0x1F7A, // GREEK CAPITAL LETTER UPSILON WITH VARIA
0x1FEB, 0x1F7B, // GREEK CAPITAL LETTER UPSILON WITH OXIA
0x1FEC, 0x1FE5, // GREEK CAPITAL LETTER RHO WITH DASIA
0x1FF8, 0x1F78, // GREEK CAPITAL LETTER OMICRON WITH VARIA
0x1FF9, 0x1F79, // GREEK CAPITAL LETTER OMICRON WITH OXIA
0x1FFA, 0x1F7C, // GREEK CAPITAL LETTER OMEGA WITH VARIA
0x1FFB, 0x1F7D, // GREEK CAPITAL LETTER OMEGA WITH OXIA
0x1FFC, 0x1FF3, // GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI
0x2126, 0x03C9, // OHM SIGN
0x212A, 0x006B, // KELVIN SIGN
0x212B, 0x00E5, // ANGSTROM SIGN
0x2132, 0x214E, // TURNED CAPITAL F
0x2160, 0x2170, // ROMAN NUMERAL ONE
0x2161, 0x2171, // ROMAN NUMERAL TWO
0x2162, 0x2172, // ROMAN NUMERAL THREE
0x2163, 0x2173, // ROMAN NUMERAL FOUR
0x2164, 0x2174, // ROMAN NUMERAL FIVE
0x2165, 0x2175, // ROMAN NUMERAL SIX
0x2166, 0x2176, // ROMAN NUMERAL SEVEN
0x2167, 0x2177, // ROMAN NUMERAL EIGHT
0x2168, 0x2178, // ROMAN NUMERAL NINE
0x2169, 0x2179, // ROMAN NUMERAL TEN
0x216A, 0x217A, // ROMAN NUMERAL ELEVEN
0x216B, 0x217B, // ROMAN NUMERAL TWELVE
0x216C, 0x217C, // ROMAN NUMERAL FIFTY
0x216D, 0x217D, // ROMAN NUMERAL ONE HUNDRED
0x216E, 0x217E, // ROMAN NUMERAL FIVE HUNDRED
0x216F, 0x217F, // ROMAN NUMERAL ONE THOUSAND
0x2183, 0x2184, // ROMAN NUMERAL REVERSED ONE HUNDRED
0x24B6, 0x24D0, // CIRCLED LATIN CAPITAL LETTER A
0x24B7, 0x24D1, // CIRCLED LATIN CAPITAL LETTER B
0x24B8, 0x24D2, // CIRCLED LATIN CAPITAL LETTER C
0x24B9, 0x24D3, // CIRCLED LATIN CAPITAL LETTER D
0x24BA, 0x24D4, // CIRCLED LATIN CAPITAL LETTER E
0x24BB, 0x24D5, // CIRCLED LATIN CAPITAL LETTER F
0x24BC, 0x24D6, // CIRCLED LATIN CAPITAL LETTER G
0x24BD, 0x24D7, // CIRCLED LATIN CAPITAL LETTER H
0x24BE, 0x24D8, // CIRCLED LATIN CAPITAL LETTER I
0x24BF, 0x24D9, // CIRCLED LATIN CAPITAL LETTER J
0x24C0, 0x24DA, // CIRCLED LATIN CAPITAL LETTER K
0x24C1, 0x24DB, // CIRCLED LATIN CAPITAL LETTER L
0x24C2, 0x24DC, // CIRCLED LATIN CAPITAL LETTER M
0x24C3, 0x24DD, // CIRCLED LATIN CAPITAL LETTER N
0x24C4, 0x24DE, // CIRCLED LATIN CAPITAL LETTER O
0x24C5, 0x24DF, // CIRCLED LATIN CAPITAL LETTER P
0x24C6, 0x24E0, // CIRCLED LATIN CAPITAL LETTER Q
0x24C7, 0x24E1, // CIRCLED LATIN CAPITAL LETTER R
0x24C8, 0x24E2, // CIRCLED LATIN CAPITAL LETTER S
0x24C9, 0x24E3, // CIRCLED LATIN CAPITAL LETTER T
0x24CA, 0x24E4, // CIRCLED LATIN CAPITAL LETTER U
0x24CB, 0x24E5, // CIRCLED LATIN CAPITAL LETTER V
0x24CC, 0x24E6, // CIRCLED LATIN CAPITAL LETTER W
0x24CD, 0x24E7, // CIRCLED LATIN CAPITAL LETTER X
0x24CE, 0x24E8, // CIRCLED LATIN CAPITAL LETTER Y
0x24CF, 0x24E9, // CIRCLED LATIN CAPITAL LETTER Z
0x2C00, 0x2C30, // GLAGOLITIC CAPITAL LETTER AZU
0x2C01, 0x2C31, // GLAGOLITIC CAPITAL LETTER BUKY
0x2C02, 0x2C32, // GLAGOLITIC CAPITAL LETTER VEDE
0x2C03, 0x2C33, // GLAGOLITIC CAPITAL LETTER GLAGOLI
0x2C04, 0x2C34, // GLAGOLITIC CAPITAL LETTER DOBRO
0x2C05, 0x2C35, // GLAGOLITIC CAPITAL LETTER YESTU
0x2C06, 0x2C36, // GLAGOLITIC CAPITAL LETTER ZHIVETE
0x2C07, 0x2C37, // GLAGOLITIC CAPITAL LETTER DZELO
0x2C08, 0x2C38, // GLAGOLITIC CAPITAL LETTER ZEMLJA
0x2C09, 0x2C39, // GLAGOLITIC CAPITAL LETTER IZHE
0x2C0A, 0x2C3A, // GLAGOLITIC CAPITAL LETTER INITIAL IZHE
0x2C0B, 0x2C3B, // GLAGOLITIC CAPITAL LETTER I
0x2C0C, 0x2C3C, // GLAGOLITIC CAPITAL LETTER DJERVI
0x2C0D, 0x2C3D, // GLAGOLITIC CAPITAL LETTER KAKO
0x2C0E, 0x2C3E, // GLAGOLITIC CAPITAL LETTER LJUDIJE
0x2C0F, 0x2C3F, // GLAGOLITIC CAPITAL LETTER MYSLITE
0x2C10, 0x2C40, // GLAGOLITIC CAPITAL LETTER NASHI
0x2C11, 0x2C41, // GLAGOLITIC CAPITAL LETTER ONU
0x2C12, 0x2C42, // GLAGOLITIC CAPITAL LETTER POKOJI
0x2C13, 0x2C43, // GLAGOLITIC CAPITAL LETTER RITSI
0x2C14, 0x2C44, // GLAGOLITIC CAPITAL LETTER SLOVO
0x2C15, 0x2C45, // GLAGOLITIC CAPITAL LETTER TVRIDO
0x2C16, 0x2C46, // GLAGOLITIC CAPITAL LETTER UKU
0x2C17, 0x2C47, // GLAGOLITIC CAPITAL LETTER FRITU
0x2C18, 0x2C48, // GLAGOLITIC CAPITAL LETTER HERU
0x2C19, 0x2C49, // GLAGOLITIC CAPITAL LETTER OTU
0x2C1A, 0x2C4A, // GLAGOLITIC CAPITAL LETTER PE
0x2C1B, 0x2C4B, // GLAGOLITIC CAPITAL LETTER SHTA
0x2C1C, 0x2C4C, // GLAGOLITIC CAPITAL LETTER TSI
0x2C1D, 0x2C4D, // GLAGOLITIC CAPITAL LETTER CHRIVI
0x2C1E, 0x2C4E, // GLAGOLITIC CAPITAL LETTER SHA
0x2C1F, 0x2C4F, // GLAGOLITIC CAPITAL LETTER YERU
0x2C20, 0x2C50, // GLAGOLITIC CAPITAL LETTER YERI
0x2C21, 0x2C51, // GLAGOLITIC CAPITAL LETTER YATI
0x2C22, 0x2C52, // GLAGOLITIC CAPITAL LETTER SPIDERY HA
0x2C23, 0x2C53, // GLAGOLITIC CAPITAL LETTER YU
0x2C24, 0x2C54, // GLAGOLITIC CAPITAL LETTER SMALL YUS
0x2C25, 0x2C55, // GLAGOLITIC CAPITAL LETTER SMALL YUS WITH TAIL
0x2C26, 0x2C56, // GLAGOLITIC CAPITAL LETTER YO
0x2C27, 0x2C57, // GLAGOLITIC CAPITAL LETTER IOTATED SMALL YUS
0x2C28, 0x2C58, // GLAGOLITIC CAPITAL LETTER BIG YUS
0x2C29, 0x2C59, // GLAGOLITIC CAPITAL LETTER IOTATED BIG YUS
0x2C2A, 0x2C5A, // GLAGOLITIC CAPITAL LETTER FITA
0x2C2B, 0x2C5B, // GLAGOLITIC CAPITAL LETTER IZHITSA
0x2C2C, 0x2C5C, // GLAGOLITIC CAPITAL LETTER SHTAPIC
0x2C2D, 0x2C5D, // GLAGOLITIC CAPITAL LETTER TROKUTASTI A
0x2C2E, 0x2C5E, // GLAGOLITIC CAPITAL LETTER LATINATE MYSLITE
0x2C60, 0x2C61, // LATIN CAPITAL LETTER L WITH DOUBLE BAR
0x2C62, 0x026B, // LATIN CAPITAL LETTER L WITH MIDDLE TILDE
0x2C63, 0x1D7D, // LATIN CAPITAL LETTER P WITH STROKE
0x2C64, 0x027D, // LATIN CAPITAL LETTER R WITH TAIL
0x2C67, 0x2C68, // LATIN CAPITAL LETTER H WITH DESCENDER
0x2C69, 0x2C6A, // LATIN CAPITAL LETTER K WITH DESCENDER
0x2C6B, 0x2C6C, // LATIN CAPITAL LETTER Z WITH DESCENDER
0x2C6D, 0x0251, // LATIN CAPITAL LETTER ALPHA
0x2C6E, 0x0271, // LATIN CAPITAL LETTER M WITH HOOK
0x2C6F, 0x0250, // LATIN CAPITAL LETTER TURNED A
0x2C70, 0x0252, // LATIN CAPITAL LETTER TURNED ALPHA
0x2C72, 0x2C73, // LATIN CAPITAL LETTER W WITH HOOK
0x2C75, 0x2C76, // LATIN CAPITAL LETTER HALF H
0x2C7E, 0x023F, // LATIN CAPITAL LETTER S WITH SWASH TAIL
0x2C7F, 0x0240, // LATIN CAPITAL LETTER Z WITH SWASH TAIL
0x2C80, 0x2C81, // COPTIC CAPITAL LETTER ALFA
0x2C82, 0x2C83, // COPTIC CAPITAL LETTER VIDA
0x2C84, 0x2C85, // COPTIC CAPITAL LETTER GAMMA
0x2C86, 0x2C87, // COPTIC CAPITAL LETTER DALDA
0x2C88, 0x2C89, // COPTIC CAPITAL LETTER EIE
0x2C8A, 0x2C8B, // COPTIC CAPITAL LETTER SOU
0x2C8C, 0x2C8D, // COPTIC CAPITAL LETTER ZATA
0x2C8E, 0x2C8F, // COPTIC CAPITAL LETTER HATE
0x2C90, 0x2C91, // COPTIC CAPITAL LETTER THETHE
0x2C92, 0x2C93, // COPTIC CAPITAL LETTER IAUDA
0x2C94, 0x2C95, // COPTIC CAPITAL LETTER KAPA
0x2C96, 0x2C97, // COPTIC CAPITAL LETTER LAULA
0x2C98, 0x2C99, // COPTIC CAPITAL LETTER MI
0x2C9A, 0x2C9B, // COPTIC CAPITAL LETTER NI
0x2C9C, 0x2C9D, // COPTIC CAPITAL LETTER KSI
0x2C9E, 0x2C9F, // COPTIC CAPITAL LETTER O
0x2CA0, 0x2CA1, // COPTIC CAPITAL LETTER PI
0x2CA2, 0x2CA3, // COPTIC CAPITAL LETTER RO
0x2CA4, 0x2CA5, // COPTIC CAPITAL LETTER SIMA
0x2CA6, 0x2CA7, // COPTIC CAPITAL LETTER TAU
0x2CA8, 0x2CA9, // COPTIC CAPITAL LETTER UA
0x2CAA, 0x2CAB, // COPTIC CAPITAL LETTER FI
0x2CAC, 0x2CAD, // COPTIC CAPITAL LETTER KHI
0x2CAE, 0x2CAF, // COPTIC CAPITAL LETTER PSI
0x2CB0, 0x2CB1, // COPTIC CAPITAL LETTER OOU
0x2CB2, 0x2CB3, // COPTIC CAPITAL LETTER DIALECT-P ALEF
0x2CB4, 0x2CB5, // COPTIC CAPITAL LETTER OLD COPTIC AIN
0x2CB6, 0x2CB7, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC EIE
0x2CB8, 0x2CB9, // COPTIC CAPITAL LETTER DIALECT-P KAPA
0x2CBA, 0x2CBB, // COPTIC CAPITAL LETTER DIALECT-P NI
0x2CBC, 0x2CBD, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC NI
0x2CBE, 0x2CBF, // COPTIC CAPITAL LETTER OLD COPTIC OOU
0x2CC0, 0x2CC1, // COPTIC CAPITAL LETTER SAMPI
0x2CC2, 0x2CC3, // COPTIC CAPITAL LETTER CROSSED SHEI
0x2CC4, 0x2CC5, // COPTIC CAPITAL LETTER OLD COPTIC SHEI
0x2CC6, 0x2CC7, // COPTIC CAPITAL LETTER OLD COPTIC ESH
0x2CC8, 0x2CC9, // COPTIC CAPITAL LETTER AKHMIMIC KHEI
0x2CCA, 0x2CCB, // COPTIC CAPITAL LETTER DIALECT-P HORI
0x2CCC, 0x2CCD, // COPTIC CAPITAL LETTER OLD COPTIC HORI
0x2CCE, 0x2CCF, // COPTIC CAPITAL LETTER OLD COPTIC HA
0x2CD0, 0x2CD1, // COPTIC CAPITAL LETTER L-SHAPED HA
0x2CD2, 0x2CD3, // COPTIC CAPITAL LETTER OLD COPTIC HEI
0x2CD4, 0x2CD5, // COPTIC CAPITAL LETTER OLD COPTIC HAT
0x2CD6, 0x2CD7, // COPTIC CAPITAL LETTER OLD COPTIC GANGIA
0x2CD8, 0x2CD9, // COPTIC CAPITAL LETTER OLD COPTIC DJA
0x2CDA, 0x2CDB, // COPTIC CAPITAL LETTER OLD COPTIC SHIMA
0x2CDC, 0x2CDD, // COPTIC CAPITAL LETTER OLD NUBIAN SHIMA
0x2CDE, 0x2CDF, // COPTIC CAPITAL LETTER OLD NUBIAN NGI
0x2CE0, 0x2CE1, // COPTIC CAPITAL LETTER OLD NUBIAN NYI
0x2CE2, 0x2CE3, // COPTIC CAPITAL LETTER OLD NUBIAN WAU
0x2CEB, 0x2CEC, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC SHEI
0x2CED, 0x2CEE, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC GANGIA
0xA640, 0xA641, // CYRILLIC CAPITAL LETTER ZEMLYA
0xA642, 0xA643, // CYRILLIC CAPITAL LETTER DZELO
0xA644, 0xA645, // CYRILLIC CAPITAL LETTER REVERSED DZE
0xA646, 0xA647, // CYRILLIC CAPITAL LETTER IOTA
0xA648, 0xA649, // CYRILLIC CAPITAL LETTER DJERV
0xA64A, 0xA64B, // CYRILLIC CAPITAL LETTER MONOGRAPH UK
0xA64C, 0xA64D, // CYRILLIC CAPITAL LETTER BROAD OMEGA
0xA64E, 0xA64F, // CYRILLIC CAPITAL LETTER NEUTRAL YER
0xA650, 0xA651, // CYRILLIC CAPITAL LETTER YERU WITH BACK YER
0xA652, 0xA653, // CYRILLIC CAPITAL LETTER IOTIFIED YAT
0xA654, 0xA655, // CYRILLIC CAPITAL LETTER REVERSED YU
0xA656, 0xA657, // CYRILLIC CAPITAL LETTER IOTIFIED A
0xA658, 0xA659, // CYRILLIC CAPITAL LETTER CLOSED LITTLE YUS
0xA65A, 0xA65B, // CYRILLIC CAPITAL LETTER BLENDED YUS
0xA65C, 0xA65D, // CYRILLIC CAPITAL LETTER IOTIFIED CLOSED LITTLE YUS
0xA65E, 0xA65F, // CYRILLIC CAPITAL LETTER YN
0xA662, 0xA663, // CYRILLIC CAPITAL LETTER SOFT DE
0xA664, 0xA665, // CYRILLIC CAPITAL LETTER SOFT EL
0xA666, 0xA667, // CYRILLIC CAPITAL LETTER SOFT EM
0xA668, 0xA669, // CYRILLIC CAPITAL LETTER MONOCULAR O
0xA66A, 0xA66B, // CYRILLIC CAPITAL LETTER BINOCULAR O
0xA66C, 0xA66D, // CYRILLIC CAPITAL LETTER DOUBLE MONOCULAR O
0xA680, 0xA681, // CYRILLIC CAPITAL LETTER DWE
0xA682, 0xA683, // CYRILLIC CAPITAL LETTER DZWE
0xA684, 0xA685, // CYRILLIC CAPITAL LETTER ZHWE
0xA686, 0xA687, // CYRILLIC CAPITAL LETTER CCHE
0xA688, 0xA689, // CYRILLIC CAPITAL LETTER DZZE
0xA68A, 0xA68B, // CYRILLIC CAPITAL LETTER TE WITH MIDDLE HOOK
0xA68C, 0xA68D, // CYRILLIC CAPITAL LETTER TWE
0xA68E, 0xA68F, // CYRILLIC CAPITAL LETTER TSWE
0xA690, 0xA691, // CYRILLIC CAPITAL LETTER TSSE
0xA692, 0xA693, // CYRILLIC CAPITAL LETTER TCHE
0xA694, 0xA695, // CYRILLIC CAPITAL LETTER HWE
0xA696, 0xA697, // CYRILLIC CAPITAL LETTER SHWE
0xA722, 0xA723, // LATIN CAPITAL LETTER EGYPTOLOGICAL ALEF
0xA724, 0xA725, // LATIN CAPITAL LETTER EGYPTOLOGICAL AIN
0xA726, 0xA727, // LATIN CAPITAL LETTER HENG
0xA728, 0xA729, // LATIN CAPITAL LETTER TZ
0xA72A, 0xA72B, // LATIN CAPITAL LETTER TRESILLO
0xA72C, 0xA72D, // LATIN CAPITAL LETTER CUATRILLO
0xA72E, 0xA72F, // LATIN CAPITAL LETTER CUATRILLO WITH COMMA
0xA732, 0xA733, // LATIN CAPITAL LETTER AA
0xA734, 0xA735, // LATIN CAPITAL LETTER AO
0xA736, 0xA737, // LATIN CAPITAL LETTER AU
0xA738, 0xA739, // LATIN CAPITAL LETTER AV
0xA73A, 0xA73B, // LATIN CAPITAL LETTER AV WITH HORIZONTAL BAR
0xA73C, 0xA73D, // LATIN CAPITAL LETTER AY
0xA73E, 0xA73F, // LATIN CAPITAL LETTER REVERSED C WITH DOT
0xA740, 0xA741, // LATIN CAPITAL LETTER K WITH STROKE
0xA742, 0xA743, // LATIN CAPITAL LETTER K WITH DIAGONAL STROKE
0xA744, 0xA745, // LATIN CAPITAL LETTER K WITH STROKE AND DIAGONAL STROKE
0xA746, 0xA747, // LATIN CAPITAL LETTER BROKEN L
0xA748, 0xA749, // LATIN CAPITAL LETTER L WITH HIGH STROKE
0xA74A, 0xA74B, // LATIN CAPITAL LETTER O WITH LONG STROKE OVERLAY
0xA74C, 0xA74D, // LATIN CAPITAL LETTER O WITH LOOP
0xA74E, 0xA74F, // LATIN CAPITAL LETTER OO
0xA750, 0xA751, // LATIN CAPITAL LETTER P WITH STROKE THROUGH DESCENDER
0xA752, 0xA753, // LATIN CAPITAL LETTER P WITH FLOURISH
0xA754, 0xA755, // LATIN CAPITAL LETTER P WITH SQUIRREL TAIL
0xA756, 0xA757, // LATIN CAPITAL LETTER Q WITH STROKE THROUGH DESCENDER
0xA758, 0xA759, // LATIN CAPITAL LETTER Q WITH DIAGONAL STROKE
0xA75A, 0xA75B, // LATIN CAPITAL LETTER R ROTUNDA
0xA75C, 0xA75D, // LATIN CAPITAL LETTER RUM ROTUNDA
0xA75E, 0xA75F, // LATIN CAPITAL LETTER V WITH DIAGONAL STROKE
0xA760, 0xA761, // LATIN CAPITAL LETTER VY
0xA762, 0xA763, // LATIN CAPITAL LETTER VISIGOTHIC Z
0xA764, 0xA765, // LATIN CAPITAL LETTER THORN WITH STROKE
0xA766, 0xA767, // LATIN CAPITAL LETTER THORN WITH STROKE THROUGH DESCENDER
0xA768, 0xA769, // LATIN CAPITAL LETTER VEND
0xA76A, 0xA76B, // LATIN CAPITAL LETTER ET
0xA76C, 0xA76D, // LATIN CAPITAL LETTER IS
0xA76E, 0xA76F, // LATIN CAPITAL LETTER CON
0xA779, 0xA77A, // LATIN CAPITAL LETTER INSULAR D
0xA77B, 0xA77C, // LATIN CAPITAL LETTER INSULAR F
0xA77D, 0x1D79, // LATIN CAPITAL LETTER INSULAR G
0xA77E, 0xA77F, // LATIN CAPITAL LETTER TURNED INSULAR G
0xA780, 0xA781, // LATIN CAPITAL LETTER TURNED L
0xA782, 0xA783, // LATIN CAPITAL LETTER INSULAR R
0xA784, 0xA785, // LATIN CAPITAL LETTER INSULAR S
0xA786, 0xA787, // LATIN CAPITAL LETTER INSULAR T
0xA78B, 0xA78C, // LATIN CAPITAL LETTER SALTILLO
0xFF21, 0xFF41, // FULLWIDTH LATIN CAPITAL LETTER A
0xFF22, 0xFF42, // FULLWIDTH LATIN CAPITAL LETTER B
0xFF23, 0xFF43, // FULLWIDTH LATIN CAPITAL LETTER C
0xFF24, 0xFF44, // FULLWIDTH LATIN CAPITAL LETTER D
0xFF25, 0xFF45, // FULLWIDTH LATIN CAPITAL LETTER E
0xFF26, 0xFF46, // FULLWIDTH LATIN CAPITAL LETTER F
0xFF27, 0xFF47, // FULLWIDTH LATIN CAPITAL LETTER G
0xFF28, 0xFF48, // FULLWIDTH LATIN CAPITAL LETTER H
0xFF29, 0xFF49, // FULLWIDTH LATIN CAPITAL LETTER I
0xFF2A, 0xFF4A, // FULLWIDTH LATIN CAPITAL LETTER J
0xFF2B, 0xFF4B, // FULLWIDTH LATIN CAPITAL LETTER K
0xFF2C, 0xFF4C, // FULLWIDTH LATIN CAPITAL LETTER L
0xFF2D, 0xFF4D, // FULLWIDTH LATIN CAPITAL LETTER M
0xFF2E, 0xFF4E, // FULLWIDTH LATIN CAPITAL LETTER N
0xFF2F, 0xFF4F, // FULLWIDTH LATIN CAPITAL LETTER O
0xFF30, 0xFF50, // FULLWIDTH LATIN CAPITAL LETTER P
0xFF31, 0xFF51, // FULLWIDTH LATIN CAPITAL LETTER Q
0xFF32, 0xFF52, // FULLWIDTH LATIN CAPITAL LETTER R
0xFF33, 0xFF53, // FULLWIDTH LATIN CAPITAL LETTER S
0xFF34, 0xFF54, // FULLWIDTH LATIN CAPITAL LETTER T
0xFF35, 0xFF55, // FULLWIDTH LATIN CAPITAL LETTER U
0xFF36, 0xFF56, // FULLWIDTH LATIN CAPITAL LETTER V
0xFF37, 0xFF57, // FULLWIDTH LATIN CAPITAL LETTER W
0xFF38, 0xFF58, // FULLWIDTH LATIN CAPITAL LETTER X
0xFF39, 0xFF59, // FULLWIDTH LATIN CAPITAL LETTER Y
0xFF3A, 0xFF5A // FULLWIDTH LATIN CAPITAL LETTER Z
};

// Yuk... Extended plane...
// These are really U+0x10000 + the values below
static const CharSetCvt::MapEnt extendedCaseFold[] = {
0x0400, 0x0428, // DESERET CAPITAL LETTER LONG I
0x0401, 0x0429, // DESERET CAPITAL LETTER LONG E
0x0402, 0x042A, // DESERET CAPITAL LETTER LONG A
0x0403, 0x042B, // DESERET CAPITAL LETTER LONG AH
0x0404, 0x042C, // DESERET CAPITAL LETTER LONG O
0x0405, 0x042D, // DESERET CAPITAL LETTER LONG OO
0x0406, 0x042E, // DESERET CAPITAL LETTER SHORT I
0x0407, 0x042F, // DESERET CAPITAL LETTER SHORT E
0x0408, 0x0430, // DESERET CAPITAL LETTER SHORT A
0x0409, 0x0431, // DESERET CAPITAL LETTER SHORT AH
0x040A, 0x0432, // DESERET CAPITAL LETTER SHORT O
0x040B, 0x0433, // DESERET CAPITAL LETTER SHORT OO
0x040C, 0x0434, // DESERET CAPITAL LETTER AY
0x040D, 0x0435, // DESERET CAPITAL LETTER OW
0x040E, 0x0436, // DESERET CAPITAL LETTER WU
0x040F, 0x0437, // DESERET CAPITAL LETTER YEE
0x0410, 0x0438, // DESERET CAPITAL LETTER H
0x0411, 0x0439, // DESERET CAPITAL LETTER PEE
0x0412, 0x043A, // DESERET CAPITAL LETTER BEE
0x0413, 0x043B, // DESERET CAPITAL LETTER TEE
0x0414, 0x043C, // DESERET CAPITAL LETTER DEE
0x0415, 0x043D, // DESERET CAPITAL LETTER CHEE
0x0416, 0x043E, // DESERET CAPITAL LETTER JEE
0x0417, 0x043F, // DESERET CAPITAL LETTER KAY
0x0418, 0x0440, // DESERET CAPITAL LETTER GAY
0x0419, 0x0441, // DESERET CAPITAL LETTER EF
0x041A, 0x0442, // DESERET CAPITAL LETTER VEE
0x041B, 0x0443, // DESERET CAPITAL LETTER ETH
0x041C, 0x0444, // DESERET CAPITAL LETTER THEE
0x041D, 0x0445, // DESERET CAPITAL LETTER ES
0x041E, 0x0446, // DESERET CAPITAL LETTER ZEE
0x041F, 0x0447, // DESERET CAPITAL LETTER ESH
0x0420, 0x0448, // DESERET CAPITAL LETTER ZHEE
0x0421, 0x0449, // DESERET CAPITAL LETTER ER
0x0422, 0x044A, // DESERET CAPITAL LETTER EL
0x0423, 0x044B, // DESERET CAPITAL LETTER EM
0x0424, 0x044C, // DESERET CAPITAL LETTER EN
0x0425, 0x044D, // DESERET CAPITAL LETTER ENG
0x0426, 0x044E, // DESERET CAPITAL LETTER OI
0x0427, 0x044F  // DESERET CAPITAL LETTER EW
};

static const int caseFoldCnt = sizeof( caseFold ) / sizeof( *caseFold );
static const int extendedCaseFoldCnt =
    sizeof( extendedCaseFold ) / sizeof( *extendedCaseFold );

int
CharSetCvt::Utf8Fold( const StrPtr *s, StrBuf *d )
{
	const unsigned char *t = s->UText();
	const unsigned char *e = s->UEnd();
	int v, l;

	while( t < e )
	{
	    v = *t++;
	    if( v & 0x80 )
	    {
		l = CharSetCvt::bytesFromUTF8[ v ];
		if ( l + t > e )
		{
		    // lasterr = PARTIALCHAR;
		    return 1;
		}
		switch( l )
		{
		default:
		    // lasterr = NOMAPPING;
		    return 1;
		case 3:
		    v <<= 6;
		    v += 0xff & *t++;
		    // fall through...
		case 2:
		    v <<= 6;
		    v += 0xff & *t++;
		    // fall through...
		case 1:
		    v <<= 6;
		    v += 0xff & *t++;
		    v -= CharSetCvt::offsetsFromUTF8[l];
# ifdef STRICT_UTF8
		    if( v < CharSetCvt::minimumFromUTF8[l] )
		    {
			// illegal over long UTF8 sequence
			// lasterr = NOMAPPING;
			return 1;
		    }
#endif
		}
		// v is unicode position
# ifdef STRICT_UTF8
		// check for invalid unicode positions
		if( ( v & 0xfffe ) == 0xfffe || ( v & 0x1ff800 ) == 0xd800 ||
		    ( v >= 0xfdd0 && v <= 0xfdef ) )
		{
		    //lasterr = NOMAPPING;
		    return 1;
		}
# endif
		// fold and extend
		if( v < 0x10000 )
		{
		    unsigned short f = CharSetCvt::MapThru( v,
			caseFold, caseFoldCnt, 0 );
		    if( f )
			v = f;
		    // extend emited multi-byte character
		    if( v >= 0x800 )
		    {
			d->Extend( 0xe0 | ( v >> 12 ) );
			d->Extend( 0x80 | (( v >> 6 ) & 0x3f ) );
		    }
		    else
		    {
			d->Extend( 0xc0 | ( v >> 6 ) );
		    }
		}
		else
		{
		    // extended Plane
		    // note we can not fold into here or from here...
		    unsigned short f = CharSetCvt::MapThru( v & 0xffff,
			extendedCaseFold, extendedCaseFoldCnt, 0 );
		    if( f )
			v = f | 0x10000;
		    d->Extend( 0xf0 | ( v >> 18 ) );
		    d->Extend( 0x80 | ( ( v >> 12 ) & 0x3f ) );
		    d->Extend( 0x80 | ( ( v >> 6 ) & 0x3f ) );
		}
		d->Extend( 0x80 | ( v & 0x3f ) );
	    }
	    else
	    {
		// simple ASCII fold
		if( v >= 'A' && v <= 'Z' )
		    d->Extend( v + 'a' - 'A' );
		else
		    d->Extend( v );
	    }
	}

	return 0;
}