| OLD | NEW |
|---|
| (Empty) | |
| 1 /* |
| 2 * Copyright (C) 2010 The Android Open Source Project |
| 3 * |
| 4 * Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 * you may not use this file except in compliance with the License. |
| 6 * You may obtain a copy of the License at |
| 7 * |
| 8 * http://www.apache.org/licenses/LICENSE-2.0 |
| 9 * |
| 10 * Unless required by applicable law or agreed to in writing, software |
| 11 * distributed under the License is distributed on an "AS IS" BASIS, |
| 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 * See the License for the specific language governing permissions and |
| 14 * limitations under the License. |
| 15 */ |
| 16 |
| 17 #include "third_party/prediction/utils/char_utils.h" |
| 18 |
| 19 #include <cstdlib> |
| 20 |
| 21 #include "third_party/prediction/defines.h" |
| 22 |
| 23 namespace latinime { |
| 24 |
| 25 const int CharUtils::MIN_UNICODE_CODE_POINT = 0; |
| 26 const int CharUtils::MAX_UNICODE_CODE_POINT = 0x10FFFF; |
| 27 |
| 28 struct LatinCapitalSmallPair { |
| 29 unsigned short capital; |
| 30 unsigned short small; |
| 31 }; |
| 32 |
| 33 /* |
| 34 * How to update the SORTED_CHAR_MAP[] array. |
| 35 * |
| 36 * 1. Download http://unicode.org/Public/UNIDATA/UnicodeData.txt |
| 37 * |
| 38 * 2. Have a latest version of ICU4C dev package installed |
| 39 * (Note: the current data has been generated with version 4.8) |
| 40 * $ apt-get install libicu-dev |
| 41 * |
| 42 * 3. Build the following code |
| 43 * $ g++ -o char_utils -I.. -DUPDATING_CHAR_UTILS char_utils.cpp -licuuc |
| 44 */ |
| 45 #ifdef UPDATING_CHAR_UTILS |
| 46 #include <stdio.h> |
| 47 #include <unicode/uchar.h> // ICU4C |
| 48 |
| 49 extern "C" int main() { |
| 50 for (unsigned short c = 0; c < 0xFFFF; c++) { |
| 51 if (c <= 0x7F) |
| 52 continue; |
| 53 const unsigned short icu4cLowerC = u_tolower(c); |
| 54 const unsigned short myLowerC = CharUtils::latin_tolower(c); |
| 55 if (c != icu4cLowerC) { |
| 56 #ifdef CONFIRMING_CHAR_UTILS |
| 57 if (icu4cLowerC != myLowerC) { |
| 58 fprintf(stderr, "icu4cLowerC != myLowerC, 0x%04X, 0x%04X\n", |
| 59 icu4cLowerC, myLowerC); |
| 60 } |
| 61 #else // CONFIRMING_CHAR_UTILS |
| 62 printf("0x%04X, 0x%04X\n", c, icu4cLowerC); |
| 63 #endif // CONFIRMING_CHAR_UTILS |
| 64 } |
| 65 } |
| 66 } |
| 67 #endif // UPDATING_CHAR_UTILS |
| 68 /* |
| 69 * 4. Process the list with UnicodeData.txt |
| 70 * (You need UnicodeData.txt in the current directory) |
| 71 * $ ./char_utils | sort -u | \ |
| 72 * perl -e 'open(FH, "UnicodeData.txt"); @buf = <FH>; close(FH); \ |
| 73 * while(<>){/0x(\w*), 0x(\w*)/; @lines = grep(/^$1/, @buf); @cols
= |
| 74 *split(/;/, $lines[0]); \ |
| 75 * print " { 0x$1, 0x$cols[13] }, // $cols[1]\n";}' |
| 76 * |
| 77 * 5. Update the SORTED_CHAR_MAP[] array below with the output above. |
| 78 * Then, rebuild with -DCONFIRMING_CHAR_UTILS and confirm the program
exits |
| 79 *successfully. |
| 80 * $ g++ -o char_utils -I.. -DUPDATING_CHAR_UTILS -DCONFIRMING_CHAR_U
TILS |
| 81 *char_utils.cpp -licuuc |
| 82 * $ ./char_utils |
| 83 * $ |
| 84 */ |
| 85 static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = { |
| 86 {0x00C0, 0x00E0}, // LATIN CAPITAL LETTER A WITH GRAVE |
| 87 {0x00C1, 0x00E1}, // LATIN CAPITAL LETTER A WITH ACUTE |
| 88 {0x00C2, 0x00E2}, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX |
| 89 {0x00C3, 0x00E3}, // LATIN CAPITAL LETTER A WITH TILDE |
| 90 {0x00C4, 0x00E4}, // LATIN CAPITAL LETTER A WITH DIAERESIS |
| 91 {0x00C5, 0x00E5}, // LATIN CAPITAL LETTER A WITH RING ABOVE |
| 92 {0x00C6, 0x00E6}, // LATIN CAPITAL LETTER AE |
| 93 {0x00C7, 0x00E7}, // LATIN CAPITAL LETTER C WITH CEDILLA |
| 94 {0x00C8, 0x00E8}, // LATIN CAPITAL LETTER E WITH GRAVE |
| 95 {0x00C9, 0x00E9}, // LATIN CAPITAL LETTER E WITH ACUTE |
| 96 {0x00CA, 0x00EA}, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX |
| 97 {0x00CB, 0x00EB}, // LATIN CAPITAL LETTER E WITH DIAERESIS |
| 98 {0x00CC, 0x00EC}, // LATIN CAPITAL LETTER I WITH GRAVE |
| 99 {0x00CD, 0x00ED}, // LATIN CAPITAL LETTER I WITH ACUTE |
| 100 {0x00CE, 0x00EE}, // LATIN CAPITAL LETTER I WITH CIRCUMFLEX |
| 101 {0x00CF, 0x00EF}, // LATIN CAPITAL LETTER I WITH DIAERESIS |
| 102 {0x00D0, 0x00F0}, // LATIN CAPITAL LETTER ETH |
| 103 {0x00D1, 0x00F1}, // LATIN CAPITAL LETTER N WITH TILDE |
| 104 {0x00D2, 0x00F2}, // LATIN CAPITAL LETTER O WITH GRAVE |
| 105 {0x00D3, 0x00F3}, // LATIN CAPITAL LETTER O WITH ACUTE |
| 106 {0x00D4, 0x00F4}, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX |
| 107 {0x00D5, 0x00F5}, // LATIN CAPITAL LETTER O WITH TILDE |
| 108 {0x00D6, 0x00F6}, // LATIN CAPITAL LETTER O WITH DIAERESIS |
| 109 {0x00D8, 0x00F8}, // LATIN CAPITAL LETTER O WITH STROKE |
| 110 {0x00D9, 0x00F9}, // LATIN CAPITAL LETTER U WITH GRAVE |
| 111 {0x00DA, 0x00FA}, // LATIN CAPITAL LETTER U WITH ACUTE |
| 112 {0x00DB, 0x00FB}, // LATIN CAPITAL LETTER U WITH CIRCUMFLEX |
| 113 {0x00DC, 0x00FC}, // LATIN CAPITAL LETTER U WITH DIAERESIS |
| 114 {0x00DD, 0x00FD}, // LATIN CAPITAL LETTER Y WITH ACUTE |
| 115 {0x00DE, 0x00FE}, // LATIN CAPITAL LETTER THORN |
| 116 {0x0100, 0x0101}, // LATIN CAPITAL LETTER A WITH MACRON |
| 117 {0x0102, 0x0103}, // LATIN CAPITAL LETTER A WITH BREVE |
| 118 {0x0104, 0x0105}, // LATIN CAPITAL LETTER A WITH OGONEK |
| 119 {0x0106, 0x0107}, // LATIN CAPITAL LETTER C WITH ACUTE |
| 120 {0x0108, 0x0109}, // LATIN CAPITAL LETTER C WITH CIRCUMFLEX |
| 121 {0x010A, 0x010B}, // LATIN CAPITAL LETTER C WITH DOT ABOVE |
| 122 {0x010C, 0x010D}, // LATIN CAPITAL LETTER C WITH CARON |
| 123 {0x010E, 0x010F}, // LATIN CAPITAL LETTER D WITH CARON |
| 124 {0x0110, 0x0111}, // LATIN CAPITAL LETTER D WITH STROKE |
| 125 {0x0112, 0x0113}, // LATIN CAPITAL LETTER E WITH MACRON |
| 126 {0x0114, 0x0115}, // LATIN CAPITAL LETTER E WITH BREVE |
| 127 {0x0116, 0x0117}, // LATIN CAPITAL LETTER E WITH DOT ABOVE |
| 128 {0x0118, 0x0119}, // LATIN CAPITAL LETTER E WITH OGONEK |
| 129 {0x011A, 0x011B}, // LATIN CAPITAL LETTER E WITH CARON |
| 130 {0x011C, 0x011D}, // LATIN CAPITAL LETTER G WITH CIRCUMFLEX |
| 131 {0x011E, 0x011F}, // LATIN CAPITAL LETTER G WITH BREVE |
| 132 {0x0120, 0x0121}, // LATIN CAPITAL LETTER G WITH DOT ABOVE |
| 133 {0x0122, 0x0123}, // LATIN CAPITAL LETTER G WITH CEDILLA |
| 134 {0x0124, 0x0125}, // LATIN CAPITAL LETTER H WITH CIRCUMFLEX |
| 135 {0x0126, 0x0127}, // LATIN CAPITAL LETTER H WITH STROKE |
| 136 {0x0128, 0x0129}, // LATIN CAPITAL LETTER I WITH TILDE |
| 137 {0x012A, 0x012B}, // LATIN CAPITAL LETTER I WITH MACRON |
| 138 {0x012C, 0x012D}, // LATIN CAPITAL LETTER I WITH BREVE |
| 139 {0x012E, 0x012F}, // LATIN CAPITAL LETTER I WITH OGONEK |
| 140 {0x0130, 0x0069}, // LATIN CAPITAL LETTER I WITH DOT ABOVE |
| 141 {0x0132, 0x0133}, // LATIN CAPITAL LIGATURE IJ |
| 142 {0x0134, 0x0135}, // LATIN CAPITAL LETTER J WITH CIRCUMFLEX |
| 143 {0x0136, 0x0137}, // LATIN CAPITAL LETTER K WITH CEDILLA |
| 144 {0x0139, 0x013A}, // LATIN CAPITAL LETTER L WITH ACUTE |
| 145 {0x013B, 0x013C}, // LATIN CAPITAL LETTER L WITH CEDILLA |
| 146 {0x013D, 0x013E}, // LATIN CAPITAL LETTER L WITH CARON |
| 147 {0x013F, 0x0140}, // LATIN CAPITAL LETTER L WITH MIDDLE DOT |
| 148 {0x0141, 0x0142}, // LATIN CAPITAL LETTER L WITH STROKE |
| 149 {0x0143, 0x0144}, // LATIN CAPITAL LETTER N WITH ACUTE |
| 150 {0x0145, 0x0146}, // LATIN CAPITAL LETTER N WITH CEDILLA |
| 151 {0x0147, 0x0148}, // LATIN CAPITAL LETTER N WITH CARON |
| 152 {0x014A, 0x014B}, // LATIN CAPITAL LETTER ENG |
| 153 {0x014C, 0x014D}, // LATIN CAPITAL LETTER O WITH MACRON |
| 154 {0x014E, 0x014F}, // LATIN CAPITAL LETTER O WITH BREVE |
| 155 {0x0150, 0x0151}, // LATIN CAPITAL LETTER O WITH DOUBLE ACUTE |
| 156 {0x0152, 0x0153}, // LATIN CAPITAL LIGATURE OE |
| 157 {0x0154, 0x0155}, // LATIN CAPITAL LETTER R WITH ACUTE |
| 158 {0x0156, 0x0157}, // LATIN CAPITAL LETTER R WITH CEDILLA |
| 159 {0x0158, 0x0159}, // LATIN CAPITAL LETTER R WITH CARON |
| 160 {0x015A, 0x015B}, // LATIN CAPITAL LETTER S WITH ACUTE |
| 161 {0x015C, 0x015D}, // LATIN CAPITAL LETTER S WITH CIRCUMFLEX |
| 162 {0x015E, 0x015F}, // LATIN CAPITAL LETTER S WITH CEDILLA |
| 163 {0x0160, 0x0161}, // LATIN CAPITAL LETTER S WITH CARON |
| 164 {0x0162, 0x0163}, // LATIN CAPITAL LETTER T WITH CEDILLA |
| 165 {0x0164, 0x0165}, // LATIN CAPITAL LETTER T WITH CARON |
| 166 {0x0166, 0x0167}, // LATIN CAPITAL LETTER T WITH STROKE |
| 167 {0x0168, 0x0169}, // LATIN CAPITAL LETTER U WITH TILDE |
| 168 {0x016A, 0x016B}, // LATIN CAPITAL LETTER U WITH MACRON |
| 169 {0x016C, 0x016D}, // LATIN CAPITAL LETTER U WITH BREVE |
| 170 {0x016E, 0x016F}, // LATIN CAPITAL LETTER U WITH RING ABOVE |
| 171 {0x0170, 0x0171}, // LATIN CAPITAL LETTER U WITH DOUBLE ACUTE |
| 172 {0x0172, 0x0173}, // LATIN CAPITAL LETTER U WITH OGONEK |
| 173 {0x0174, 0x0175}, // LATIN CAPITAL LETTER W WITH CIRCUMFLEX |
| 174 {0x0176, 0x0177}, // LATIN CAPITAL LETTER Y WITH CIRCUMFLEX |
| 175 {0x0178, 0x00FF}, // LATIN CAPITAL LETTER Y WITH DIAERESIS |
| 176 {0x0179, 0x017A}, // LATIN CAPITAL LETTER Z WITH ACUTE |
| 177 {0x017B, 0x017C}, // LATIN CAPITAL LETTER Z WITH DOT ABOVE |
| 178 {0x017D, 0x017E}, // LATIN CAPITAL LETTER Z WITH CARON |
| 179 {0x0181, 0x0253}, // LATIN CAPITAL LETTER B WITH HOOK |
| 180 {0x0182, 0x0183}, // LATIN CAPITAL LETTER B WITH TOPBAR |
| 181 {0x0184, 0x0185}, // LATIN CAPITAL LETTER TONE SIX |
| 182 {0x0186, 0x0254}, // LATIN CAPITAL LETTER OPEN O |
| 183 {0x0187, 0x0188}, // LATIN CAPITAL LETTER C WITH HOOK |
| 184 {0x0189, 0x0256}, // LATIN CAPITAL LETTER AFRICAN D |
| 185 {0x018A, 0x0257}, // LATIN CAPITAL LETTER D WITH HOOK |
| 186 {0x018B, 0x018C}, // LATIN CAPITAL LETTER D WITH TOPBAR |
| 187 {0x018E, 0x01DD}, // LATIN CAPITAL LETTER REVERSED E |
| 188 {0x018F, 0x0259}, // LATIN CAPITAL LETTER SCHWA |
| 189 {0x0190, 0x025B}, // LATIN CAPITAL LETTER OPEN E |
| 190 {0x0191, 0x0192}, // LATIN CAPITAL LETTER F WITH HOOK |
| 191 {0x0193, 0x0260}, // LATIN CAPITAL LETTER G WITH HOOK |
| 192 {0x0194, 0x0263}, // LATIN CAPITAL LETTER GAMMA |
| 193 {0x0196, 0x0269}, // LATIN CAPITAL LETTER IOTA |
| 194 {0x0197, 0x0268}, // LATIN CAPITAL LETTER I WITH STROKE |
| 195 {0x0198, 0x0199}, // LATIN CAPITAL LETTER K WITH HOOK |
| 196 {0x019C, 0x026F}, // LATIN CAPITAL LETTER TURNED M |
| 197 {0x019D, 0x0272}, // LATIN CAPITAL LETTER N WITH LEFT HOOK |
| 198 {0x019F, 0x0275}, // LATIN CAPITAL LETTER O WITH MIDDLE TILDE |
| 199 {0x01A0, 0x01A1}, // LATIN CAPITAL LETTER O WITH HORN |
| 200 {0x01A2, 0x01A3}, // LATIN CAPITAL LETTER OI |
| 201 {0x01A4, 0x01A5}, // LATIN CAPITAL LETTER P WITH HOOK |
| 202 {0x01A6, 0x0280}, // LATIN LETTER YR |
| 203 {0x01A7, 0x01A8}, // LATIN CAPITAL LETTER TONE TWO |
| 204 {0x01A9, 0x0283}, // LATIN CAPITAL LETTER ESH |
| 205 {0x01AC, 0x01AD}, // LATIN CAPITAL LETTER T WITH HOOK |
| 206 {0x01AE, 0x0288}, // LATIN CAPITAL LETTER T WITH RETROFLEX HOOK |
| 207 {0x01AF, 0x01B0}, // LATIN CAPITAL LETTER U WITH HORN |
| 208 {0x01B1, 0x028A}, // LATIN CAPITAL LETTER UPSILON |
| 209 {0x01B2, 0x028B}, // LATIN CAPITAL LETTER V WITH HOOK |
| 210 {0x01B3, 0x01B4}, // LATIN CAPITAL LETTER Y WITH HOOK |
| 211 {0x01B5, 0x01B6}, // LATIN CAPITAL LETTER Z WITH STROKE |
| 212 {0x01B7, 0x0292}, // LATIN CAPITAL LETTER EZH |
| 213 {0x01B8, 0x01B9}, // LATIN CAPITAL LETTER EZH REVERSED |
| 214 {0x01BC, 0x01BD}, // LATIN CAPITAL LETTER TONE FIVE |
| 215 {0x01C4, 0x01C6}, // LATIN CAPITAL LETTER DZ WITH CARON |
| 216 {0x01C5, 0x01C6}, // LATIN CAPITAL LETTER D WITH SMALL LETTER Z WITH CARON |
| 217 {0x01C7, 0x01C9}, // LATIN CAPITAL LETTER LJ |
| 218 {0x01C8, 0x01C9}, // LATIN CAPITAL LETTER L WITH SMALL LETTER J |
| 219 {0x01CA, 0x01CC}, // LATIN CAPITAL LETTER NJ |
| 220 {0x01CB, 0x01CC}, // LATIN CAPITAL LETTER N WITH SMALL LETTER J |
| 221 {0x01CD, 0x01CE}, // LATIN CAPITAL LETTER A WITH CARON |
| 222 {0x01CF, 0x01D0}, // LATIN CAPITAL LETTER I WITH CARON |
| 223 {0x01D1, 0x01D2}, // LATIN CAPITAL LETTER O WITH CARON |
| 224 {0x01D3, 0x01D4}, // LATIN CAPITAL LETTER U WITH CARON |
| 225 {0x01D5, 0x01D6}, // LATIN CAPITAL LETTER U WITH DIAERESIS AND MACRON |
| 226 {0x01D7, 0x01D8}, // LATIN CAPITAL LETTER U WITH DIAERESIS AND ACUTE |
| 227 {0x01D9, 0x01DA}, // LATIN CAPITAL LETTER U WITH DIAERESIS AND CARON |
| 228 {0x01DB, 0x01DC}, // LATIN CAPITAL LETTER U WITH DIAERESIS AND GRAVE |
| 229 {0x01DE, 0x01DF}, // LATIN CAPITAL LETTER A WITH DIAERESIS AND MACRON |
| 230 {0x01E0, 0x01E1}, // LATIN CAPITAL LETTER A WITH DOT ABOVE AND MACRON |
| 231 {0x01E2, 0x01E3}, // LATIN CAPITAL LETTER AE WITH MACRON |
| 232 {0x01E4, 0x01E5}, // LATIN CAPITAL LETTER G WITH STROKE |
| 233 {0x01E6, 0x01E7}, // LATIN CAPITAL LETTER G WITH CARON |
| 234 {0x01E8, 0x01E9}, // LATIN CAPITAL LETTER K WITH CARON |
| 235 {0x01EA, 0x01EB}, // LATIN CAPITAL LETTER O WITH OGONEK |
| 236 {0x01EC, 0x01ED}, // LATIN CAPITAL LETTER O WITH OGONEK AND MACRON |
| 237 {0x01EE, 0x01EF}, // LATIN CAPITAL LETTER EZH WITH CARON |
| 238 {0x01F1, 0x01F3}, // LATIN CAPITAL LETTER DZ |
| 239 {0x01F2, 0x01F3}, // LATIN CAPITAL LETTER D WITH SMALL LETTER Z |
| 240 {0x01F4, 0x01F5}, // LATIN CAPITAL LETTER G WITH ACUTE |
| 241 {0x01F6, 0x0195}, // LATIN CAPITAL LETTER HWAIR |
| 242 {0x01F7, 0x01BF}, // LATIN CAPITAL LETTER WYNN |
| 243 {0x01F8, 0x01F9}, // LATIN CAPITAL LETTER N WITH GRAVE |
| 244 {0x01FA, 0x01FB}, // LATIN CAPITAL LETTER A WITH RING ABOVE AND ACUTE |
| 245 {0x01FC, 0x01FD}, // LATIN CAPITAL LETTER AE WITH ACUTE |
| 246 {0x01FE, 0x01FF}, // LATIN CAPITAL LETTER O WITH STROKE AND ACUTE |
| 247 {0x0200, 0x0201}, // LATIN CAPITAL LETTER A WITH DOUBLE GRAVE |
| 248 {0x0202, 0x0203}, // LATIN CAPITAL LETTER A WITH INVERTED BREVE |
| 249 {0x0204, 0x0205}, // LATIN CAPITAL LETTER E WITH DOUBLE GRAVE |
| 250 {0x0206, 0x0207}, // LATIN CAPITAL LETTER E WITH INVERTED BREVE |
| 251 {0x0208, 0x0209}, // LATIN CAPITAL LETTER I WITH DOUBLE GRAVE |
| 252 {0x020A, 0x020B}, // LATIN CAPITAL LETTER I WITH INVERTED BREVE |
| 253 {0x020C, 0x020D}, // LATIN CAPITAL LETTER O WITH DOUBLE GRAVE |
| 254 {0x020E, 0x020F}, // LATIN CAPITAL LETTER O WITH INVERTED BREVE |
| 255 {0x0210, 0x0211}, // LATIN CAPITAL LETTER R WITH DOUBLE GRAVE |
| 256 {0x0212, 0x0213}, // LATIN CAPITAL LETTER R WITH INVERTED BREVE |
| 257 {0x0214, 0x0215}, // LATIN CAPITAL LETTER U WITH DOUBLE GRAVE |
| 258 {0x0216, 0x0217}, // LATIN CAPITAL LETTER U WITH INVERTED BREVE |
| 259 {0x0218, 0x0219}, // LATIN CAPITAL LETTER S WITH COMMA BELOW |
| 260 {0x021A, 0x021B}, // LATIN CAPITAL LETTER T WITH COMMA BELOW |
| 261 {0x021C, 0x021D}, // LATIN CAPITAL LETTER YOGH |
| 262 {0x021E, 0x021F}, // LATIN CAPITAL LETTER H WITH CARON |
| 263 {0x0220, 0x019E}, // LATIN CAPITAL LETTER N WITH LONG RIGHT LEG |
| 264 {0x0222, 0x0223}, // LATIN CAPITAL LETTER OU |
| 265 {0x0224, 0x0225}, // LATIN CAPITAL LETTER Z WITH HOOK |
| 266 {0x0226, 0x0227}, // LATIN CAPITAL LETTER A WITH DOT ABOVE |
| 267 {0x0228, 0x0229}, // LATIN CAPITAL LETTER E WITH CEDILLA |
| 268 {0x022A, 0x022B}, // LATIN CAPITAL LETTER O WITH DIAERESIS AND MACRON |
| 269 {0x022C, 0x022D}, // LATIN CAPITAL LETTER O WITH TILDE AND MACRON |
| 270 {0x022E, 0x022F}, // LATIN CAPITAL LETTER O WITH DOT ABOVE |
| 271 {0x0230, 0x0231}, // LATIN CAPITAL LETTER O WITH DOT ABOVE AND MACRON |
| 272 {0x0232, 0x0233}, // LATIN CAPITAL LETTER Y WITH MACRON |
| 273 {0x023A, 0x2C65}, // LATIN CAPITAL LETTER A WITH STROKE |
| 274 {0x023B, 0x023C}, // LATIN CAPITAL LETTER C WITH STROKE |
| 275 {0x023D, 0x019A}, // LATIN CAPITAL LETTER L WITH BAR |
| 276 {0x023E, 0x2C66}, // LATIN CAPITAL LETTER T WITH DIAGONAL STROKE |
| 277 {0x0241, 0x0242}, // LATIN CAPITAL LETTER GLOTTAL STOP |
| 278 {0x0243, 0x0180}, // LATIN CAPITAL LETTER B WITH STROKE |
| 279 {0x0244, 0x0289}, // LATIN CAPITAL LETTER U BAR |
| 280 {0x0245, 0x028C}, // LATIN CAPITAL LETTER TURNED V |
| 281 {0x0246, 0x0247}, // LATIN CAPITAL LETTER E WITH STROKE |
| 282 {0x0248, 0x0249}, // LATIN CAPITAL LETTER J WITH STROKE |
| 283 {0x024A, 0x024B}, // LATIN CAPITAL LETTER SMALL Q WITH HOOK TAIL |
| 284 {0x024C, 0x024D}, // LATIN CAPITAL LETTER R WITH STROKE |
| 285 {0x024E, 0x024F}, // LATIN CAPITAL LETTER Y WITH STROKE |
| 286 {0x0370, 0x0371}, // GREEK CAPITAL LETTER HETA |
| 287 {0x0372, 0x0373}, // GREEK CAPITAL LETTER ARCHAIC SAMPI |
| 288 {0x0376, 0x0377}, // GREEK CAPITAL LETTER PAMPHYLIAN DIGAMMA |
| 289 {0x0386, 0x03AC}, // GREEK CAPITAL LETTER ALPHA WITH TONOS |
| 290 {0x0388, 0x03AD}, // GREEK CAPITAL LETTER EPSILON WITH TONOS |
| 291 {0x0389, 0x03AE}, // GREEK CAPITAL LETTER ETA WITH TONOS |
| 292 {0x038A, 0x03AF}, // GREEK CAPITAL LETTER IOTA WITH TONOS |
| 293 {0x038C, 0x03CC}, // GREEK CAPITAL LETTER OMICRON WITH TONOS |
| 294 {0x038E, 0x03CD}, // GREEK CAPITAL LETTER UPSILON WITH TONOS |
| 295 {0x038F, 0x03CE}, // GREEK CAPITAL LETTER OMEGA WITH TONOS |
| 296 {0x0391, 0x03B1}, // GREEK CAPITAL LETTER ALPHA |
| 297 {0x0392, 0x03B2}, // GREEK CAPITAL LETTER BETA |
| 298 {0x0393, 0x03B3}, // GREEK CAPITAL LETTER GAMMA |
| 299 {0x0394, 0x03B4}, // GREEK CAPITAL LETTER DELTA |
| 300 {0x0395, 0x03B5}, // GREEK CAPITAL LETTER EPSILON |
| 301 {0x0396, 0x03B6}, // GREEK CAPITAL LETTER ZETA |
| 302 {0x0397, 0x03B7}, // GREEK CAPITAL LETTER ETA |
| 303 {0x0398, 0x03B8}, // GREEK CAPITAL LETTER THETA |
| 304 {0x0399, 0x03B9}, // GREEK CAPITAL LETTER IOTA |
| 305 {0x039A, 0x03BA}, // GREEK CAPITAL LETTER KAPPA |
| 306 {0x039B, 0x03BB}, // GREEK CAPITAL LETTER LAMDA |
| 307 {0x039C, 0x03BC}, // GREEK CAPITAL LETTER MU |
| 308 {0x039D, 0x03BD}, // GREEK CAPITAL LETTER NU |
| 309 {0x039E, 0x03BE}, // GREEK CAPITAL LETTER XI |
| 310 {0x039F, 0x03BF}, // GREEK CAPITAL LETTER OMICRON |
| 311 {0x03A0, 0x03C0}, // GREEK CAPITAL LETTER PI |
| 312 {0x03A1, 0x03C1}, // GREEK CAPITAL LETTER RHO |
| 313 {0x03A3, 0x03C3}, // GREEK CAPITAL LETTER SIGMA |
| 314 {0x03A4, 0x03C4}, // GREEK CAPITAL LETTER TAU |
| 315 {0x03A5, 0x03C5}, // GREEK CAPITAL LETTER UPSILON |
| 316 {0x03A6, 0x03C6}, // GREEK CAPITAL LETTER PHI |
| 317 {0x03A7, 0x03C7}, // GREEK CAPITAL LETTER CHI |
| 318 {0x03A8, 0x03C8}, // GREEK CAPITAL LETTER PSI |
| 319 {0x03A9, 0x03C9}, // GREEK CAPITAL LETTER OMEGA |
| 320 {0x03AA, 0x03CA}, // GREEK CAPITAL LETTER IOTA WITH DIALYTIKA |
| 321 {0x03AB, 0x03CB}, // GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA |
| 322 {0x03CF, 0x03D7}, // GREEK CAPITAL KAI SYMBOL |
| 323 {0x03D8, 0x03D9}, // GREEK LETTER ARCHAIC KOPPA |
| 324 {0x03DA, 0x03DB}, // GREEK LETTER STIGMA |
| 325 {0x03DC, 0x03DD}, // GREEK LETTER DIGAMMA |
| 326 {0x03DE, 0x03DF}, // GREEK LETTER KOPPA |
| 327 {0x03E0, 0x03E1}, // GREEK LETTER SAMPI |
| 328 {0x03E2, 0x03E3}, // COPTIC CAPITAL LETTER SHEI |
| 329 {0x03E4, 0x03E5}, // COPTIC CAPITAL LETTER FEI |
| 330 {0x03E6, 0x03E7}, // COPTIC CAPITAL LETTER KHEI |
| 331 {0x03E8, 0x03E9}, // COPTIC CAPITAL LETTER HORI |
| 332 {0x03EA, 0x03EB}, // COPTIC CAPITAL LETTER GANGIA |
| 333 {0x03EC, 0x03ED}, // COPTIC CAPITAL LETTER SHIMA |
| 334 {0x03EE, 0x03EF}, // COPTIC CAPITAL LETTER DEI |
| 335 {0x03F4, 0x03B8}, // GREEK CAPITAL THETA SYMBOL |
| 336 {0x03F7, 0x03F8}, // GREEK CAPITAL LETTER SHO |
| 337 {0x03F9, 0x03F2}, // GREEK CAPITAL LUNATE SIGMA SYMBOL |
| 338 {0x03FA, 0x03FB}, // GREEK CAPITAL LETTER SAN |
| 339 {0x03FD, 0x037B}, // GREEK CAPITAL REVERSED LUNATE SIGMA SYMBOL |
| 340 {0x03FE, 0x037C}, // GREEK CAPITAL DOTTED LUNATE SIGMA SYMBOL |
| 341 {0x03FF, 0x037D}, // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL |
| 342 {0x0400, 0x0450}, // CYRILLIC CAPITAL LETTER IE WITH GRAVE |
| 343 {0x0401, 0x0451}, // CYRILLIC CAPITAL LETTER IO |
| 344 {0x0402, 0x0452}, // CYRILLIC CAPITAL LETTER DJE |
| 345 {0x0403, 0x0453}, // CYRILLIC CAPITAL LETTER GJE |
| 346 {0x0404, 0x0454}, // CYRILLIC CAPITAL LETTER UKRAINIAN IE |
| 347 {0x0405, 0x0455}, // CYRILLIC CAPITAL LETTER DZE |
| 348 {0x0406, 0x0456}, // CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I |
| 349 {0x0407, 0x0457}, // CYRILLIC CAPITAL LETTER YI |
| 350 {0x0408, 0x0458}, // CYRILLIC CAPITAL LETTER JE |
| 351 {0x0409, 0x0459}, // CYRILLIC CAPITAL LETTER LJE |
| 352 {0x040A, 0x045A}, // CYRILLIC CAPITAL LETTER NJE |
| 353 {0x040B, 0x045B}, // CYRILLIC CAPITAL LETTER TSHE |
| 354 {0x040C, 0x045C}, // CYRILLIC CAPITAL LETTER KJE |
| 355 {0x040D, 0x045D}, // CYRILLIC CAPITAL LETTER I WITH GRAVE |
| 356 {0x040E, 0x045E}, // CYRILLIC CAPITAL LETTER SHORT U |
| 357 {0x040F, 0x045F}, // CYRILLIC CAPITAL LETTER DZHE |
| 358 {0x0410, 0x0430}, // CYRILLIC CAPITAL LETTER A |
| 359 {0x0411, 0x0431}, // CYRILLIC CAPITAL LETTER BE |
| 360 {0x0412, 0x0432}, // CYRILLIC CAPITAL LETTER VE |
| 361 {0x0413, 0x0433}, // CYRILLIC CAPITAL LETTER GHE |
| 362 {0x0414, 0x0434}, // CYRILLIC CAPITAL LETTER DE |
| 363 {0x0415, 0x0435}, // CYRILLIC CAPITAL LETTER IE |
| 364 {0x0416, 0x0436}, // CYRILLIC CAPITAL LETTER ZHE |
| 365 {0x0417, 0x0437}, // CYRILLIC CAPITAL LETTER ZE |
| 366 {0x0418, 0x0438}, // CYRILLIC CAPITAL LETTER I |
| 367 {0x0419, 0x0439}, // CYRILLIC CAPITAL LETTER SHORT I |
| 368 {0x041A, 0x043A}, // CYRILLIC CAPITAL LETTER KA |
| 369 {0x041B, 0x043B}, // CYRILLIC CAPITAL LETTER EL |
| 370 {0x041C, 0x043C}, // CYRILLIC CAPITAL LETTER EM |
| 371 {0x041D, 0x043D}, // CYRILLIC CAPITAL LETTER EN |
| 372 {0x041E, 0x043E}, // CYRILLIC CAPITAL LETTER O |
| 373 {0x041F, 0x043F}, // CYRILLIC CAPITAL LETTER PE |
| 374 {0x0420, 0x0440}, // CYRILLIC CAPITAL LETTER ER |
| 375 {0x0421, 0x0441}, // CYRILLIC CAPITAL LETTER ES |
| 376 {0x0422, 0x0442}, // CYRILLIC CAPITAL LETTER TE |
| 377 {0x0423, 0x0443}, // CYRILLIC CAPITAL LETTER U |
| 378 {0x0424, 0x0444}, // CYRILLIC CAPITAL LETTER EF |
| 379 {0x0425, 0x0445}, // CYRILLIC CAPITAL LETTER HA |
| 380 {0x0426, 0x0446}, // CYRILLIC CAPITAL LETTER TSE |
| 381 {0x0427, 0x0447}, // CYRILLIC CAPITAL LETTER CHE |
| 382 {0x0428, 0x0448}, // CYRILLIC CAPITAL LETTER SHA |
| 383 {0x0429, 0x0449}, // CYRILLIC CAPITAL LETTER SHCHA |
| 384 {0x042A, 0x044A}, // CYRILLIC CAPITAL LETTER HARD SIGN |
| 385 {0x042B, 0x044B}, // CYRILLIC CAPITAL LETTER YERU |
| 386 {0x042C, 0x044C}, // CYRILLIC CAPITAL LETTER SOFT SIGN |
| 387 {0x042D, 0x044D}, // CYRILLIC CAPITAL LETTER E |
| 388 {0x042E, 0x044E}, // CYRILLIC CAPITAL LETTER YU |
| 389 {0x042F, 0x044F}, // CYRILLIC CAPITAL LETTER YA |
| 390 {0x0460, 0x0461}, // CYRILLIC CAPITAL LETTER OMEGA |
| 391 {0x0462, 0x0463}, // CYRILLIC CAPITAL LETTER YAT |
| 392 {0x0464, 0x0465}, // CYRILLIC CAPITAL LETTER IOTIFIED E |
| 393 {0x0466, 0x0467}, // CYRILLIC CAPITAL LETTER LITTLE YUS |
| 394 {0x0468, 0x0469}, // CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS |
| 395 {0x046A, 0x046B}, // CYRILLIC CAPITAL LETTER BIG YUS |
| 396 {0x046C, 0x046D}, // CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS |
| 397 {0x046E, 0x046F}, // CYRILLIC CAPITAL LETTER KSI |
| 398 {0x0470, 0x0471}, // CYRILLIC CAPITAL LETTER PSI |
| 399 {0x0472, 0x0473}, // CYRILLIC CAPITAL LETTER FITA |
| 400 {0x0474, 0x0475}, // CYRILLIC CAPITAL LETTER IZHITSA |
| 401 {0x0476, |
| 402 0x0477}, // CYRILLIC CAPITAL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT |
| 403 {0x0478, 0x0479}, // CYRILLIC CAPITAL LETTER UK |
| 404 {0x047A, 0x047B}, // CYRILLIC CAPITAL LETTER ROUND OMEGA |
| 405 {0x047C, 0x047D}, // CYRILLIC CAPITAL LETTER OMEGA WITH TITLO |
| 406 {0x047E, 0x047F}, // CYRILLIC CAPITAL LETTER OT |
| 407 {0x0480, 0x0481}, // CYRILLIC CAPITAL LETTER KOPPA |
| 408 {0x048A, 0x048B}, // CYRILLIC CAPITAL LETTER SHORT I WITH TAIL |
| 409 {0x048C, 0x048D}, // CYRILLIC CAPITAL LETTER SEMISOFT SIGN |
| 410 {0x048E, 0x048F}, // CYRILLIC CAPITAL LETTER ER WITH TICK |
| 411 {0x0490, 0x0491}, // CYRILLIC CAPITAL LETTER GHE WITH UPTURN |
| 412 {0x0492, 0x0493}, // CYRILLIC CAPITAL LETTER GHE WITH STROKE |
| 413 {0x0494, 0x0495}, // CYRILLIC CAPITAL LETTER GHE WITH MIDDLE HOOK |
| 414 {0x0496, 0x0497}, // CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER |
| 415 {0x0498, 0x0499}, // CYRILLIC CAPITAL LETTER ZE WITH DESCENDER |
| 416 {0x049A, 0x049B}, // CYRILLIC CAPITAL LETTER KA WITH DESCENDER |
| 417 {0x049C, 0x049D}, // CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE |
| 418 {0x049E, 0x049F}, // CYRILLIC CAPITAL LETTER KA WITH STROKE |
| 419 {0x04A0, 0x04A1}, // CYRILLIC CAPITAL LETTER BASHKIR KA |
| 420 {0x04A2, 0x04A3}, // CYRILLIC CAPITAL LETTER EN WITH DESCENDER |
| 421 {0x04A4, 0x04A5}, // CYRILLIC CAPITAL LIGATURE EN GHE |
| 422 {0x04A6, 0x04A7}, // CYRILLIC CAPITAL LETTER PE WITH MIDDLE HOOK |
| 423 {0x04A8, 0x04A9}, // CYRILLIC CAPITAL LETTER ABKHASIAN HA |
| 424 {0x04AA, 0x04AB}, // CYRILLIC CAPITAL LETTER ES WITH DESCENDER |
| 425 {0x04AC, 0x04AD}, // CYRILLIC CAPITAL LETTER TE WITH DESCENDER |
| 426 {0x04AE, 0x04AF}, // CYRILLIC CAPITAL LETTER STRAIGHT U |
| 427 {0x04B0, 0x04B1}, // CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE |
| 428 {0x04B2, 0x04B3}, // CYRILLIC CAPITAL LETTER HA WITH DESCENDER |
| 429 {0x04B4, 0x04B5}, // CYRILLIC CAPITAL LIGATURE TE TSE |
| 430 {0x04B6, 0x04B7}, // CYRILLIC CAPITAL LETTER CHE WITH DESCENDER |
| 431 {0x04B8, 0x04B9}, // CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE |
| 432 {0x04BA, 0x04BB}, // CYRILLIC CAPITAL LETTER SHHA |
| 433 {0x04BC, 0x04BD}, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE |
| 434 {0x04BE, 0x04BF}, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER |
| 435 {0x04C0, 0x04CF}, // CYRILLIC LETTER PALOCHKA |
| 436 {0x04C1, 0x04C2}, // CYRILLIC CAPITAL LETTER ZHE WITH BREVE |
| 437 {0x04C3, 0x04C4}, // CYRILLIC CAPITAL LETTER KA WITH HOOK |
| 438 {0x04C5, 0x04C6}, // CYRILLIC CAPITAL LETTER EL WITH TAIL |
| 439 {0x04C7, 0x04C8}, // CYRILLIC CAPITAL LETTER EN WITH HOOK |
| 440 {0x04C9, 0x04CA}, // CYRILLIC CAPITAL LETTER EN WITH TAIL |
| 441 {0x04CB, 0x04CC}, // CYRILLIC CAPITAL LETTER KHAKASSIAN CHE |
| 442 {0x04CD, 0x04CE}, // CYRILLIC CAPITAL LETTER EM WITH TAIL |
| 443 {0x04D0, 0x04D1}, // CYRILLIC CAPITAL LETTER A WITH BREVE |
| 444 {0x04D2, 0x04D3}, // CYRILLIC CAPITAL LETTER A WITH DIAERESIS |
| 445 {0x04D4, 0x04D5}, // CYRILLIC CAPITAL LIGATURE A IE |
| 446 {0x04D6, 0x04D7}, // CYRILLIC CAPITAL LETTER IE WITH BREVE |
| 447 {0x04D8, 0x04D9}, // CYRILLIC CAPITAL LETTER SCHWA |
| 448 {0x04DA, 0x04DB}, // CYRILLIC CAPITAL LETTER SCHWA WITH DIAERESIS |
| 449 {0x04DC, 0x04DD}, // CYRILLIC CAPITAL LETTER ZHE WITH DIAERESIS |
| 450 {0x04DE, 0x04DF}, // CYRILLIC CAPITAL LETTER ZE WITH DIAERESIS |
| 451 {0x04E0, 0x04E1}, // CYRILLIC CAPITAL LETTER ABKHASIAN DZE |
| 452 {0x04E2, 0x04E3}, // CYRILLIC CAPITAL LETTER I WITH MACRON |
| 453 {0x04E4, 0x04E5}, // CYRILLIC CAPITAL LETTER I WITH DIAERESIS |
| 454 {0x04E6, 0x04E7}, // CYRILLIC CAPITAL LETTER O WITH DIAERESIS |
| 455 {0x04E8, 0x04E9}, // CYRILLIC CAPITAL LETTER BARRED O |
| 456 {0x04EA, 0x04EB}, // CYRILLIC CAPITAL LETTER BARRED O WITH DIAERESIS |
| 457 {0x04EC, 0x04ED}, // CYRILLIC CAPITAL LETTER E WITH DIAERESIS |
| 458 {0x04EE, 0x04EF}, // CYRILLIC CAPITAL LETTER U WITH MACRON |
| 459 {0x04F0, 0x04F1}, // CYRILLIC CAPITAL LETTER U WITH DIAERESIS |
| 460 {0x04F2, 0x04F3}, // CYRILLIC CAPITAL LETTER U WITH DOUBLE ACUTE |
| 461 {0x04F4, 0x04F5}, // CYRILLIC CAPITAL LETTER CHE WITH DIAERESIS |
| 462 {0x04F6, 0x04F7}, // CYRILLIC CAPITAL LETTER GHE WITH DESCENDER |
| 463 {0x04F8, 0x04F9}, // CYRILLIC CAPITAL LETTER YERU WITH DIAERESIS |
| 464 {0x04FA, 0x04FB}, // CYRILLIC CAPITAL LETTER GHE WITH STROKE AND HOOK |
| 465 {0x04FC, 0x04FD}, // CYRILLIC CAPITAL LETTER HA WITH HOOK |
| 466 {0x04FE, 0x04FF}, // CYRILLIC CAPITAL LETTER HA WITH STROKE |
| 467 {0x0500, 0x0501}, // CYRILLIC CAPITAL LETTER KOMI DE |
| 468 {0x0502, 0x0503}, // CYRILLIC CAPITAL LETTER KOMI DJE |
| 469 {0x0504, 0x0505}, // CYRILLIC CAPITAL LETTER KOMI ZJE |
| 470 {0x0506, 0x0507}, // CYRILLIC CAPITAL LETTER KOMI DZJE |
| 471 {0x0508, 0x0509}, // CYRILLIC CAPITAL LETTER KOMI LJE |
| 472 {0x050A, 0x050B}, // CYRILLIC CAPITAL LETTER KOMI NJE |
| 473 {0x050C, 0x050D}, // CYRILLIC CAPITAL LETTER KOMI SJE |
| 474 {0x050E, 0x050F}, // CYRILLIC CAPITAL LETTER KOMI TJE |
| 475 {0x0510, 0x0511}, // CYRILLIC CAPITAL LETTER REVERSED ZE |
| 476 {0x0512, 0x0513}, // CYRILLIC CAPITAL LETTER EL WITH HOOK |
| 477 {0x0514, 0x0515}, // CYRILLIC CAPITAL LETTER LHA |
| 478 {0x0516, 0x0517}, // CYRILLIC CAPITAL LETTER RHA |
| 479 {0x0518, 0x0519}, // CYRILLIC CAPITAL LETTER YAE |
| 480 {0x051A, 0x051B}, // CYRILLIC CAPITAL LETTER QA |
| 481 {0x051C, 0x051D}, // CYRILLIC CAPITAL LETTER WE |
| 482 {0x051E, 0x051F}, // CYRILLIC CAPITAL LETTER ALEUT KA |
| 483 {0x0520, 0x0521}, // CYRILLIC CAPITAL LETTER EL WITH MIDDLE HOOK |
| 484 {0x0522, 0x0523}, // CYRILLIC CAPITAL LETTER EN WITH MIDDLE HOOK |
| 485 {0x0524, 0x0525}, // CYRILLIC CAPITAL LETTER PE WITH DESCENDER |
| 486 {0x0526, 0x0527}, // CYRILLIC CAPITAL LETTER SHHA WITH DESCENDER |
| 487 {0x0531, 0x0561}, // ARMENIAN CAPITAL LETTER AYB |
| 488 {0x0532, 0x0562}, // ARMENIAN CAPITAL LETTER BEN |
| 489 {0x0533, 0x0563}, // ARMENIAN CAPITAL LETTER GIM |
| 490 {0x0534, 0x0564}, // ARMENIAN CAPITAL LETTER DA |
| 491 {0x0535, 0x0565}, // ARMENIAN CAPITAL LETTER ECH |
| 492 {0x0536, 0x0566}, // ARMENIAN CAPITAL LETTER ZA |
| 493 {0x0537, 0x0567}, // ARMENIAN CAPITAL LETTER EH |
| 494 {0x0538, 0x0568}, // ARMENIAN CAPITAL LETTER ET |
| 495 {0x0539, 0x0569}, // ARMENIAN CAPITAL LETTER TO |
| 496 {0x053A, 0x056A}, // ARMENIAN CAPITAL LETTER ZHE |
| 497 {0x053B, 0x056B}, // ARMENIAN CAPITAL LETTER INI |
| 498 {0x053C, 0x056C}, // ARMENIAN CAPITAL LETTER LIWN |
| 499 {0x053D, 0x056D}, // ARMENIAN CAPITAL LETTER XEH |
| 500 {0x053E, 0x056E}, // ARMENIAN CAPITAL LETTER CA |
| 501 {0x053F, 0x056F}, // ARMENIAN CAPITAL LETTER KEN |
| 502 {0x0540, 0x0570}, // ARMENIAN CAPITAL LETTER HO |
| 503 {0x0541, 0x0571}, // ARMENIAN CAPITAL LETTER JA |
| 504 {0x0542, 0x0572}, // ARMENIAN CAPITAL LETTER GHAD |
| 505 {0x0543, 0x0573}, // ARMENIAN CAPITAL LETTER CHEH |
| 506 {0x0544, 0x0574}, // ARMENIAN CAPITAL LETTER MEN |
| 507 {0x0545, 0x0575}, // ARMENIAN CAPITAL LETTER YI |
| 508 {0x0546, 0x0576}, // ARMENIAN CAPITAL LETTER NOW |
| 509 {0x0547, 0x0577}, // ARMENIAN CAPITAL LETTER SHA |
| 510 {0x0548, 0x0578}, // ARMENIAN CAPITAL LETTER VO |
| 511 {0x0549, 0x0579}, // ARMENIAN CAPITAL LETTER CHA |
| 512 {0x054A, 0x057A}, // ARMENIAN CAPITAL LETTER PEH |
| 513 {0x054B, 0x057B}, // ARMENIAN CAPITAL LETTER JHEH |
| 514 {0x054C, 0x057C}, // ARMENIAN CAPITAL LETTER RA |
| 515 {0x054D, 0x057D}, // ARMENIAN CAPITAL LETTER SEH |
| 516 {0x054E, 0x057E}, // ARMENIAN CAPITAL LETTER VEW |
| 517 {0x054F, 0x057F}, // ARMENIAN CAPITAL LETTER TIWN |
| 518 {0x0550, 0x0580}, // ARMENIAN CAPITAL LETTER REH |
| 519 {0x0551, 0x0581}, // ARMENIAN CAPITAL LETTER CO |
| 520 {0x0552, 0x0582}, // ARMENIAN CAPITAL LETTER YIWN |
| 521 {0x0553, 0x0583}, // ARMENIAN CAPITAL LETTER PIWR |
| 522 {0x0554, 0x0584}, // ARMENIAN CAPITAL LETTER KEH |
| 523 {0x0555, 0x0585}, // ARMENIAN CAPITAL LETTER OH |
| 524 {0x0556, 0x0586}, // ARMENIAN CAPITAL LETTER FEH |
| 525 {0x10A0, 0x2D00}, // GEORGIAN CAPITAL LETTER AN |
| 526 {0x10A1, 0x2D01}, // GEORGIAN CAPITAL LETTER BAN |
| 527 {0x10A2, 0x2D02}, // GEORGIAN CAPITAL LETTER GAN |
| 528 {0x10A3, 0x2D03}, // GEORGIAN CAPITAL LETTER DON |
| 529 {0x10A4, 0x2D04}, // GEORGIAN CAPITAL LETTER EN |
| 530 {0x10A5, 0x2D05}, // GEORGIAN CAPITAL LETTER VIN |
| 531 {0x10A6, 0x2D06}, // GEORGIAN CAPITAL LETTER ZEN |
| 532 {0x10A7, 0x2D07}, // GEORGIAN CAPITAL LETTER TAN |
| 533 {0x10A8, 0x2D08}, // GEORGIAN CAPITAL LETTER IN |
| 534 {0x10A9, 0x2D09}, // GEORGIAN CAPITAL LETTER KAN |
| 535 {0x10AA, 0x2D0A}, // GEORGIAN CAPITAL LETTER LAS |
| 536 {0x10AB, 0x2D0B}, // GEORGIAN CAPITAL LETTER MAN |
| 537 {0x10AC, 0x2D0C}, // GEORGIAN CAPITAL LETTER NAR |
| 538 {0x10AD, 0x2D0D}, // GEORGIAN CAPITAL LETTER ON |
| 539 {0x10AE, 0x2D0E}, // GEORGIAN CAPITAL LETTER PAR |
| 540 {0x10AF, 0x2D0F}, // GEORGIAN CAPITAL LETTER ZHAR |
| 541 {0x10B0, 0x2D10}, // GEORGIAN CAPITAL LETTER RAE |
| 542 {0x10B1, 0x2D11}, // GEORGIAN CAPITAL LETTER SAN |
| 543 {0x10B2, 0x2D12}, // GEORGIAN CAPITAL LETTER TAR |
| 544 {0x10B3, 0x2D13}, // GEORGIAN CAPITAL LETTER UN |
| 545 {0x10B4, 0x2D14}, // GEORGIAN CAPITAL LETTER PHAR |
| 546 {0x10B5, 0x2D15}, // GEORGIAN CAPITAL LETTER KHAR |
| 547 {0x10B6, 0x2D16}, // GEORGIAN CAPITAL LETTER GHAN |
| 548 {0x10B7, 0x2D17}, // GEORGIAN CAPITAL LETTER QAR |
| 549 {0x10B8, 0x2D18}, // GEORGIAN CAPITAL LETTER SHIN |
| 550 {0x10B9, 0x2D19}, // GEORGIAN CAPITAL LETTER CHIN |
| 551 {0x10BA, 0x2D1A}, // GEORGIAN CAPITAL LETTER CAN |
| 552 {0x10BB, 0x2D1B}, // GEORGIAN CAPITAL LETTER JIL |
| 553 {0x10BC, 0x2D1C}, // GEORGIAN CAPITAL LETTER CIL |
| 554 {0x10BD, 0x2D1D}, // GEORGIAN CAPITAL LETTER CHAR |
| 555 {0x10BE, 0x2D1E}, // GEORGIAN CAPITAL LETTER XAN |
| 556 {0x10BF, 0x2D1F}, // GEORGIAN CAPITAL LETTER JHAN |
| 557 {0x10C0, 0x2D20}, // GEORGIAN CAPITAL LETTER HAE |
| 558 {0x10C1, 0x2D21}, // GEORGIAN CAPITAL LETTER HE |
| 559 {0x10C2, 0x2D22}, // GEORGIAN CAPITAL LETTER HIE |
| 560 {0x10C3, 0x2D23}, // GEORGIAN CAPITAL LETTER WE |
| 561 {0x10C4, 0x2D24}, // GEORGIAN CAPITAL LETTER HAR |
| 562 {0x10C5, 0x2D25}, // GEORGIAN CAPITAL LETTER HOE |
| 563 {0x1E00, 0x1E01}, // LATIN CAPITAL LETTER A WITH RING BELOW |
| 564 {0x1E02, 0x1E03}, // LATIN CAPITAL LETTER B WITH DOT ABOVE |
| 565 {0x1E04, 0x1E05}, // LATIN CAPITAL LETTER B WITH DOT BELOW |
| 566 {0x1E06, 0x1E07}, // LATIN CAPITAL LETTER B WITH LINE BELOW |
| 567 {0x1E08, 0x1E09}, // LATIN CAPITAL LETTER C WITH CEDILLA AND ACUTE |
| 568 {0x1E0A, 0x1E0B}, // LATIN CAPITAL LETTER D WITH DOT ABOVE |
| 569 {0x1E0C, 0x1E0D}, // LATIN CAPITAL LETTER D WITH DOT BELOW |
| 570 {0x1E0E, 0x1E0F}, // LATIN CAPITAL LETTER D WITH LINE BELOW |
| 571 {0x1E10, 0x1E11}, // LATIN CAPITAL LETTER D WITH CEDILLA |
| 572 {0x1E12, 0x1E13}, // LATIN CAPITAL LETTER D WITH CIRCUMFLEX BELOW |
| 573 {0x1E14, 0x1E15}, // LATIN CAPITAL LETTER E WITH MACRON AND GRAVE |
| 574 {0x1E16, 0x1E17}, // LATIN CAPITAL LETTER E WITH MACRON AND ACUTE |
| 575 {0x1E18, 0x1E19}, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX BELOW |
| 576 {0x1E1A, 0x1E1B}, // LATIN CAPITAL LETTER E WITH TILDE BELOW |
| 577 {0x1E1C, 0x1E1D}, // LATIN CAPITAL LETTER E WITH CEDILLA AND BREVE |
| 578 {0x1E1E, 0x1E1F}, // LATIN CAPITAL LETTER F WITH DOT ABOVE |
| 579 {0x1E20, 0x1E21}, // LATIN CAPITAL LETTER G WITH MACRON |
| 580 {0x1E22, 0x1E23}, // LATIN CAPITAL LETTER H WITH DOT ABOVE |
| 581 {0x1E24, 0x1E25}, // LATIN CAPITAL LETTER H WITH DOT BELOW |
| 582 {0x1E26, 0x1E27}, // LATIN CAPITAL LETTER H WITH DIAERESIS |
| 583 {0x1E28, 0x1E29}, // LATIN CAPITAL LETTER H WITH CEDILLA |
| 584 {0x1E2A, 0x1E2B}, // LATIN CAPITAL LETTER H WITH BREVE BELOW |
| 585 {0x1E2C, 0x1E2D}, // LATIN CAPITAL LETTER I WITH TILDE BELOW |
| 586 {0x1E2E, 0x1E2F}, // LATIN CAPITAL LETTER I WITH DIAERESIS AND ACUTE |
| 587 {0x1E30, 0x1E31}, // LATIN CAPITAL LETTER K WITH ACUTE |
| 588 {0x1E32, 0x1E33}, // LATIN CAPITAL LETTER K WITH DOT BELOW |
| 589 {0x1E34, 0x1E35}, // LATIN CAPITAL LETTER K WITH LINE BELOW |
| 590 {0x1E36, 0x1E37}, // LATIN CAPITAL LETTER L WITH DOT BELOW |
| 591 {0x1E38, 0x1E39}, // LATIN CAPITAL LETTER L WITH DOT BELOW AND MACRON |
| 592 {0x1E3A, 0x1E3B}, // LATIN CAPITAL LETTER L WITH LINE BELOW |
| 593 {0x1E3C, 0x1E3D}, // LATIN CAPITAL LETTER L WITH CIRCUMFLEX BELOW |
| 594 {0x1E3E, 0x1E3F}, // LATIN CAPITAL LETTER M WITH ACUTE |
| 595 {0x1E40, 0x1E41}, // LATIN CAPITAL LETTER M WITH DOT ABOVE |
| 596 {0x1E42, 0x1E43}, // LATIN CAPITAL LETTER M WITH DOT BELOW |
| 597 {0x1E44, 0x1E45}, // LATIN CAPITAL LETTER N WITH DOT ABOVE |
| 598 {0x1E46, 0x1E47}, // LATIN CAPITAL LETTER N WITH DOT BELOW |
| 599 {0x1E48, 0x1E49}, // LATIN CAPITAL LETTER N WITH LINE BELOW |
| 600 {0x1E4A, 0x1E4B}, // LATIN CAPITAL LETTER N WITH CIRCUMFLEX BELOW |
| 601 {0x1E4C, 0x1E4D}, // LATIN CAPITAL LETTER O WITH TILDE AND ACUTE |
| 602 {0x1E4E, 0x1E4F}, // LATIN CAPITAL LETTER O WITH TILDE AND DIAERESIS |
| 603 {0x1E50, 0x1E51}, // LATIN CAPITAL LETTER O WITH MACRON AND GRAVE |
| 604 {0x1E52, 0x1E53}, // LATIN CAPITAL LETTER O WITH MACRON AND ACUTE |
| 605 {0x1E54, 0x1E55}, // LATIN CAPITAL LETTER P WITH ACUTE |
| 606 {0x1E56, 0x1E57}, // LATIN CAPITAL LETTER P WITH DOT ABOVE |
| 607 {0x1E58, 0x1E59}, // LATIN CAPITAL LETTER R WITH DOT ABOVE |
| 608 {0x1E5A, 0x1E5B}, // LATIN CAPITAL LETTER R WITH DOT BELOW |
| 609 {0x1E5C, 0x1E5D}, // LATIN CAPITAL LETTER R WITH DOT BELOW AND MACRON |
| 610 {0x1E5E, 0x1E5F}, // LATIN CAPITAL LETTER R WITH LINE BELOW |
| 611 {0x1E60, 0x1E61}, // LATIN CAPITAL LETTER S WITH DOT ABOVE |
| 612 {0x1E62, 0x1E63}, // LATIN CAPITAL LETTER S WITH DOT BELOW |
| 613 {0x1E64, 0x1E65}, // LATIN CAPITAL LETTER S WITH ACUTE AND DOT ABOVE |
| 614 {0x1E66, 0x1E67}, // LATIN CAPITAL LETTER S WITH CARON AND DOT ABOVE |
| 615 {0x1E68, 0x1E69}, // LATIN CAPITAL LETTER S WITH DOT BELOW AND DOT ABOVE |
| 616 {0x1E6A, 0x1E6B}, // LATIN CAPITAL LETTER T WITH DOT ABOVE |
| 617 {0x1E6C, 0x1E6D}, // LATIN CAPITAL LETTER T WITH DOT BELOW |
| 618 {0x1E6E, 0x1E6F}, // LATIN CAPITAL LETTER T WITH LINE BELOW |
| 619 {0x1E70, 0x1E71}, // LATIN CAPITAL LETTER T WITH CIRCUMFLEX BELOW |
| 620 {0x1E72, 0x1E73}, // LATIN CAPITAL LETTER U WITH DIAERESIS BELOW |
| 621 {0x1E74, 0x1E75}, // LATIN CAPITAL LETTER U WITH TILDE BELOW |
| 622 {0x1E76, 0x1E77}, // LATIN CAPITAL LETTER U WITH CIRCUMFLEX BELOW |
| 623 {0x1E78, 0x1E79}, // LATIN CAPITAL LETTER U WITH TILDE AND ACUTE |
| 624 {0x1E7A, 0x1E7B}, // LATIN CAPITAL LETTER U WITH MACRON AND DIAERESIS |
| 625 {0x1E7C, 0x1E7D}, // LATIN CAPITAL LETTER V WITH TILDE |
| 626 {0x1E7E, 0x1E7F}, // LATIN CAPITAL LETTER V WITH DOT BELOW |
| 627 {0x1E80, 0x1E81}, // LATIN CAPITAL LETTER W WITH GRAVE |
| 628 {0x1E82, 0x1E83}, // LATIN CAPITAL LETTER W WITH ACUTE |
| 629 {0x1E84, 0x1E85}, // LATIN CAPITAL LETTER W WITH DIAERESIS |
| 630 {0x1E86, 0x1E87}, // LATIN CAPITAL LETTER W WITH DOT ABOVE |
| 631 {0x1E88, 0x1E89}, // LATIN CAPITAL LETTER W WITH DOT BELOW |
| 632 {0x1E8A, 0x1E8B}, // LATIN CAPITAL LETTER X WITH DOT ABOVE |
| 633 {0x1E8C, 0x1E8D}, // LATIN CAPITAL LETTER X WITH DIAERESIS |
| 634 {0x1E8E, 0x1E8F}, // LATIN CAPITAL LETTER Y WITH DOT ABOVE |
| 635 {0x1E90, 0x1E91}, // LATIN CAPITAL LETTER Z WITH CIRCUMFLEX |
| 636 {0x1E92, 0x1E93}, // LATIN CAPITAL LETTER Z WITH DOT BELOW |
| 637 {0x1E94, 0x1E95}, // LATIN CAPITAL LETTER Z WITH LINE BELOW |
| 638 {0x1E9E, 0x00DF}, // LATIN CAPITAL LETTER SHARP S |
| 639 {0x1EA0, 0x1EA1}, // LATIN CAPITAL LETTER A WITH DOT BELOW |
| 640 {0x1EA2, 0x1EA3}, // LATIN CAPITAL LETTER A WITH HOOK ABOVE |
| 641 {0x1EA4, 0x1EA5}, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND ACUTE |
| 642 {0x1EA6, 0x1EA7}, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND GRAVE |
| 643 {0x1EA8, 0x1EA9}, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND HOOK ABOVE |
| 644 {0x1EAA, 0x1EAB}, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND TILDE |
| 645 {0x1EAC, 0x1EAD}, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND DOT BELOW |
| 646 {0x1EAE, 0x1EAF}, // LATIN CAPITAL LETTER A WITH BREVE AND ACUTE |
| 647 {0x1EB0, 0x1EB1}, // LATIN CAPITAL LETTER A WITH BREVE AND GRAVE |
| 648 {0x1EB2, 0x1EB3}, // LATIN CAPITAL LETTER A WITH BREVE AND HOOK ABOVE |
| 649 {0x1EB4, 0x1EB5}, // LATIN CAPITAL LETTER A WITH BREVE AND TILDE |
| 650 {0x1EB6, 0x1EB7}, // LATIN CAPITAL LETTER A WITH BREVE AND DOT BELOW |
| 651 {0x1EB8, 0x1EB9}, // LATIN CAPITAL LETTER E WITH DOT BELOW |
| 652 {0x1EBA, 0x1EBB}, // LATIN CAPITAL LETTER E WITH HOOK ABOVE |
| 653 {0x1EBC, 0x1EBD}, // LATIN CAPITAL LETTER E WITH TILDE |
| 654 {0x1EBE, 0x1EBF}, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND ACUTE |
| 655 {0x1EC0, 0x1EC1}, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND GRAVE |
| 656 {0x1EC2, 0x1EC3}, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND HOOK ABOVE |
| 657 {0x1EC4, 0x1EC5}, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND TILDE |
| 658 {0x1EC6, 0x1EC7}, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND DOT BELOW |
| 659 {0x1EC8, 0x1EC9}, // LATIN CAPITAL LETTER I WITH HOOK ABOVE |
| 660 {0x1ECA, 0x1ECB}, // LATIN CAPITAL LETTER I WITH DOT BELOW |
| 661 {0x1ECC, 0x1ECD}, // LATIN CAPITAL LETTER O WITH DOT BELOW |
| 662 {0x1ECE, 0x1ECF}, // LATIN CAPITAL LETTER O WITH HOOK ABOVE |
| 663 {0x1ED0, 0x1ED1}, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND ACUTE |
| 664 {0x1ED2, 0x1ED3}, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND GRAVE |
| 665 {0x1ED4, 0x1ED5}, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND HOOK ABOVE |
| 666 {0x1ED6, 0x1ED7}, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND TILDE |
| 667 {0x1ED8, 0x1ED9}, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND DOT BELOW |
| 668 {0x1EDA, 0x1EDB}, // LATIN CAPITAL LETTER O WITH HORN AND ACUTE |
| 669 {0x1EDC, 0x1EDD}, // LATIN CAPITAL LETTER O WITH HORN AND GRAVE |
| 670 {0x1EDE, 0x1EDF}, // LATIN CAPITAL LETTER O WITH HORN AND HOOK ABOVE |
| 671 {0x1EE0, 0x1EE1}, // LATIN CAPITAL LETTER O WITH HORN AND TILDE |
| 672 {0x1EE2, 0x1EE3}, // LATIN CAPITAL LETTER O WITH HORN AND DOT BELOW |
| 673 {0x1EE4, 0x1EE5}, // LATIN CAPITAL LETTER U WITH DOT BELOW |
| 674 {0x1EE6, 0x1EE7}, // LATIN CAPITAL LETTER U WITH HOOK ABOVE |
| 675 {0x1EE8, 0x1EE9}, // LATIN CAPITAL LETTER U WITH HORN AND ACUTE |
| 676 {0x1EEA, 0x1EEB}, // LATIN CAPITAL LETTER U WITH HORN AND GRAVE |
| 677 {0x1EEC, 0x1EED}, // LATIN CAPITAL LETTER U WITH HORN AND HOOK ABOVE |
| 678 {0x1EEE, 0x1EEF}, // LATIN CAPITAL LETTER U WITH HORN AND TILDE |
| 679 {0x1EF0, 0x1EF1}, // LATIN CAPITAL LETTER U WITH HORN AND DOT BELOW |
| 680 {0x1EF2, 0x1EF3}, // LATIN CAPITAL LETTER Y WITH GRAVE |
| 681 {0x1EF4, 0x1EF5}, // LATIN CAPITAL LETTER Y WITH DOT BELOW |
| 682 {0x1EF6, 0x1EF7}, // LATIN CAPITAL LETTER Y WITH HOOK ABOVE |
| 683 {0x1EF8, 0x1EF9}, // LATIN CAPITAL LETTER Y WITH TILDE |
| 684 {0x1EFA, 0x1EFB}, // LATIN CAPITAL LETTER MIDDLE-WELSH LL |
| 685 {0x1EFC, 0x1EFD}, // LATIN CAPITAL LETTER MIDDLE-WELSH V |
| 686 {0x1EFE, 0x1EFF}, // LATIN CAPITAL LETTER Y WITH LOOP |
| 687 {0x1F08, 0x1F00}, // GREEK CAPITAL LETTER ALPHA WITH PSILI |
| 688 {0x1F09, 0x1F01}, // GREEK CAPITAL LETTER ALPHA WITH DASIA |
| 689 {0x1F0A, 0x1F02}, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA |
| 690 {0x1F0B, 0x1F03}, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA |
| 691 {0x1F0C, 0x1F04}, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA |
| 692 {0x1F0D, 0x1F05}, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA |
| 693 {0x1F0E, 0x1F06}, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI |
| 694 {0x1F0F, 0x1F07}, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI |
| 695 {0x1F18, 0x1F10}, // GREEK CAPITAL LETTER EPSILON WITH PSILI |
| 696 {0x1F19, 0x1F11}, // GREEK CAPITAL LETTER EPSILON WITH DASIA |
| 697 {0x1F1A, 0x1F12}, // GREEK CAPITAL LETTER EPSILON WITH PSILI AND VARIA |
| 698 {0x1F1B, 0x1F13}, // GREEK CAPITAL LETTER EPSILON WITH DASIA AND VARIA |
| 699 {0x1F1C, 0x1F14}, // GREEK CAPITAL LETTER EPSILON WITH PSILI AND OXIA |
| 700 {0x1F1D, 0x1F15}, // GREEK CAPITAL LETTER EPSILON WITH DASIA AND OXIA |
| 701 {0x1F28, 0x1F20}, // GREEK CAPITAL LETTER ETA WITH PSILI |
| 702 {0x1F29, 0x1F21}, // GREEK CAPITAL LETTER ETA WITH DASIA |
| 703 {0x1F2A, 0x1F22}, // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA |
| 704 {0x1F2B, 0x1F23}, // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA |
| 705 {0x1F2C, 0x1F24}, // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA |
| 706 {0x1F2D, 0x1F25}, // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA |
| 707 {0x1F2E, 0x1F26}, // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI |
| 708 {0x1F2F, 0x1F27}, // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI |
| 709 {0x1F38, 0x1F30}, // GREEK CAPITAL LETTER IOTA WITH PSILI |
| 710 {0x1F39, 0x1F31}, // GREEK CAPITAL LETTER IOTA WITH DASIA |
| 711 {0x1F3A, 0x1F32}, // GREEK CAPITAL LETTER IOTA WITH PSILI AND VARIA |
| 712 {0x1F3B, 0x1F33}, // GREEK CAPITAL LETTER IOTA WITH DASIA AND VARIA |
| 713 {0x1F3C, 0x1F34}, // GREEK CAPITAL LETTER IOTA WITH PSILI AND OXIA |
| 714 {0x1F3D, 0x1F35}, // GREEK CAPITAL LETTER IOTA WITH DASIA AND OXIA |
| 715 {0x1F3E, 0x1F36}, // GREEK CAPITAL LETTER IOTA WITH PSILI AND PERISPOMENI |
| 716 {0x1F3F, 0x1F37}, // GREEK CAPITAL LETTER IOTA WITH DASIA AND PERISPOMENI |
| 717 {0x1F48, 0x1F40}, // GREEK CAPITAL LETTER OMICRON WITH PSILI |
| 718 {0x1F49, 0x1F41}, // GREEK CAPITAL LETTER OMICRON WITH DASIA |
| 719 {0x1F4A, 0x1F42}, // GREEK CAPITAL LETTER OMICRON WITH PSILI AND VARIA |
| 720 {0x1F4B, 0x1F43}, // GREEK CAPITAL LETTER OMICRON WITH DASIA AND VARIA |
| 721 {0x1F4C, 0x1F44}, // GREEK CAPITAL LETTER OMICRON WITH PSILI AND OXIA |
| 722 {0x1F4D, 0x1F45}, // GREEK CAPITAL LETTER OMICRON WITH DASIA AND OXIA |
| 723 {0x1F59, 0x1F51}, // GREEK CAPITAL LETTER UPSILON WITH DASIA |
| 724 {0x1F5B, 0x1F53}, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND VARIA |
| 725 {0x1F5D, 0x1F55}, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND OXIA |
| 726 {0x1F5F, |
| 727 0x1F57}, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND PERISPOMENI |
| 728 {0x1F68, 0x1F60}, // GREEK CAPITAL LETTER OMEGA WITH PSILI |
| 729 {0x1F69, 0x1F61}, // GREEK CAPITAL LETTER OMEGA WITH DASIA |
| 730 {0x1F6A, 0x1F62}, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA |
| 731 {0x1F6B, 0x1F63}, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA |
| 732 {0x1F6C, 0x1F64}, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA |
| 733 {0x1F6D, 0x1F65}, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA |
| 734 {0x1F6E, 0x1F66}, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI |
| 735 {0x1F6F, 0x1F67}, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI |
| 736 {0x1F88, |
| 737 0x1F80}, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PROSGEGRAMMENI |
| 738 {0x1F89, |
| 739 0x1F81}, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PROSGEGRAMMENI |
| 740 {0x1F8A, 0x1F82}, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA AND |
| 741 // PROSGEGRAMMENI |
| 742 {0x1F8B, 0x1F83}, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA AND |
| 743 // PROSGEGRAMMENI |
| 744 {0x1F8C, 0x1F84}, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA AND |
| 745 // PROSGEGRAMMENI |
| 746 {0x1F8D, 0x1F85}, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA AND |
| 747 // PROSGEGRAMMENI |
| 748 {0x1F8E, 0x1F86}, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI |
| 749 // AND PROSGEGRAMMENI |
| 750 {0x1F8F, 0x1F87}, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI |
| 751 // AND PROSGEGRAMMENI |
| 752 {0x1F98, 0x1F90}, // GREEK CAPITAL LETTER ETA WITH PSILI AND PROSGEGRAMMENI |
| 753 {0x1F99, 0x1F91}, // GREEK CAPITAL LETTER ETA WITH DASIA AND PROSGEGRAMMENI |
| 754 {0x1F9A, 0x1F92}, // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA AND |
| 755 // PROSGEGRAMMENI |
| 756 {0x1F9B, 0x1F93}, // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA AND |
| 757 // PROSGEGRAMMENI |
| 758 {0x1F9C, 0x1F94}, // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA AND |
| 759 // PROSGEGRAMMENI |
| 760 {0x1F9D, 0x1F95}, // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA AND |
| 761 // PROSGEGRAMMENI |
| 762 {0x1F9E, 0x1F96}, // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI |
| 763 // AND PROSGEGRAMMENI |
| 764 {0x1F9F, 0x1F97}, // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI |
| 765 // AND PROSGEGRAMMENI |
| 766 {0x1FA8, |
| 767 0x1FA0}, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PROSGEGRAMMENI |
| 768 {0x1FA9, |
| 769 0x1FA1}, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PROSGEGRAMMENI |
| 770 {0x1FAA, 0x1FA2}, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA AND |
| 771 // PROSGEGRAMMENI |
| 772 {0x1FAB, 0x1FA3}, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA AND |
| 773 // PROSGEGRAMMENI |
| 774 {0x1FAC, 0x1FA4}, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA AND |
| 775 // PROSGEGRAMMENI |
| 776 {0x1FAD, 0x1FA5}, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA AND |
| 777 // PROSGEGRAMMENI |
| 778 {0x1FAE, 0x1FA6}, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI |
| 779 // AND PROSGEGRAMMENI |
| 780 {0x1FAF, 0x1FA7}, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI |
| 781 // AND PROSGEGRAMMENI |
| 782 {0x1FB8, 0x1FB0}, // GREEK CAPITAL LETTER ALPHA WITH VRACHY |
| 783 {0x1FB9, 0x1FB1}, // GREEK CAPITAL LETTER ALPHA WITH MACRON |
| 784 {0x1FBA, 0x1F70}, // GREEK CAPITAL LETTER ALPHA WITH VARIA |
| 785 {0x1FBB, 0x1F71}, // GREEK CAPITAL LETTER ALPHA WITH OXIA |
| 786 {0x1FBC, 0x1FB3}, // GREEK CAPITAL LETTER ALPHA WITH PROSGEGRAMMENI |
| 787 {0x1FC8, 0x1F72}, // GREEK CAPITAL LETTER EPSILON WITH VARIA |
| 788 {0x1FC9, 0x1F73}, // GREEK CAPITAL LETTER EPSILON WITH OXIA |
| 789 {0x1FCA, 0x1F74}, // GREEK CAPITAL LETTER ETA WITH VARIA |
| 790 {0x1FCB, 0x1F75}, // GREEK CAPITAL LETTER ETA WITH OXIA |
| 791 {0x1FCC, 0x1FC3}, // GREEK CAPITAL LETTER ETA WITH PROSGEGRAMMENI |
| 792 {0x1FD8, 0x1FD0}, // GREEK CAPITAL LETTER IOTA WITH VRACHY |
| 793 {0x1FD9, 0x1FD1}, // GREEK CAPITAL LETTER IOTA WITH MACRON |
| 794 {0x1FDA, 0x1F76}, // GREEK CAPITAL LETTER IOTA WITH VARIA |
| 795 {0x1FDB, 0x1F77}, // GREEK CAPITAL LETTER IOTA WITH OXIA |
| 796 {0x1FE8, 0x1FE0}, // GREEK CAPITAL LETTER UPSILON WITH VRACHY |
| 797 {0x1FE9, 0x1FE1}, // GREEK CAPITAL LETTER UPSILON WITH MACRON |
| 798 {0x1FEA, 0x1F7A}, // GREEK CAPITAL LETTER UPSILON WITH VARIA |
| 799 {0x1FEB, 0x1F7B}, // GREEK CAPITAL LETTER UPSILON WITH OXIA |
| 800 {0x1FEC, 0x1FE5}, // GREEK CAPITAL LETTER RHO WITH DASIA |
| 801 {0x1FF8, 0x1F78}, // GREEK CAPITAL LETTER OMICRON WITH VARIA |
| 802 {0x1FF9, 0x1F79}, // GREEK CAPITAL LETTER OMICRON WITH OXIA |
| 803 {0x1FFA, 0x1F7C}, // GREEK CAPITAL LETTER OMEGA WITH VARIA |
| 804 {0x1FFB, 0x1F7D}, // GREEK CAPITAL LETTER OMEGA WITH OXIA |
| 805 {0x1FFC, 0x1FF3}, // GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI |
| 806 {0x2126, 0x03C9}, // OHM SIGN |
| 807 {0x212A, 0x006B}, // KELVIN SIGN |
| 808 {0x212B, 0x00E5}, // ANGSTROM SIGN |
| 809 {0x2132, 0x214E}, // TURNED CAPITAL F |
| 810 {0x2160, 0x2170}, // ROMAN NUMERAL ONE |
| 811 {0x2161, 0x2171}, // ROMAN NUMERAL TWO |
| 812 {0x2162, 0x2172}, // ROMAN NUMERAL THREE |
| 813 {0x2163, 0x2173}, // ROMAN NUMERAL FOUR |
| 814 {0x2164, 0x2174}, // ROMAN NUMERAL FIVE |
| 815 {0x2165, 0x2175}, // ROMAN NUMERAL SIX |
| 816 {0x2166, 0x2176}, // ROMAN NUMERAL SEVEN |
| 817 {0x2167, 0x2177}, // ROMAN NUMERAL EIGHT |
| 818 {0x2168, 0x2178}, // ROMAN NUMERAL NINE |
| 819 {0x2169, 0x2179}, // ROMAN NUMERAL TEN |
| 820 {0x216A, 0x217A}, // ROMAN NUMERAL ELEVEN |
| 821 {0x216B, 0x217B}, // ROMAN NUMERAL TWELVE |
| 822 {0x216C, 0x217C}, // ROMAN NUMERAL FIFTY |
| 823 {0x216D, 0x217D}, // ROMAN NUMERAL ONE HUNDRED |
| 824 {0x216E, 0x217E}, // ROMAN NUMERAL FIVE HUNDRED |
| 825 {0x216F, 0x217F}, // ROMAN NUMERAL ONE THOUSAND |
| 826 {0x2183, 0x2184}, // ROMAN NUMERAL REVERSED ONE HUNDRED |
| 827 {0x24B6, 0x24D0}, // CIRCLED LATIN CAPITAL LETTER A |
| 828 {0x24B7, 0x24D1}, // CIRCLED LATIN CAPITAL LETTER B |
| 829 {0x24B8, 0x24D2}, // CIRCLED LATIN CAPITAL LETTER C |
| 830 {0x24B9, 0x24D3}, // CIRCLED LATIN CAPITAL LETTER D |
| 831 {0x24BA, 0x24D4}, // CIRCLED LATIN CAPITAL LETTER E |
| 832 {0x24BB, 0x24D5}, // CIRCLED LATIN CAPITAL LETTER F |
| 833 {0x24BC, 0x24D6}, // CIRCLED LATIN CAPITAL LETTER G |
| 834 {0x24BD, 0x24D7}, // CIRCLED LATIN CAPITAL LETTER H |
| 835 {0x24BE, 0x24D8}, // CIRCLED LATIN CAPITAL LETTER I |
| 836 {0x24BF, 0x24D9}, // CIRCLED LATIN CAPITAL LETTER J |
| 837 {0x24C0, 0x24DA}, // CIRCLED LATIN CAPITAL LETTER K |
| 838 {0x24C1, 0x24DB}, // CIRCLED LATIN CAPITAL LETTER L |
| 839 {0x24C2, 0x24DC}, // CIRCLED LATIN CAPITAL LETTER M |
| 840 {0x24C3, 0x24DD}, // CIRCLED LATIN CAPITAL LETTER N |
| 841 {0x24C4, 0x24DE}, // CIRCLED LATIN CAPITAL LETTER O |
| 842 {0x24C5, 0x24DF}, // CIRCLED LATIN CAPITAL LETTER P |
| 843 {0x24C6, 0x24E0}, // CIRCLED LATIN CAPITAL LETTER Q |
| 844 {0x24C7, 0x24E1}, // CIRCLED LATIN CAPITAL LETTER R |
| 845 {0x24C8, 0x24E2}, // CIRCLED LATIN CAPITAL LETTER S |
| 846 {0x24C9, 0x24E3}, // CIRCLED LATIN CAPITAL LETTER T |
| 847 {0x24CA, 0x24E4}, // CIRCLED LATIN CAPITAL LETTER U |
| 848 {0x24CB, 0x24E5}, // CIRCLED LATIN CAPITAL LETTER V |
| 849 {0x24CC, 0x24E6}, // CIRCLED LATIN CAPITAL LETTER W |
| 850 {0x24CD, 0x24E7}, // CIRCLED LATIN CAPITAL LETTER X |
| 851 {0x24CE, 0x24E8}, // CIRCLED LATIN CAPITAL LETTER Y |
| 852 {0x24CF, 0x24E9}, // CIRCLED LATIN CAPITAL LETTER Z |
| 853 {0x2C00, 0x2C30}, // GLAGOLITIC CAPITAL LETTER AZU |
| 854 {0x2C01, 0x2C31}, // GLAGOLITIC CAPITAL LETTER BUKY |
| 855 {0x2C02, 0x2C32}, // GLAGOLITIC CAPITAL LETTER VEDE |
| 856 {0x2C03, 0x2C33}, // GLAGOLITIC CAPITAL LETTER GLAGOLI |
| 857 {0x2C04, 0x2C34}, // GLAGOLITIC CAPITAL LETTER DOBRO |
| 858 {0x2C05, 0x2C35}, // GLAGOLITIC CAPITAL LETTER YESTU |
| 859 {0x2C06, 0x2C36}, // GLAGOLITIC CAPITAL LETTER ZHIVETE |
| 860 {0x2C07, 0x2C37}, // GLAGOLITIC CAPITAL LETTER DZELO |
| 861 {0x2C08, 0x2C38}, // GLAGOLITIC CAPITAL LETTER ZEMLJA |
| 862 {0x2C09, 0x2C39}, // GLAGOLITIC CAPITAL LETTER IZHE |
| 863 {0x2C0A, 0x2C3A}, // GLAGOLITIC CAPITAL LETTER INITIAL IZHE |
| 864 {0x2C0B, 0x2C3B}, // GLAGOLITIC CAPITAL LETTER I |
| 865 {0x2C0C, 0x2C3C}, // GLAGOLITIC CAPITAL LETTER DJERVI |
| 866 {0x2C0D, 0x2C3D}, // GLAGOLITIC CAPITAL LETTER KAKO |
| 867 {0x2C0E, 0x2C3E}, // GLAGOLITIC CAPITAL LETTER LJUDIJE |
| 868 {0x2C0F, 0x2C3F}, // GLAGOLITIC CAPITAL LETTER MYSLITE |
| 869 {0x2C10, 0x2C40}, // GLAGOLITIC CAPITAL LETTER NASHI |
| 870 {0x2C11, 0x2C41}, // GLAGOLITIC CAPITAL LETTER ONU |
| 871 {0x2C12, 0x2C42}, // GLAGOLITIC CAPITAL LETTER POKOJI |
| 872 {0x2C13, 0x2C43}, // GLAGOLITIC CAPITAL LETTER RITSI |
| 873 {0x2C14, 0x2C44}, // GLAGOLITIC CAPITAL LETTER SLOVO |
| 874 {0x2C15, 0x2C45}, // GLAGOLITIC CAPITAL LETTER TVRIDO |
| 875 {0x2C16, 0x2C46}, // GLAGOLITIC CAPITAL LETTER UKU |
| 876 {0x2C17, 0x2C47}, // GLAGOLITIC CAPITAL LETTER FRITU |
| 877 {0x2C18, 0x2C48}, // GLAGOLITIC CAPITAL LETTER HERU |
| 878 {0x2C19, 0x2C49}, // GLAGOLITIC CAPITAL LETTER OTU |
| 879 {0x2C1A, 0x2C4A}, // GLAGOLITIC CAPITAL LETTER PE |
| 880 {0x2C1B, 0x2C4B}, // GLAGOLITIC CAPITAL LETTER SHTA |
| 881 {0x2C1C, 0x2C4C}, // GLAGOLITIC CAPITAL LETTER TSI |
| 882 {0x2C1D, 0x2C4D}, // GLAGOLITIC CAPITAL LETTER CHRIVI |
| 883 {0x2C1E, 0x2C4E}, // GLAGOLITIC CAPITAL LETTER SHA |
| 884 {0x2C1F, 0x2C4F}, // GLAGOLITIC CAPITAL LETTER YERU |
| 885 {0x2C20, 0x2C50}, // GLAGOLITIC CAPITAL LETTER YERI |
| 886 {0x2C21, 0x2C51}, // GLAGOLITIC CAPITAL LETTER YATI |
| 887 {0x2C22, 0x2C52}, // GLAGOLITIC CAPITAL LETTER SPIDERY HA |
| 888 {0x2C23, 0x2C53}, // GLAGOLITIC CAPITAL LETTER YU |
| 889 {0x2C24, 0x2C54}, // GLAGOLITIC CAPITAL LETTER SMALL YUS |
| 890 {0x2C25, 0x2C55}, // GLAGOLITIC CAPITAL LETTER SMALL YUS WITH TAIL |
| 891 {0x2C26, 0x2C56}, // GLAGOLITIC CAPITAL LETTER YO |
| 892 {0x2C27, 0x2C57}, // GLAGOLITIC CAPITAL LETTER IOTATED SMALL YUS |
| 893 {0x2C28, 0x2C58}, // GLAGOLITIC CAPITAL LETTER BIG YUS |
| 894 {0x2C29, 0x2C59}, // GLAGOLITIC CAPITAL LETTER IOTATED BIG YUS |
| 895 {0x2C2A, 0x2C5A}, // GLAGOLITIC CAPITAL LETTER FITA |
| 896 {0x2C2B, 0x2C5B}, // GLAGOLITIC CAPITAL LETTER IZHITSA |
| 897 {0x2C2C, 0x2C5C}, // GLAGOLITIC CAPITAL LETTER SHTAPIC |
| 898 {0x2C2D, 0x2C5D}, // GLAGOLITIC CAPITAL LETTER TROKUTASTI A |
| 899 {0x2C2E, 0x2C5E}, // GLAGOLITIC CAPITAL LETTER LATINATE MYSLITE |
| 900 {0x2C60, 0x2C61}, // LATIN CAPITAL LETTER L WITH DOUBLE BAR |
| 901 {0x2C62, 0x026B}, // LATIN CAPITAL LETTER L WITH MIDDLE TILDE |
| 902 {0x2C63, 0x1D7D}, // LATIN CAPITAL LETTER P WITH STROKE |
| 903 {0x2C64, 0x027D}, // LATIN CAPITAL LETTER R WITH TAIL |
| 904 {0x2C67, 0x2C68}, // LATIN CAPITAL LETTER H WITH DESCENDER |
| 905 {0x2C69, 0x2C6A}, // LATIN CAPITAL LETTER K WITH DESCENDER |
| 906 {0x2C6B, 0x2C6C}, // LATIN CAPITAL LETTER Z WITH DESCENDER |
| 907 {0x2C6D, 0x0251}, // LATIN CAPITAL LETTER ALPHA |
| 908 {0x2C6E, 0x0271}, // LATIN CAPITAL LETTER M WITH HOOK |
| 909 {0x2C6F, 0x0250}, // LATIN CAPITAL LETTER TURNED A |
| 910 {0x2C70, 0x0252}, // LATIN CAPITAL LETTER TURNED ALPHA |
| 911 {0x2C72, 0x2C73}, // LATIN CAPITAL LETTER W WITH HOOK |
| 912 {0x2C75, 0x2C76}, // LATIN CAPITAL LETTER HALF H |
| 913 {0x2C7E, 0x023F}, // LATIN CAPITAL LETTER S WITH SWASH TAIL |
| 914 {0x2C7F, 0x0240}, // LATIN CAPITAL LETTER Z WITH SWASH TAIL |
| 915 {0x2C80, 0x2C81}, // COPTIC CAPITAL LETTER ALFA |
| 916 {0x2C82, 0x2C83}, // COPTIC CAPITAL LETTER VIDA |
| 917 {0x2C84, 0x2C85}, // COPTIC CAPITAL LETTER GAMMA |
| 918 {0x2C86, 0x2C87}, // COPTIC CAPITAL LETTER DALDA |
| 919 {0x2C88, 0x2C89}, // COPTIC CAPITAL LETTER EIE |
| 920 {0x2C8A, 0x2C8B}, // COPTIC CAPITAL LETTER SOU |
| 921 {0x2C8C, 0x2C8D}, // COPTIC CAPITAL LETTER ZATA |
| 922 {0x2C8E, 0x2C8F}, // COPTIC CAPITAL LETTER HATE |
| 923 {0x2C90, 0x2C91}, // COPTIC CAPITAL LETTER THETHE |
| 924 {0x2C92, 0x2C93}, // COPTIC CAPITAL LETTER IAUDA |
| 925 {0x2C94, 0x2C95}, // COPTIC CAPITAL LETTER KAPA |
| 926 {0x2C96, 0x2C97}, // COPTIC CAPITAL LETTER LAULA |
| 927 {0x2C98, 0x2C99}, // COPTIC CAPITAL LETTER MI |
| 928 {0x2C9A, 0x2C9B}, // COPTIC CAPITAL LETTER NI |
| 929 {0x2C9C, 0x2C9D}, // COPTIC CAPITAL LETTER KSI |
| 930 {0x2C9E, 0x2C9F}, // COPTIC CAPITAL LETTER O |
| 931 {0x2CA0, 0x2CA1}, // COPTIC CAPITAL LETTER PI |
| 932 {0x2CA2, 0x2CA3}, // COPTIC CAPITAL LETTER RO |
| 933 {0x2CA4, 0x2CA5}, // COPTIC CAPITAL LETTER SIMA |
| 934 {0x2CA6, 0x2CA7}, // COPTIC CAPITAL LETTER TAU |
| 935 {0x2CA8, 0x2CA9}, // COPTIC CAPITAL LETTER UA |
| 936 {0x2CAA, 0x2CAB}, // COPTIC CAPITAL LETTER FI |
| 937 {0x2CAC, 0x2CAD}, // COPTIC CAPITAL LETTER KHI |
| 938 {0x2CAE, 0x2CAF}, // COPTIC CAPITAL LETTER PSI |
| 939 {0x2CB0, 0x2CB1}, // COPTIC CAPITAL LETTER OOU |
| 940 {0x2CB2, 0x2CB3}, // COPTIC CAPITAL LETTER DIALECT-P ALEF |
| 941 {0x2CB4, 0x2CB5}, // COPTIC CAPITAL LETTER OLD COPTIC AIN |
| 942 {0x2CB6, 0x2CB7}, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC EIE |
| 943 {0x2CB8, 0x2CB9}, // COPTIC CAPITAL LETTER DIALECT-P KAPA |
| 944 {0x2CBA, 0x2CBB}, // COPTIC CAPITAL LETTER DIALECT-P NI |
| 945 {0x2CBC, 0x2CBD}, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC NI |
| 946 {0x2CBE, 0x2CBF}, // COPTIC CAPITAL LETTER OLD COPTIC OOU |
| 947 {0x2CC0, 0x2CC1}, // COPTIC CAPITAL LETTER SAMPI |
| 948 {0x2CC2, 0x2CC3}, // COPTIC CAPITAL LETTER CROSSED SHEI |
| 949 {0x2CC4, 0x2CC5}, // COPTIC CAPITAL LETTER OLD COPTIC SHEI |
| 950 {0x2CC6, 0x2CC7}, // COPTIC CAPITAL LETTER OLD COPTIC ESH |
| 951 {0x2CC8, 0x2CC9}, // COPTIC CAPITAL LETTER AKHMIMIC KHEI |
| 952 {0x2CCA, 0x2CCB}, // COPTIC CAPITAL LETTER DIALECT-P HORI |
| 953 {0x2CCC, 0x2CCD}, // COPTIC CAPITAL LETTER OLD COPTIC HORI |
| 954 {0x2CCE, 0x2CCF}, // COPTIC CAPITAL LETTER OLD COPTIC HA |
| 955 {0x2CD0, 0x2CD1}, // COPTIC CAPITAL LETTER L-SHAPED HA |
| 956 {0x2CD2, 0x2CD3}, // COPTIC CAPITAL LETTER OLD COPTIC HEI |
| 957 {0x2CD4, 0x2CD5}, // COPTIC CAPITAL LETTER OLD COPTIC HAT |
| 958 {0x2CD6, 0x2CD7}, // COPTIC CAPITAL LETTER OLD COPTIC GANGIA |
| 959 {0x2CD8, 0x2CD9}, // COPTIC CAPITAL LETTER OLD COPTIC DJA |
| 960 {0x2CDA, 0x2CDB}, // COPTIC CAPITAL LETTER OLD COPTIC SHIMA |
| 961 {0x2CDC, 0x2CDD}, // COPTIC CAPITAL LETTER OLD NUBIAN SHIMA |
| 962 {0x2CDE, 0x2CDF}, // COPTIC CAPITAL LETTER OLD NUBIAN NGI |
| 963 {0x2CE0, 0x2CE1}, // COPTIC CAPITAL LETTER OLD NUBIAN NYI |
| 964 {0x2CE2, 0x2CE3}, // COPTIC CAPITAL LETTER OLD NUBIAN WAU |
| 965 {0x2CEB, 0x2CEC}, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC SHEI |
| 966 {0x2CED, 0x2CEE}, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC GANGIA |
| 967 {0xA640, 0xA641}, // CYRILLIC CAPITAL LETTER ZEMLYA |
| 968 {0xA642, 0xA643}, // CYRILLIC CAPITAL LETTER DZELO |
| 969 {0xA644, 0xA645}, // CYRILLIC CAPITAL LETTER REVERSED DZE |
| 970 {0xA646, 0xA647}, // CYRILLIC CAPITAL LETTER IOTA |
| 971 {0xA648, 0xA649}, // CYRILLIC CAPITAL LETTER DJERV |
| 972 {0xA64A, 0xA64B}, // CYRILLIC CAPITAL LETTER MONOGRAPH UK |
| 973 {0xA64C, 0xA64D}, // CYRILLIC CAPITAL LETTER BROAD OMEGA |
| 974 {0xA64E, 0xA64F}, // CYRILLIC CAPITAL LETTER NEUTRAL YER |
| 975 {0xA650, 0xA651}, // CYRILLIC CAPITAL LETTER YERU WITH BACK YER |
| 976 {0xA652, 0xA653}, // CYRILLIC CAPITAL LETTER IOTIFIED YAT |
| 977 {0xA654, 0xA655}, // CYRILLIC CAPITAL LETTER REVERSED YU |
| 978 {0xA656, 0xA657}, // CYRILLIC CAPITAL LETTER IOTIFIED A |
| 979 {0xA658, 0xA659}, // CYRILLIC CAPITAL LETTER CLOSED LITTLE YUS |
| 980 {0xA65A, 0xA65B}, // CYRILLIC CAPITAL LETTER BLENDED YUS |
| 981 {0xA65C, 0xA65D}, // CYRILLIC CAPITAL LETTER IOTIFIED CLOSED LITTLE YUS |
| 982 {0xA65E, 0xA65F}, // CYRILLIC CAPITAL LETTER YN |
| 983 {0xA660, 0xA661}, // CYRILLIC CAPITAL LETTER REVERSED TSE |
| 984 {0xA662, 0xA663}, // CYRILLIC CAPITAL LETTER SOFT DE |
| 985 {0xA664, 0xA665}, // CYRILLIC CAPITAL LETTER SOFT EL |
| 986 {0xA666, 0xA667}, // CYRILLIC CAPITAL LETTER SOFT EM |
| 987 {0xA668, 0xA669}, // CYRILLIC CAPITAL LETTER MONOCULAR O |
| 988 {0xA66A, 0xA66B}, // CYRILLIC CAPITAL LETTER BINOCULAR O |
| 989 {0xA66C, 0xA66D}, // CYRILLIC CAPITAL LETTER DOUBLE MONOCULAR O |
| 990 {0xA680, 0xA681}, // CYRILLIC CAPITAL LETTER DWE |
| 991 {0xA682, 0xA683}, // CYRILLIC CAPITAL LETTER DZWE |
| 992 {0xA684, 0xA685}, // CYRILLIC CAPITAL LETTER ZHWE |
| 993 {0xA686, 0xA687}, // CYRILLIC CAPITAL LETTER CCHE |
| 994 {0xA688, 0xA689}, // CYRILLIC CAPITAL LETTER DZZE |
| 995 {0xA68A, 0xA68B}, // CYRILLIC CAPITAL LETTER TE WITH MIDDLE HOOK |
| 996 {0xA68C, 0xA68D}, // CYRILLIC CAPITAL LETTER TWE |
| 997 {0xA68E, 0xA68F}, // CYRILLIC CAPITAL LETTER TSWE |
| 998 {0xA690, 0xA691}, // CYRILLIC CAPITAL LETTER TSSE |
| 999 {0xA692, 0xA693}, // CYRILLIC CAPITAL LETTER TCHE |
| 1000 {0xA694, 0xA695}, // CYRILLIC CAPITAL LETTER HWE |
| 1001 {0xA696, 0xA697}, // CYRILLIC CAPITAL LETTER SHWE |
| 1002 {0xA722, 0xA723}, // LATIN CAPITAL LETTER EGYPTOLOGICAL ALEF |
| 1003 {0xA724, 0xA725}, // LATIN CAPITAL LETTER EGYPTOLOGICAL AIN |
| 1004 {0xA726, 0xA727}, // LATIN CAPITAL LETTER HENG |
| 1005 {0xA728, 0xA729}, // LATIN CAPITAL LETTER TZ |
| 1006 {0xA72A, 0xA72B}, // LATIN CAPITAL LETTER TRESILLO |
| 1007 {0xA72C, 0xA72D}, // LATIN CAPITAL LETTER CUATRILLO |
| 1008 {0xA72E, 0xA72F}, // LATIN CAPITAL LETTER CUATRILLO WITH COMMA |
| 1009 {0xA732, 0xA733}, // LATIN CAPITAL LETTER AA |
| 1010 {0xA734, 0xA735}, // LATIN CAPITAL LETTER AO |
| 1011 {0xA736, 0xA737}, // LATIN CAPITAL LETTER AU |
| 1012 {0xA738, 0xA739}, // LATIN CAPITAL LETTER AV |
| 1013 {0xA73A, 0xA73B}, // LATIN CAPITAL LETTER AV WITH HORIZONTAL BAR |
| 1014 {0xA73C, 0xA73D}, // LATIN CAPITAL LETTER AY |
| 1015 {0xA73E, 0xA73F}, // LATIN CAPITAL LETTER REVERSED C WITH DOT |
| 1016 {0xA740, 0xA741}, // LATIN CAPITAL LETTER K WITH STROKE |
| 1017 {0xA742, 0xA743}, // LATIN CAPITAL LETTER K WITH DIAGONAL STROKE |
| 1018 {0xA744, 0xA745}, // LATIN CAPITAL LETTER K WITH STROKE AND DIAGONAL STROKE |
| 1019 {0xA746, 0xA747}, // LATIN CAPITAL LETTER BROKEN L |
| 1020 {0xA748, 0xA749}, // LATIN CAPITAL LETTER L WITH HIGH STROKE |
| 1021 {0xA74A, 0xA74B}, // LATIN CAPITAL LETTER O WITH LONG STROKE OVERLAY |
| 1022 {0xA74C, 0xA74D}, // LATIN CAPITAL LETTER O WITH LOOP |
| 1023 {0xA74E, 0xA74F}, // LATIN CAPITAL LETTER OO |
| 1024 {0xA750, 0xA751}, // LATIN CAPITAL LETTER P WITH STROKE THROUGH DESCENDER |
| 1025 {0xA752, 0xA753}, // LATIN CAPITAL LETTER P WITH FLOURISH |
| 1026 {0xA754, 0xA755}, // LATIN CAPITAL LETTER P WITH SQUIRREL TAIL |
| 1027 {0xA756, 0xA757}, // LATIN CAPITAL LETTER Q WITH STROKE THROUGH DESCENDER |
| 1028 {0xA758, 0xA759}, // LATIN CAPITAL LETTER Q WITH DIAGONAL STROKE |
| 1029 {0xA75A, 0xA75B}, // LATIN CAPITAL LETTER R ROTUNDA |
| 1030 {0xA75C, 0xA75D}, // LATIN CAPITAL LETTER RUM ROTUNDA |
| 1031 {0xA75E, 0xA75F}, // LATIN CAPITAL LETTER V WITH DIAGONAL STROKE |
| 1032 {0xA760, 0xA761}, // LATIN CAPITAL LETTER VY |
| 1033 {0xA762, 0xA763}, // LATIN CAPITAL LETTER VISIGOTHIC Z |
| 1034 {0xA764, 0xA765}, // LATIN CAPITAL LETTER THORN WITH STROKE |
| 1035 {0xA766, |
| 1036 0xA767}, // LATIN CAPITAL LETTER THORN WITH STROKE THROUGH DESCENDER |
| 1037 {0xA768, 0xA769}, // LATIN CAPITAL LETTER VEND |
| 1038 {0xA76A, 0xA76B}, // LATIN CAPITAL LETTER ET |
| 1039 {0xA76C, 0xA76D}, // LATIN CAPITAL LETTER IS |
| 1040 {0xA76E, 0xA76F}, // LATIN CAPITAL LETTER CON |
| 1041 {0xA779, 0xA77A}, // LATIN CAPITAL LETTER INSULAR D |
| 1042 {0xA77B, 0xA77C}, // LATIN CAPITAL LETTER INSULAR F |
| 1043 {0xA77D, 0x1D79}, // LATIN CAPITAL LETTER INSULAR G |
| 1044 {0xA77E, 0xA77F}, // LATIN CAPITAL LETTER TURNED INSULAR G |
| 1045 {0xA780, 0xA781}, // LATIN CAPITAL LETTER TURNED L |
| 1046 {0xA782, 0xA783}, // LATIN CAPITAL LETTER INSULAR R |
| 1047 {0xA784, 0xA785}, // LATIN CAPITAL LETTER INSULAR S |
| 1048 {0xA786, 0xA787}, // LATIN CAPITAL LETTER INSULAR T |
| 1049 {0xA78B, 0xA78C}, // LATIN CAPITAL LETTER SALTILLO |
| 1050 {0xA78D, 0x0265}, // LATIN CAPITAL LETTER TURNED H |
| 1051 {0xA790, 0xA791}, // LATIN CAPITAL LETTER N WITH DESCENDER |
| 1052 {0xA7A0, 0xA7A1}, // LATIN CAPITAL LETTER G WITH OBLIQUE STROKE |
| 1053 {0xA7A2, 0xA7A3}, // LATIN CAPITAL LETTER K WITH OBLIQUE STROKE |
| 1054 {0xA7A4, 0xA7A5}, // LATIN CAPITAL LETTER N WITH OBLIQUE STROKE |
| 1055 {0xA7A6, 0xA7A7}, // LATIN CAPITAL LETTER R WITH OBLIQUE STROKE |
| 1056 {0xA7A8, 0xA7A9}, // LATIN CAPITAL LETTER S WITH OBLIQUE STROKE |
| 1057 {0xFF21, 0xFF41}, // FULLWIDTH LATIN CAPITAL LETTER A |
| 1058 {0xFF22, 0xFF42}, // FULLWIDTH LATIN CAPITAL LETTER B |
| 1059 {0xFF23, 0xFF43}, // FULLWIDTH LATIN CAPITAL LETTER C |
| 1060 {0xFF24, 0xFF44}, // FULLWIDTH LATIN CAPITAL LETTER D |
| 1061 {0xFF25, 0xFF45}, // FULLWIDTH LATIN CAPITAL LETTER E |
| 1062 {0xFF26, 0xFF46}, // FULLWIDTH LATIN CAPITAL LETTER F |
| 1063 {0xFF27, 0xFF47}, // FULLWIDTH LATIN CAPITAL LETTER G |
| 1064 {0xFF28, 0xFF48}, // FULLWIDTH LATIN CAPITAL LETTER H |
| 1065 {0xFF29, 0xFF49}, // FULLWIDTH LATIN CAPITAL LETTER I |
| 1066 {0xFF2A, 0xFF4A}, // FULLWIDTH LATIN CAPITAL LETTER J |
| 1067 {0xFF2B, 0xFF4B}, // FULLWIDTH LATIN CAPITAL LETTER K |
| 1068 {0xFF2C, 0xFF4C}, // FULLWIDTH LATIN CAPITAL LETTER L |
| 1069 {0xFF2D, 0xFF4D}, // FULLWIDTH LATIN CAPITAL LETTER M |
| 1070 {0xFF2E, 0xFF4E}, // FULLWIDTH LATIN CAPITAL LETTER N |
| 1071 {0xFF2F, 0xFF4F}, // FULLWIDTH LATIN CAPITAL LETTER O |
| 1072 {0xFF30, 0xFF50}, // FULLWIDTH LATIN CAPITAL LETTER P |
| 1073 {0xFF31, 0xFF51}, // FULLWIDTH LATIN CAPITAL LETTER Q |
| 1074 {0xFF32, 0xFF52}, // FULLWIDTH LATIN CAPITAL LETTER R |
| 1075 {0xFF33, 0xFF53}, // FULLWIDTH LATIN CAPITAL LETTER S |
| 1076 {0xFF34, 0xFF54}, // FULLWIDTH LATIN CAPITAL LETTER T |
| 1077 {0xFF35, 0xFF55}, // FULLWIDTH LATIN CAPITAL LETTER U |
| 1078 {0xFF36, 0xFF56}, // FULLWIDTH LATIN CAPITAL LETTER V |
| 1079 {0xFF37, 0xFF57}, // FULLWIDTH LATIN CAPITAL LETTER W |
| 1080 {0xFF38, 0xFF58}, // FULLWIDTH LATIN CAPITAL LETTER X |
| 1081 {0xFF39, 0xFF59}, // FULLWIDTH LATIN CAPITAL LETTER Y |
| 1082 {0xFF3A, 0xFF5A} // FULLWIDTH LATIN CAPITAL LETTER Z |
| 1083 }; |
| 1084 |
| 1085 static int compare_pair_capital(const void* a, const void* b) { |
| 1086 return static_cast<int>(*static_cast<const unsigned short*>(a)) - |
| 1087 static_cast<int>( |
| 1088 (static_cast<const struct LatinCapitalSmallPair*>(b))->capital); |
| 1089 } |
| 1090 |
| 1091 /* static */ unsigned short CharUtils::latin_tolower(const unsigned short c) { |
| 1092 struct LatinCapitalSmallPair* p = static_cast<struct LatinCapitalSmallPair*>( |
| 1093 bsearch(&c, SORTED_CHAR_MAP, NELEMS(SORTED_CHAR_MAP), |
| 1094 sizeof(SORTED_CHAR_MAP[0]), compare_pair_capital)); |
| 1095 return p ? p->small : c; |
| 1096 } |
| 1097 |
| 1098 /* |
| 1099 * Table mapping most combined Latin, Greek, and Cyrillic characters |
| 1100 * to their base characters. If c is in range, CharUtils::BASE_CHARS[c] == c |
| 1101 * if c is not a combined character, or the base character if it |
| 1102 * is combined. |
| 1103 * |
| 1104 * Generated with: |
| 1105 * cat UnicodeData.txt | perl -e 'while (<>) { @foo = split(/;/); $foo[5] =~ |
| 1106 *s/<.*> //; \ |
| 1107 * $base[hex($foo[0])] = hex($foo[5]);} \ |
| 1108 * for ($i = 0; $i < 0x500; $i += 8) { printf("/" . "* U+%04X *" . "/ ", $i); |
| 1109 *\ |
| 1110 * for ($j = $i; $j < $i + 8; $j++) { \ |
| 1111 * printf("0x%04X, ", $base[$j] ? $base[$j] : $j)}; print "\n"; }' |
| 1112 */ |
| 1113 /* static */ const unsigned short |
| 1114 CharUtils::BASE_CHARS[CharUtils::BASE_CHARS_SIZE] = { |
| 1115 /* U+0000 */ 0x0000, |
| 1116 0x0001, |
| 1117 0x0002, |
| 1118 0x0003, |
| 1119 0x0004, |
| 1120 0x0005, |
| 1121 0x0006, |
| 1122 0x0007, |
| 1123 /* U+0008 */ 0x0008, |
| 1124 0x0009, |
| 1125 0x000A, |
| 1126 0x000B, |
| 1127 0x000C, |
| 1128 0x000D, |
| 1129 0x000E, |
| 1130 0x000F, |
| 1131 /* U+0010 */ 0x0010, |
| 1132 0x0011, |
| 1133 0x0012, |
| 1134 0x0013, |
| 1135 0x0014, |
| 1136 0x0015, |
| 1137 0x0016, |
| 1138 0x0017, |
| 1139 /* U+0018 */ 0x0018, |
| 1140 0x0019, |
| 1141 0x001A, |
| 1142 0x001B, |
| 1143 0x001C, |
| 1144 0x001D, |
| 1145 0x001E, |
| 1146 0x001F, |
| 1147 /* U+0020 */ 0x0020, |
| 1148 0x0021, |
| 1149 0x0022, |
| 1150 0x0023, |
| 1151 0x0024, |
| 1152 0x0025, |
| 1153 0x0026, |
| 1154 0x0027, |
| 1155 /* U+0028 */ 0x0028, |
| 1156 0x0029, |
| 1157 0x002A, |
| 1158 0x002B, |
| 1159 0x002C, |
| 1160 0x002D, |
| 1161 0x002E, |
| 1162 0x002F, |
| 1163 /* U+0030 */ 0x0030, |
| 1164 0x0031, |
| 1165 0x0032, |
| 1166 0x0033, |
| 1167 0x0034, |
| 1168 0x0035, |
| 1169 0x0036, |
| 1170 0x0037, |
| 1171 /* U+0038 */ 0x0038, |
| 1172 0x0039, |
| 1173 0x003A, |
| 1174 0x003B, |
| 1175 0x003C, |
| 1176 0x003D, |
| 1177 0x003E, |
| 1178 0x003F, |
| 1179 /* U+0040 */ 0x0040, |
| 1180 0x0041, |
| 1181 0x0042, |
| 1182 0x0043, |
| 1183 0x0044, |
| 1184 0x0045, |
| 1185 0x0046, |
| 1186 0x0047, |
| 1187 /* U+0048 */ 0x0048, |
| 1188 0x0049, |
| 1189 0x004A, |
| 1190 0x004B, |
| 1191 0x004C, |
| 1192 0x004D, |
| 1193 0x004E, |
| 1194 0x004F, |
| 1195 /* U+0050 */ 0x0050, |
| 1196 0x0051, |
| 1197 0x0052, |
| 1198 0x0053, |
| 1199 0x0054, |
| 1200 0x0055, |
| 1201 0x0056, |
| 1202 0x0057, |
| 1203 /* U+0058 */ 0x0058, |
| 1204 0x0059, |
| 1205 0x005A, |
| 1206 0x005B, |
| 1207 0x005C, |
| 1208 0x005D, |
| 1209 0x005E, |
| 1210 0x005F, |
| 1211 /* U+0060 */ 0x0060, |
| 1212 0x0061, |
| 1213 0x0062, |
| 1214 0x0063, |
| 1215 0x0064, |
| 1216 0x0065, |
| 1217 0x0066, |
| 1218 0x0067, |
| 1219 /* U+0068 */ 0x0068, |
| 1220 0x0069, |
| 1221 0x006A, |
| 1222 0x006B, |
| 1223 0x006C, |
| 1224 0x006D, |
| 1225 0x006E, |
| 1226 0x006F, |
| 1227 /* U+0070 */ 0x0070, |
| 1228 0x0071, |
| 1229 0x0072, |
| 1230 0x0073, |
| 1231 0x0074, |
| 1232 0x0075, |
| 1233 0x0076, |
| 1234 0x0077, |
| 1235 /* U+0078 */ 0x0078, |
| 1236 0x0079, |
| 1237 0x007A, |
| 1238 0x007B, |
| 1239 0x007C, |
| 1240 0x007D, |
| 1241 0x007E, |
| 1242 0x007F, |
| 1243 /* U+0080 */ 0x0080, |
| 1244 0x0081, |
| 1245 0x0082, |
| 1246 0x0083, |
| 1247 0x0084, |
| 1248 0x0085, |
| 1249 0x0086, |
| 1250 0x0087, |
| 1251 /* U+0088 */ 0x0088, |
| 1252 0x0089, |
| 1253 0x008A, |
| 1254 0x008B, |
| 1255 0x008C, |
| 1256 0x008D, |
| 1257 0x008E, |
| 1258 0x008F, |
| 1259 /* U+0090 */ 0x0090, |
| 1260 0x0091, |
| 1261 0x0092, |
| 1262 0x0093, |
| 1263 0x0094, |
| 1264 0x0095, |
| 1265 0x0096, |
| 1266 0x0097, |
| 1267 /* U+0098 */ 0x0098, |
| 1268 0x0099, |
| 1269 0x009A, |
| 1270 0x009B, |
| 1271 0x009C, |
| 1272 0x009D, |
| 1273 0x009E, |
| 1274 0x009F, |
| 1275 /* U+00A0 */ 0x0020, |
| 1276 0x00A1, |
| 1277 0x00A2, |
| 1278 0x00A3, |
| 1279 0x00A4, |
| 1280 0x00A5, |
| 1281 0x00A6, |
| 1282 0x00A7, |
| 1283 /* U+00A8 */ 0x0020, |
| 1284 0x00A9, |
| 1285 0x0061, |
| 1286 0x00AB, |
| 1287 0x00AC, |
| 1288 0x00AD, |
| 1289 0x00AE, |
| 1290 0x0020, |
| 1291 /* U+00B0 */ 0x00B0, |
| 1292 0x00B1, |
| 1293 0x0032, |
| 1294 0x0033, |
| 1295 0x0020, |
| 1296 0x03BC, |
| 1297 0x00B6, |
| 1298 0x00B7, |
| 1299 /* U+00B8 */ 0x0020, |
| 1300 0x0031, |
| 1301 0x006F, |
| 1302 0x00BB, |
| 1303 0x0031, |
| 1304 0x0031, |
| 1305 0x0033, |
| 1306 0x00BF, |
| 1307 /* U+00C0 */ 0x0041, |
| 1308 0x0041, |
| 1309 0x0041, |
| 1310 0x0041, |
| 1311 0x0041, |
| 1312 0x0041, |
| 1313 0x00C6, |
| 1314 0x0043, |
| 1315 /* U+00C8 */ 0x0045, |
| 1316 0x0045, |
| 1317 0x0045, |
| 1318 0x0045, |
| 1319 0x0049, |
| 1320 0x0049, |
| 1321 0x0049, |
| 1322 0x0049, |
| 1323 /* U+00D0 */ 0x00D0, |
| 1324 0x004E, |
| 1325 0x004F, |
| 1326 0x004F, |
| 1327 0x004F, |
| 1328 0x004F, |
| 1329 0x004F, |
| 1330 0x00D7, |
| 1331 /* U+00D8 */ 0x004F, |
| 1332 0x0055, |
| 1333 0x0055, |
| 1334 0x0055, |
| 1335 0x0055, |
| 1336 0x0059, |
| 1337 0x00DE, |
| 1338 0x0073, |
| 1339 // U+00D8: Manually changed from 00D8 to 004F |
| 1340 // TODO: Check if it's really acceptable to consider Ø a diacritical |
| 1341 // variant of O |
| 1342 // U+00DF: Manually changed from 00DF to 0073 |
| 1343 /* U+00E0 */ 0x0061, |
| 1344 0x0061, |
| 1345 0x0061, |
| 1346 0x0061, |
| 1347 0x0061, |
| 1348 0x0061, |
| 1349 0x00E6, |
| 1350 0x0063, |
| 1351 /* U+00E8 */ 0x0065, |
| 1352 0x0065, |
| 1353 0x0065, |
| 1354 0x0065, |
| 1355 0x0069, |
| 1356 0x0069, |
| 1357 0x0069, |
| 1358 0x0069, |
| 1359 /* U+00F0 */ 0x00F0, |
| 1360 0x006E, |
| 1361 0x006F, |
| 1362 0x006F, |
| 1363 0x006F, |
| 1364 0x006F, |
| 1365 0x006F, |
| 1366 0x00F7, |
| 1367 /* U+00F8 */ 0x006F, |
| 1368 0x0075, |
| 1369 0x0075, |
| 1370 0x0075, |
| 1371 0x0075, |
| 1372 0x0079, |
| 1373 0x00FE, |
| 1374 0x0079, |
| 1375 // U+00F8: Manually changed from 00F8 to 006F |
| 1376 // TODO: Check if it's really acceptable to consider ø a diacritical |
| 1377 // variant of o |
| 1378 /* U+0100 */ 0x0041, |
| 1379 0x0061, |
| 1380 0x0041, |
| 1381 0x0061, |
| 1382 0x0041, |
| 1383 0x0061, |
| 1384 0x0043, |
| 1385 0x0063, |
| 1386 /* U+0108 */ 0x0043, |
| 1387 0x0063, |
| 1388 0x0043, |
| 1389 0x0063, |
| 1390 0x0043, |
| 1391 0x0063, |
| 1392 0x0044, |
| 1393 0x0064, |
| 1394 /* U+0110 */ 0x0110, |
| 1395 0x0111, |
| 1396 0x0045, |
| 1397 0x0065, |
| 1398 0x0045, |
| 1399 0x0065, |
| 1400 0x0045, |
| 1401 0x0065, |
| 1402 /* U+0118 */ 0x0045, |
| 1403 0x0065, |
| 1404 0x0045, |
| 1405 0x0065, |
| 1406 0x0047, |
| 1407 0x0067, |
| 1408 0x0047, |
| 1409 0x0067, |
| 1410 /* U+0120 */ 0x0047, |
| 1411 0x0067, |
| 1412 0x0047, |
| 1413 0x0067, |
| 1414 0x0048, |
| 1415 0x0068, |
| 1416 0x0126, |
| 1417 0x0127, |
| 1418 /* U+0128 */ 0x0049, |
| 1419 0x0069, |
| 1420 0x0049, |
| 1421 0x0069, |
| 1422 0x0049, |
| 1423 0x0069, |
| 1424 0x0049, |
| 1425 0x0069, |
| 1426 // U+0131: Manually changed from 0131 to 0049 |
| 1427 /* U+0130 */ 0x0049, |
| 1428 0x0049, |
| 1429 0x0049, |
| 1430 0x0069, |
| 1431 0x004A, |
| 1432 0x006A, |
| 1433 0x004B, |
| 1434 0x006B, |
| 1435 /* U+0138 */ 0x0138, |
| 1436 0x004C, |
| 1437 0x006C, |
| 1438 0x004C, |
| 1439 0x006C, |
| 1440 0x004C, |
| 1441 0x006C, |
| 1442 0x004C, |
| 1443 /* U+0140 */ 0x006C, |
| 1444 0x004C, |
| 1445 0x006C, |
| 1446 0x004E, |
| 1447 0x006E, |
| 1448 0x004E, |
| 1449 0x006E, |
| 1450 0x004E, |
| 1451 // U+0141: Manually changed from 0141 to 004C |
| 1452 // U+0142: Manually changed from 0142 to 006C |
| 1453 /* U+0148 */ 0x006E, |
| 1454 0x02BC, |
| 1455 0x014A, |
| 1456 0x014B, |
| 1457 0x004F, |
| 1458 0x006F, |
| 1459 0x004F, |
| 1460 0x006F, |
| 1461 /* U+0150 */ 0x004F, |
| 1462 0x006F, |
| 1463 0x0152, |
| 1464 0x0153, |
| 1465 0x0052, |
| 1466 0x0072, |
| 1467 0x0052, |
| 1468 0x0072, |
| 1469 /* U+0158 */ 0x0052, |
| 1470 0x0072, |
| 1471 0x0053, |
| 1472 0x0073, |
| 1473 0x0053, |
| 1474 0x0073, |
| 1475 0x0053, |
| 1476 0x0073, |
| 1477 /* U+0160 */ 0x0053, |
| 1478 0x0073, |
| 1479 0x0054, |
| 1480 0x0074, |
| 1481 0x0054, |
| 1482 0x0074, |
| 1483 0x0166, |
| 1484 0x0167, |
| 1485 /* U+0168 */ 0x0055, |
| 1486 0x0075, |
| 1487 0x0055, |
| 1488 0x0075, |
| 1489 0x0055, |
| 1490 0x0075, |
| 1491 0x0055, |
| 1492 0x0075, |
| 1493 /* U+0170 */ 0x0055, |
| 1494 0x0075, |
| 1495 0x0055, |
| 1496 0x0075, |
| 1497 0x0057, |
| 1498 0x0077, |
| 1499 0x0059, |
| 1500 0x0079, |
| 1501 /* U+0178 */ 0x0059, |
| 1502 0x005A, |
| 1503 0x007A, |
| 1504 0x005A, |
| 1505 0x007A, |
| 1506 0x005A, |
| 1507 0x007A, |
| 1508 0x0073, |
| 1509 /* U+0180 */ 0x0180, |
| 1510 0x0181, |
| 1511 0x0182, |
| 1512 0x0183, |
| 1513 0x0184, |
| 1514 0x0185, |
| 1515 0x0186, |
| 1516 0x0187, |
| 1517 /* U+0188 */ 0x0188, |
| 1518 0x0189, |
| 1519 0x018A, |
| 1520 0x018B, |
| 1521 0x018C, |
| 1522 0x018D, |
| 1523 0x018E, |
| 1524 0x018F, |
| 1525 /* U+0190 */ 0x0190, |
| 1526 0x0191, |
| 1527 0x0192, |
| 1528 0x0193, |
| 1529 0x0194, |
| 1530 0x0195, |
| 1531 0x0196, |
| 1532 0x0197, |
| 1533 /* U+0198 */ 0x0198, |
| 1534 0x0199, |
| 1535 0x019A, |
| 1536 0x019B, |
| 1537 0x019C, |
| 1538 0x019D, |
| 1539 0x019E, |
| 1540 0x019F, |
| 1541 /* U+01A0 */ 0x004F, |
| 1542 0x006F, |
| 1543 0x01A2, |
| 1544 0x01A3, |
| 1545 0x01A4, |
| 1546 0x01A5, |
| 1547 0x01A6, |
| 1548 0x01A7, |
| 1549 /* U+01A8 */ 0x01A8, |
| 1550 0x01A9, |
| 1551 0x01AA, |
| 1552 0x01AB, |
| 1553 0x01AC, |
| 1554 0x01AD, |
| 1555 0x01AE, |
| 1556 0x0055, |
| 1557 /* U+01B0 */ 0x0075, |
| 1558 0x01B1, |
| 1559 0x01B2, |
| 1560 0x01B3, |
| 1561 0x01B4, |
| 1562 0x01B5, |
| 1563 0x01B6, |
| 1564 0x01B7, |
| 1565 /* U+01B8 */ 0x01B8, |
| 1566 0x01B9, |
| 1567 0x01BA, |
| 1568 0x01BB, |
| 1569 0x01BC, |
| 1570 0x01BD, |
| 1571 0x01BE, |
| 1572 0x01BF, |
| 1573 /* U+01C0 */ 0x01C0, |
| 1574 0x01C1, |
| 1575 0x01C2, |
| 1576 0x01C3, |
| 1577 0x0044, |
| 1578 0x0044, |
| 1579 0x0064, |
| 1580 0x004C, |
| 1581 /* U+01C8 */ 0x004C, |
| 1582 0x006C, |
| 1583 0x004E, |
| 1584 0x004E, |
| 1585 0x006E, |
| 1586 0x0041, |
| 1587 0x0061, |
| 1588 0x0049, |
| 1589 /* U+01D0 */ 0x0069, |
| 1590 0x004F, |
| 1591 0x006F, |
| 1592 0x0055, |
| 1593 0x0075, |
| 1594 0x0055, |
| 1595 0x0075, |
| 1596 0x0055, |
| 1597 // U+01D5: Manually changed from 00DC to 0055 |
| 1598 // U+01D6: Manually changed from 00FC to 0075 |
| 1599 // U+01D7: Manually changed from 00DC to 0055 |
| 1600 /* U+01D8 */ 0x0075, |
| 1601 0x0055, |
| 1602 0x0075, |
| 1603 0x0055, |
| 1604 0x0075, |
| 1605 0x01DD, |
| 1606 0x0041, |
| 1607 0x0061, |
| 1608 // U+01D8: Manually changed from 00FC to 0075 |
| 1609 // U+01D9: Manually changed from 00DC to 0055 |
| 1610 // U+01DA: Manually changed from 00FC to 0075 |
| 1611 // U+01DB: Manually changed from 00DC to 0055 |
| 1612 // U+01DC: Manually changed from 00FC to 0075 |
| 1613 // U+01DE: Manually changed from 00C4 to 0041 |
| 1614 // U+01DF: Manually changed from 00E4 to 0061 |
| 1615 /* U+01E0 */ 0x0041, |
| 1616 0x0061, |
| 1617 0x00C6, |
| 1618 0x00E6, |
| 1619 0x01E4, |
| 1620 0x01E5, |
| 1621 0x0047, |
| 1622 0x0067, |
| 1623 // U+01E0: Manually changed from 0226 to 0041 |
| 1624 // U+01E1: Manually changed from 0227 to 0061 |
| 1625 /* U+01E8 */ 0x004B, |
| 1626 0x006B, |
| 1627 0x004F, |
| 1628 0x006F, |
| 1629 0x004F, |
| 1630 0x006F, |
| 1631 0x01B7, |
| 1632 0x0292, |
| 1633 // U+01EC: Manually changed from 01EA to 004F |
| 1634 // U+01ED: Manually changed from 01EB to 006F |
| 1635 /* U+01F0 */ 0x006A, |
| 1636 0x0044, |
| 1637 0x0044, |
| 1638 0x0064, |
| 1639 0x0047, |
| 1640 0x0067, |
| 1641 0x01F6, |
| 1642 0x01F7, |
| 1643 /* U+01F8 */ 0x004E, |
| 1644 0x006E, |
| 1645 0x0041, |
| 1646 0x0061, |
| 1647 0x00C6, |
| 1648 0x00E6, |
| 1649 0x004F, |
| 1650 0x006F, |
| 1651 // U+01FA: Manually changed from 00C5 to 0041 |
| 1652 // U+01FB: Manually changed from 00E5 to 0061 |
| 1653 // U+01FE: Manually changed from 00D8 to 004F |
| 1654 // TODO: Check if it's really acceptable to consider Ø a diacritical |
| 1655 // variant of O |
| 1656 // U+01FF: Manually changed from 00F8 to 006F |
| 1657 // TODO: Check if it's really acceptable to consider ø a diacritical |
| 1658 // variant of o |
| 1659 /* U+0200 */ 0x0041, |
| 1660 0x0061, |
| 1661 0x0041, |
| 1662 0x0061, |
| 1663 0x0045, |
| 1664 0x0065, |
| 1665 0x0045, |
| 1666 0x0065, |
| 1667 /* U+0208 */ 0x0049, |
| 1668 0x0069, |
| 1669 0x0049, |
| 1670 0x0069, |
| 1671 0x004F, |
| 1672 0x006F, |
| 1673 0x004F, |
| 1674 0x006F, |
| 1675 /* U+0210 */ 0x0052, |
| 1676 0x0072, |
| 1677 0x0052, |
| 1678 0x0072, |
| 1679 0x0055, |
| 1680 0x0075, |
| 1681 0x0055, |
| 1682 0x0075, |
| 1683 /* U+0218 */ 0x0053, |
| 1684 0x0073, |
| 1685 0x0054, |
| 1686 0x0074, |
| 1687 0x021C, |
| 1688 0x021D, |
| 1689 0x0048, |
| 1690 0x0068, |
| 1691 /* U+0220 */ 0x0220, |
| 1692 0x0221, |
| 1693 0x0222, |
| 1694 0x0223, |
| 1695 0x0224, |
| 1696 0x0225, |
| 1697 0x0041, |
| 1698 0x0061, |
| 1699 /* U+0228 */ 0x0045, |
| 1700 0x0065, |
| 1701 0x004F, |
| 1702 0x006F, |
| 1703 0x004F, |
| 1704 0x006F, |
| 1705 0x004F, |
| 1706 0x006F, |
| 1707 // U+022A: Manually changed from 00D6 to 004F |
| 1708 // U+022B: Manually changed from 00F6 to 006F |
| 1709 // U+022C: Manually changed from 00D5 to 004F |
| 1710 // U+022D: Manually changed from 00F5 to 006F |
| 1711 /* U+0230 */ 0x004F, |
| 1712 0x006F, |
| 1713 0x0059, |
| 1714 0x0079, |
| 1715 0x0234, |
| 1716 0x0235, |
| 1717 0x0236, |
| 1718 0x0237, |
| 1719 // U+0230: Manually changed from 022E to 004F |
| 1720 // U+0231: Manually changed from 022F to 006F |
| 1721 /* U+0238 */ 0x0238, |
| 1722 0x0239, |
| 1723 0x023A, |
| 1724 0x023B, |
| 1725 0x023C, |
| 1726 0x023D, |
| 1727 0x023E, |
| 1728 0x023F, |
| 1729 /* U+0240 */ 0x0240, |
| 1730 0x0241, |
| 1731 0x0242, |
| 1732 0x0243, |
| 1733 0x0244, |
| 1734 0x0245, |
| 1735 0x0246, |
| 1736 0x0247, |
| 1737 /* U+0248 */ 0x0248, |
| 1738 0x0249, |
| 1739 0x024A, |
| 1740 0x024B, |
| 1741 0x024C, |
| 1742 0x024D, |
| 1743 0x024E, |
| 1744 0x024F, |
| 1745 /* U+0250 */ 0x0250, |
| 1746 0x0251, |
| 1747 0x0252, |
| 1748 0x0253, |
| 1749 0x0254, |
| 1750 0x0255, |
| 1751 0x0256, |
| 1752 0x0257, |
| 1753 /* U+0258 */ 0x0258, |
| 1754 0x0259, |
| 1755 0x025A, |
| 1756 0x025B, |
| 1757 0x025C, |
| 1758 0x025D, |
| 1759 0x025E, |
| 1760 0x025F, |
| 1761 /* U+0260 */ 0x0260, |
| 1762 0x0261, |
| 1763 0x0262, |
| 1764 0x0263, |
| 1765 0x0264, |
| 1766 0x0265, |
| 1767 0x0266, |
| 1768 0x0267, |
| 1769 /* U+0268 */ 0x0268, |
| 1770 0x0269, |
| 1771 0x026A, |
| 1772 0x026B, |
| 1773 0x026C, |
| 1774 0x026D, |
| 1775 0x026E, |
| 1776 0x026F, |
| 1777 /* U+0270 */ 0x0270, |
| 1778 0x0271, |
| 1779 0x0272, |
| 1780 0x0273, |
| 1781 0x0274, |
| 1782 0x0275, |
| 1783 0x0276, |
| 1784 0x0277, |
| 1785 /* U+0278 */ 0x0278, |
| 1786 0x0279, |
| 1787 0x027A, |
| 1788 0x027B, |
| 1789 0x027C, |
| 1790 0x027D, |
| 1791 0x027E, |
| 1792 0x027F, |
| 1793 /* U+0280 */ 0x0280, |
| 1794 0x0281, |
| 1795 0x0282, |
| 1796 0x0283, |
| 1797 0x0284, |
| 1798 0x0285, |
| 1799 0x0286, |
| 1800 0x0287, |
| 1801 /* U+0288 */ 0x0288, |
| 1802 0x0289, |
| 1803 0x028A, |
| 1804 0x028B, |
| 1805 0x028C, |
| 1806 0x028D, |
| 1807 0x028E, |
| 1808 0x028F, |
| 1809 /* U+0290 */ 0x0290, |
| 1810 0x0291, |
| 1811 0x0292, |
| 1812 0x0293, |
| 1813 0x0294, |
| 1814 0x0295, |
| 1815 0x0296, |
| 1816 0x0297, |
| 1817 /* U+0298 */ 0x0298, |
| 1818 0x0299, |
| 1819 0x029A, |
| 1820 0x029B, |
| 1821 0x029C, |
| 1822 0x029D, |
| 1823 0x029E, |
| 1824 0x029F, |
| 1825 /* U+02A0 */ 0x02A0, |
| 1826 0x02A1, |
| 1827 0x02A2, |
| 1828 0x02A3, |
| 1829 0x02A4, |
| 1830 0x02A5, |
| 1831 0x02A6, |
| 1832 0x02A7, |
| 1833 /* U+02A8 */ 0x02A8, |
| 1834 0x02A9, |
| 1835 0x02AA, |
| 1836 0x02AB, |
| 1837 0x02AC, |
| 1838 0x02AD, |
| 1839 0x02AE, |
| 1840 0x02AF, |
| 1841 /* U+02B0 */ 0x0068, |
| 1842 0x0266, |
| 1843 0x006A, |
| 1844 0x0072, |
| 1845 0x0279, |
| 1846 0x027B, |
| 1847 0x0281, |
| 1848 0x0077, |
| 1849 /* U+02B8 */ 0x0079, |
| 1850 0x02B9, |
| 1851 0x02BA, |
| 1852 0x02BB, |
| 1853 0x02BC, |
| 1854 0x02BD, |
| 1855 0x02BE, |
| 1856 0x02BF, |
| 1857 /* U+02C0 */ 0x02C0, |
| 1858 0x02C1, |
| 1859 0x02C2, |
| 1860 0x02C3, |
| 1861 0x02C4, |
| 1862 0x02C5, |
| 1863 0x02C6, |
| 1864 0x02C7, |
| 1865 /* U+02C8 */ 0x02C8, |
| 1866 0x02C9, |
| 1867 0x02CA, |
| 1868 0x02CB, |
| 1869 0x02CC, |
| 1870 0x02CD, |
| 1871 0x02CE, |
| 1872 0x02CF, |
| 1873 /* U+02D0 */ 0x02D0, |
| 1874 0x02D1, |
| 1875 0x02D2, |
| 1876 0x02D3, |
| 1877 0x02D4, |
| 1878 0x02D5, |
| 1879 0x02D6, |
| 1880 0x02D7, |
| 1881 /* U+02D8 */ 0x0020, |
| 1882 0x0020, |
| 1883 0x0020, |
| 1884 0x0020, |
| 1885 0x0020, |
| 1886 0x0020, |
| 1887 0x02DE, |
| 1888 0x02DF, |
| 1889 /* U+02E0 */ 0x0263, |
| 1890 0x006C, |
| 1891 0x0073, |
| 1892 0x0078, |
| 1893 0x0295, |
| 1894 0x02E5, |
| 1895 0x02E6, |
| 1896 0x02E7, |
| 1897 /* U+02E8 */ 0x02E8, |
| 1898 0x02E9, |
| 1899 0x02EA, |
| 1900 0x02EB, |
| 1901 0x02EC, |
| 1902 0x02ED, |
| 1903 0x02EE, |
| 1904 0x02EF, |
| 1905 /* U+02F0 */ 0x02F0, |
| 1906 0x02F1, |
| 1907 0x02F2, |
| 1908 0x02F3, |
| 1909 0x02F4, |
| 1910 0x02F5, |
| 1911 0x02F6, |
| 1912 0x02F7, |
| 1913 /* U+02F8 */ 0x02F8, |
| 1914 0x02F9, |
| 1915 0x02FA, |
| 1916 0x02FB, |
| 1917 0x02FC, |
| 1918 0x02FD, |
| 1919 0x02FE, |
| 1920 0x02FF, |
| 1921 /* U+0300 */ 0x0300, |
| 1922 0x0301, |
| 1923 0x0302, |
| 1924 0x0303, |
| 1925 0x0304, |
| 1926 0x0305, |
| 1927 0x0306, |
| 1928 0x0307, |
| 1929 /* U+0308 */ 0x0308, |
| 1930 0x0309, |
| 1931 0x030A, |
| 1932 0x030B, |
| 1933 0x030C, |
| 1934 0x030D, |
| 1935 0x030E, |
| 1936 0x030F, |
| 1937 /* U+0310 */ 0x0310, |
| 1938 0x0311, |
| 1939 0x0312, |
| 1940 0x0313, |
| 1941 0x0314, |
| 1942 0x0315, |
| 1943 0x0316, |
| 1944 0x0317, |
| 1945 /* U+0318 */ 0x0318, |
| 1946 0x0319, |
| 1947 0x031A, |
| 1948 0x031B, |
| 1949 0x031C, |
| 1950 0x031D, |
| 1951 0x031E, |
| 1952 0x031F, |
| 1953 /* U+0320 */ 0x0320, |
| 1954 0x0321, |
| 1955 0x0322, |
| 1956 0x0323, |
| 1957 0x0324, |
| 1958 0x0325, |
| 1959 0x0326, |
| 1960 0x0327, |
| 1961 /* U+0328 */ 0x0328, |
| 1962 0x0329, |
| 1963 0x032A, |
| 1964 0x032B, |
| 1965 0x032C, |
| 1966 0x032D, |
| 1967 0x032E, |
| 1968 0x032F, |
| 1969 /* U+0330 */ 0x0330, |
| 1970 0x0331, |
| 1971 0x0332, |
| 1972 0x0333, |
| 1973 0x0334, |
| 1974 0x0335, |
| 1975 0x0336, |
| 1976 0x0337, |
| 1977 /* U+0338 */ 0x0338, |
| 1978 0x0339, |
| 1979 0x033A, |
| 1980 0x033B, |
| 1981 0x033C, |
| 1982 0x033D, |
| 1983 0x033E, |
| 1984 0x033F, |
| 1985 /* U+0340 */ 0x0300, |
| 1986 0x0301, |
| 1987 0x0342, |
| 1988 0x0313, |
| 1989 0x0308, |
| 1990 0x0345, |
| 1991 0x0346, |
| 1992 0x0347, |
| 1993 /* U+0348 */ 0x0348, |
| 1994 0x0349, |
| 1995 0x034A, |
| 1996 0x034B, |
| 1997 0x034C, |
| 1998 0x034D, |
| 1999 0x034E, |
| 2000 0x034F, |
| 2001 /* U+0350 */ 0x0350, |
| 2002 0x0351, |
| 2003 0x0352, |
| 2004 0x0353, |
| 2005 0x0354, |
| 2006 0x0355, |
| 2007 0x0356, |
| 2008 0x0357, |
| 2009 /* U+0358 */ 0x0358, |
| 2010 0x0359, |
| 2011 0x035A, |
| 2012 0x035B, |
| 2013 0x035C, |
| 2014 0x035D, |
| 2015 0x035E, |
| 2016 0x035F, |
| 2017 /* U+0360 */ 0x0360, |
| 2018 0x0361, |
| 2019 0x0362, |
| 2020 0x0363, |
| 2021 0x0364, |
| 2022 0x0365, |
| 2023 0x0366, |
| 2024 0x0367, |
| 2025 /* U+0368 */ 0x0368, |
| 2026 0x0369, |
| 2027 0x036A, |
| 2028 0x036B, |
| 2029 0x036C, |
| 2030 0x036D, |
| 2031 0x036E, |
| 2032 0x036F, |
| 2033 /* U+0370 */ 0x0370, |
| 2034 0x0371, |
| 2035 0x0372, |
| 2036 0x0373, |
| 2037 0x02B9, |
| 2038 0x0375, |
| 2039 0x0376, |
| 2040 0x0377, |
| 2041 /* U+0378 */ 0x0378, |
| 2042 0x0379, |
| 2043 0x0020, |
| 2044 0x037B, |
| 2045 0x037C, |
| 2046 0x037D, |
| 2047 0x003B, |
| 2048 0x037F, |
| 2049 /* U+0380 */ 0x0380, |
| 2050 0x0381, |
| 2051 0x0382, |
| 2052 0x0383, |
| 2053 0x0020, |
| 2054 0x00A8, |
| 2055 0x0391, |
| 2056 0x00B7, |
| 2057 /* U+0388 */ 0x0395, |
| 2058 0x0397, |
| 2059 0x0399, |
| 2060 0x038B, |
| 2061 0x039F, |
| 2062 0x038D, |
| 2063 0x03A5, |
| 2064 0x03A9, |
| 2065 /* U+0390 */ 0x03CA, |
| 2066 0x0391, |
| 2067 0x0392, |
| 2068 0x0393, |
| 2069 0x0394, |
| 2070 0x0395, |
| 2071 0x0396, |
| 2072 0x0397, |
| 2073 /* U+0398 */ 0x0398, |
| 2074 0x0399, |
| 2075 0x039A, |
| 2076 0x039B, |
| 2077 0x039C, |
| 2078 0x039D, |
| 2079 0x039E, |
| 2080 0x039F, |
| 2081 /* U+03A0 */ 0x03A0, |
| 2082 0x03A1, |
| 2083 0x03A2, |
| 2084 0x03A3, |
| 2085 0x03A4, |
| 2086 0x03A5, |
| 2087 0x03A6, |
| 2088 0x03A7, |
| 2089 /* U+03A8 */ 0x03A8, |
| 2090 0x03A9, |
| 2091 0x0399, |
| 2092 0x03A5, |
| 2093 0x03B1, |
| 2094 0x03B5, |
| 2095 0x03B7, |
| 2096 0x03B9, |
| 2097 /* U+03B0 */ 0x03CB, |
| 2098 0x03B1, |
| 2099 0x03B2, |
| 2100 0x03B3, |
| 2101 0x03B4, |
| 2102 0x03B5, |
| 2103 0x03B6, |
| 2104 0x03B7, |
| 2105 /* U+03B8 */ 0x03B8, |
| 2106 0x03B9, |
| 2107 0x03BA, |
| 2108 0x03BB, |
| 2109 0x03BC, |
| 2110 0x03BD, |
| 2111 0x03BE, |
| 2112 0x03BF, |
| 2113 /* U+03C0 */ 0x03C0, |
| 2114 0x03C1, |
| 2115 0x03C2, |
| 2116 0x03C3, |
| 2117 0x03C4, |
| 2118 0x03C5, |
| 2119 0x03C6, |
| 2120 0x03C7, |
| 2121 /* U+03C8 */ 0x03C8, |
| 2122 0x03C9, |
| 2123 0x03B9, |
| 2124 0x03C5, |
| 2125 0x03BF, |
| 2126 0x03C5, |
| 2127 0x03C9, |
| 2128 0x03CF, |
| 2129 /* U+03D0 */ 0x03B2, |
| 2130 0x03B8, |
| 2131 0x03A5, |
| 2132 0x03D2, |
| 2133 0x03D2, |
| 2134 0x03C6, |
| 2135 0x03C0, |
| 2136 0x03D7, |
| 2137 /* U+03D8 */ 0x03D8, |
| 2138 0x03D9, |
| 2139 0x03DA, |
| 2140 0x03DB, |
| 2141 0x03DC, |
| 2142 0x03DD, |
| 2143 0x03DE, |
| 2144 0x03DF, |
| 2145 /* U+03E0 */ 0x03E0, |
| 2146 0x03E1, |
| 2147 0x03E2, |
| 2148 0x03E3, |
| 2149 0x03E4, |
| 2150 0x03E5, |
| 2151 0x03E6, |
| 2152 0x03E7, |
| 2153 /* U+03E8 */ 0x03E8, |
| 2154 0x03E9, |
| 2155 0x03EA, |
| 2156 0x03EB, |
| 2157 0x03EC, |
| 2158 0x03ED, |
| 2159 0x03EE, |
| 2160 0x03EF, |
| 2161 /* U+03F0 */ 0x03BA, |
| 2162 0x03C1, |
| 2163 0x03C2, |
| 2164 0x03F3, |
| 2165 0x0398, |
| 2166 0x03B5, |
| 2167 0x03F6, |
| 2168 0x03F7, |
| 2169 /* U+03F8 */ 0x03F8, |
| 2170 0x03A3, |
| 2171 0x03FA, |
| 2172 0x03FB, |
| 2173 0x03FC, |
| 2174 0x03FD, |
| 2175 0x03FE, |
| 2176 0x03FF, |
| 2177 /* U+0400 */ 0x0415, |
| 2178 0x0415, |
| 2179 0x0402, |
| 2180 0x0413, |
| 2181 0x0404, |
| 2182 0x0405, |
| 2183 0x0406, |
| 2184 0x0406, |
| 2185 /* U+0408 */ 0x0408, |
| 2186 0x0409, |
| 2187 0x040A, |
| 2188 0x040B, |
| 2189 0x041A, |
| 2190 0x0418, |
| 2191 0x0423, |
| 2192 0x040F, |
| 2193 /* U+0410 */ 0x0410, |
| 2194 0x0411, |
| 2195 0x0412, |
| 2196 0x0413, |
| 2197 0x0414, |
| 2198 0x0415, |
| 2199 0x0416, |
| 2200 0x0417, |
| 2201 /* U+0418 */ 0x0418, |
| 2202 0x0419, |
| 2203 0x041A, |
| 2204 0x041B, |
| 2205 0x041C, |
| 2206 0x041D, |
| 2207 0x041E, |
| 2208 0x041F, |
| 2209 // U+0419: Manually changed from 0418 to 0419 |
| 2210 /* U+0420 */ 0x0420, |
| 2211 0x0421, |
| 2212 0x0422, |
| 2213 0x0423, |
| 2214 0x0424, |
| 2215 0x0425, |
| 2216 0x0426, |
| 2217 0x0427, |
| 2218 /* U+0428 */ 0x0428, |
| 2219 0x0429, |
| 2220 0x042C, |
| 2221 0x042B, |
| 2222 0x042C, |
| 2223 0x042D, |
| 2224 0x042E, |
| 2225 0x042F, |
| 2226 // U+042A: Manually changed from 042A to 042C |
| 2227 /* U+0430 */ 0x0430, |
| 2228 0x0431, |
| 2229 0x0432, |
| 2230 0x0433, |
| 2231 0x0434, |
| 2232 0x0435, |
| 2233 0x0436, |
| 2234 0x0437, |
| 2235 /* U+0438 */ 0x0438, |
| 2236 0x0439, |
| 2237 0x043A, |
| 2238 0x043B, |
| 2239 0x043C, |
| 2240 0x043D, |
| 2241 0x043E, |
| 2242 0x043F, |
| 2243 // U+0439: Manually changed from 0438 to 0439 |
| 2244 /* U+0440 */ 0x0440, |
| 2245 0x0441, |
| 2246 0x0442, |
| 2247 0x0443, |
| 2248 0x0444, |
| 2249 0x0445, |
| 2250 0x0446, |
| 2251 0x0447, |
| 2252 /* U+0448 */ 0x0448, |
| 2253 0x0449, |
| 2254 0x044C, |
| 2255 0x044B, |
| 2256 0x044C, |
| 2257 0x044D, |
| 2258 0x044E, |
| 2259 0x044F, |
| 2260 // U+044A: Manually changed from 044A to 044C |
| 2261 /* U+0450 */ 0x0435, |
| 2262 0x0435, |
| 2263 0x0452, |
| 2264 0x0433, |
| 2265 0x0454, |
| 2266 0x0455, |
| 2267 0x0456, |
| 2268 0x0456, |
| 2269 /* U+0458 */ 0x0458, |
| 2270 0x0459, |
| 2271 0x045A, |
| 2272 0x045B, |
| 2273 0x043A, |
| 2274 0x0438, |
| 2275 0x0443, |
| 2276 0x045F, |
| 2277 /* U+0460 */ 0x0460, |
| 2278 0x0461, |
| 2279 0x0462, |
| 2280 0x0463, |
| 2281 0x0464, |
| 2282 0x0465, |
| 2283 0x0466, |
| 2284 0x0467, |
| 2285 /* U+0468 */ 0x0468, |
| 2286 0x0469, |
| 2287 0x046A, |
| 2288 0x046B, |
| 2289 0x046C, |
| 2290 0x046D, |
| 2291 0x046E, |
| 2292 0x046F, |
| 2293 /* U+0470 */ 0x0470, |
| 2294 0x0471, |
| 2295 0x0472, |
| 2296 0x0473, |
| 2297 0x0474, |
| 2298 0x0475, |
| 2299 0x0474, |
| 2300 0x0475, |
| 2301 /* U+0478 */ 0x0478, |
| 2302 0x0479, |
| 2303 0x047A, |
| 2304 0x047B, |
| 2305 0x047C, |
| 2306 0x047D, |
| 2307 0x047E, |
| 2308 0x047F, |
| 2309 /* U+0480 */ 0x0480, |
| 2310 0x0481, |
| 2311 0x0482, |
| 2312 0x0483, |
| 2313 0x0484, |
| 2314 0x0485, |
| 2315 0x0486, |
| 2316 0x0487, |
| 2317 /* U+0488 */ 0x0488, |
| 2318 0x0489, |
| 2319 0x048A, |
| 2320 0x048B, |
| 2321 0x048C, |
| 2322 0x048D, |
| 2323 0x048E, |
| 2324 0x048F, |
| 2325 /* U+0490 */ 0x0490, |
| 2326 0x0491, |
| 2327 0x0492, |
| 2328 0x0493, |
| 2329 0x0494, |
| 2330 0x0495, |
| 2331 0x0496, |
| 2332 0x0497, |
| 2333 /* U+0498 */ 0x0498, |
| 2334 0x0499, |
| 2335 0x049A, |
| 2336 0x049B, |
| 2337 0x049C, |
| 2338 0x049D, |
| 2339 0x049E, |
| 2340 0x049F, |
| 2341 /* U+04A0 */ 0x04A0, |
| 2342 0x04A1, |
| 2343 0x04A2, |
| 2344 0x04A3, |
| 2345 0x04A4, |
| 2346 0x04A5, |
| 2347 0x04A6, |
| 2348 0x04A7, |
| 2349 /* U+04A8 */ 0x04A8, |
| 2350 0x04A9, |
| 2351 0x04AA, |
| 2352 0x04AB, |
| 2353 0x04AC, |
| 2354 0x04AD, |
| 2355 0x04AE, |
| 2356 0x04AF, |
| 2357 /* U+04B0 */ 0x04B0, |
| 2358 0x04B1, |
| 2359 0x04B2, |
| 2360 0x04B3, |
| 2361 0x04B4, |
| 2362 0x04B5, |
| 2363 0x04B6, |
| 2364 0x04B7, |
| 2365 /* U+04B8 */ 0x04B8, |
| 2366 0x04B9, |
| 2367 0x04BA, |
| 2368 0x04BB, |
| 2369 0x04BC, |
| 2370 0x04BD, |
| 2371 0x04BE, |
| 2372 0x04BF, |
| 2373 /* U+04C0 */ 0x04C0, |
| 2374 0x0416, |
| 2375 0x0436, |
| 2376 0x04C3, |
| 2377 0x04C4, |
| 2378 0x04C5, |
| 2379 0x04C6, |
| 2380 0x04C7, |
| 2381 /* U+04C8 */ 0x04C8, |
| 2382 0x04C9, |
| 2383 0x04CA, |
| 2384 0x04CB, |
| 2385 0x04CC, |
| 2386 0x04CD, |
| 2387 0x04CE, |
| 2388 0x04CF, |
| 2389 /* U+04D0 */ 0x0410, |
| 2390 0x0430, |
| 2391 0x0410, |
| 2392 0x0430, |
| 2393 0x04D4, |
| 2394 0x04D5, |
| 2395 0x0415, |
| 2396 0x0435, |
| 2397 /* U+04D8 */ 0x04D8, |
| 2398 0x04D9, |
| 2399 0x04D8, |
| 2400 0x04D9, |
| 2401 0x0416, |
| 2402 0x0436, |
| 2403 0x0417, |
| 2404 0x0437, |
| 2405 /* U+04E0 */ 0x04E0, |
| 2406 0x04E1, |
| 2407 0x0418, |
| 2408 0x0438, |
| 2409 0x0418, |
| 2410 0x0438, |
| 2411 0x041E, |
| 2412 0x043E, |
| 2413 /* U+04E8 */ 0x04E8, |
| 2414 0x04E9, |
| 2415 0x04E8, |
| 2416 0x04E9, |
| 2417 0x042D, |
| 2418 0x044D, |
| 2419 0x0423, |
| 2420 0x0443, |
| 2421 /* U+04F0 */ 0x0423, |
| 2422 0x0443, |
| 2423 0x0423, |
| 2424 0x0443, |
| 2425 0x0427, |
| 2426 0x0447, |
| 2427 0x04F6, |
| 2428 0x04F7, |
| 2429 /* U+04F8 */ 0x042B, |
| 2430 0x044B, |
| 2431 0x04FA, |
| 2432 0x04FB, |
| 2433 0x04FC, |
| 2434 0x04FD, |
| 2435 0x04FE, |
| 2436 0x04FF, |
| 2437 }; |
| 2438 |
| 2439 /* static */ const std::vector<int> CharUtils::EMPTY_STRING(1 /* size */, |
| 2440 '\0' /* value */); |
| 2441 } // namespace latinime |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 1247903003:
Add spellcheck and word suggestion to the prediction service (Closed)
Base URL: https://github.com/domokit/mojo.git@master