Home | History | Annotate | Download | only in utils
      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 "utils/char_utils.h"
     18 
     19 #include <cstdlib>
     20 
     21 #include "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) continue;
     52         const unsigned short icu4cLowerC = u_tolower(c);
     53         const unsigned short myLowerC = CharUtils::latin_tolower(c);
     54         if (c != icu4cLowerC) {
     55 #ifdef CONFIRMING_CHAR_UTILS
     56             if (icu4cLowerC != myLowerC) {
     57                 fprintf(stderr, "icu4cLowerC != myLowerC, 0x%04X, 0x%04X\n", icu4cLowerC, myLowerC);
     58             }
     59 #else // CONFIRMING_CHAR_UTILS
     60             printf("0x%04X, 0x%04X\n", c, icu4cLowerC);
     61 #endif // CONFIRMING_CHAR_UTILS
     62         }
     63     }
     64 }
     65 #endif // UPDATING_CHAR_UTILS
     66 /*
     67  * 4. Process the list with UnicodeData.txt
     68  *    (You need UnicodeData.txt in the current directory)
     69  *    $ ./char_utils | sort -u | \
     70  *      perl -e 'open(FH, "UnicodeData.txt"); @buf = <FH>; close(FH); \
     71  *      while(<>){/0x(\w*), 0x(\w*)/; @lines = grep(/^$1/, @buf); @cols = split(/;/, $lines[0]); \
     72  *      print "    { 0x$1, 0x$cols[13] },  // $cols[1]\n";}'
     73  *
     74  * 5. Update the SORTED_CHAR_MAP[] array below with the output above.
     75  *    Then, rebuild with -DCONFIRMING_CHAR_UTILS and confirm the program exits successfully.
     76  *    $ g++ -o char_utils -I.. -DUPDATING_CHAR_UTILS -DCONFIRMING_CHAR_UTILS char_utils.cpp -licuuc
     77  *    $ ./char_utils
     78  *    $
     79  */
     80 static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
     81     { 0x00C0, 0x00E0 },  // LATIN CAPITAL LETTER A WITH GRAVE
     82     { 0x00C1, 0x00E1 },  // LATIN CAPITAL LETTER A WITH ACUTE
     83     { 0x00C2, 0x00E2 },  // LATIN CAPITAL LETTER A WITH CIRCUMFLEX
     84     { 0x00C3, 0x00E3 },  // LATIN CAPITAL LETTER A WITH TILDE
     85     { 0x00C4, 0x00E4 },  // LATIN CAPITAL LETTER A WITH DIAERESIS
     86     { 0x00C5, 0x00E5 },  // LATIN CAPITAL LETTER A WITH RING ABOVE
     87     { 0x00C6, 0x00E6 },  // LATIN CAPITAL LETTER AE
     88     { 0x00C7, 0x00E7 },  // LATIN CAPITAL LETTER C WITH CEDILLA
     89     { 0x00C8, 0x00E8 },  // LATIN CAPITAL LETTER E WITH GRAVE
     90     { 0x00C9, 0x00E9 },  // LATIN CAPITAL LETTER E WITH ACUTE
     91     { 0x00CA, 0x00EA },  // LATIN CAPITAL LETTER E WITH CIRCUMFLEX
     92     { 0x00CB, 0x00EB },  // LATIN CAPITAL LETTER E WITH DIAERESIS
     93     { 0x00CC, 0x00EC },  // LATIN CAPITAL LETTER I WITH GRAVE
     94     { 0x00CD, 0x00ED },  // LATIN CAPITAL LETTER I WITH ACUTE
     95     { 0x00CE, 0x00EE },  // LATIN CAPITAL LETTER I WITH CIRCUMFLEX
     96     { 0x00CF, 0x00EF },  // LATIN CAPITAL LETTER I WITH DIAERESIS
     97     { 0x00D0, 0x00F0 },  // LATIN CAPITAL LETTER ETH
     98     { 0x00D1, 0x00F1 },  // LATIN CAPITAL LETTER N WITH TILDE
     99     { 0x00D2, 0x00F2 },  // LATIN CAPITAL LETTER O WITH GRAVE
    100     { 0x00D3, 0x00F3 },  // LATIN CAPITAL LETTER O WITH ACUTE
    101     { 0x00D4, 0x00F4 },  // LATIN CAPITAL LETTER O WITH CIRCUMFLEX
    102     { 0x00D5, 0x00F5 },  // LATIN CAPITAL LETTER O WITH TILDE
    103     { 0x00D6, 0x00F6 },  // LATIN CAPITAL LETTER O WITH DIAERESIS
    104     { 0x00D8, 0x00F8 },  // LATIN CAPITAL LETTER O WITH STROKE
    105     { 0x00D9, 0x00F9 },  // LATIN CAPITAL LETTER U WITH GRAVE
    106     { 0x00DA, 0x00FA },  // LATIN CAPITAL LETTER U WITH ACUTE
    107     { 0x00DB, 0x00FB },  // LATIN CAPITAL LETTER U WITH CIRCUMFLEX
    108     { 0x00DC, 0x00FC },  // LATIN CAPITAL LETTER U WITH DIAERESIS
    109     { 0x00DD, 0x00FD },  // LATIN CAPITAL LETTER Y WITH ACUTE
    110     { 0x00DE, 0x00FE },  // LATIN CAPITAL LETTER THORN
    111     { 0x0100, 0x0101 },  // LATIN CAPITAL LETTER A WITH MACRON
    112     { 0x0102, 0x0103 },  // LATIN CAPITAL LETTER A WITH BREVE
    113     { 0x0104, 0x0105 },  // LATIN CAPITAL LETTER A WITH OGONEK
    114     { 0x0106, 0x0107 },  // LATIN CAPITAL LETTER C WITH ACUTE
    115     { 0x0108, 0x0109 },  // LATIN CAPITAL LETTER C WITH CIRCUMFLEX
    116     { 0x010A, 0x010B },  // LATIN CAPITAL LETTER C WITH DOT ABOVE
    117     { 0x010C, 0x010D },  // LATIN CAPITAL LETTER C WITH CARON
    118     { 0x010E, 0x010F },  // LATIN CAPITAL LETTER D WITH CARON
    119     { 0x0110, 0x0111 },  // LATIN CAPITAL LETTER D WITH STROKE
    120     { 0x0112, 0x0113 },  // LATIN CAPITAL LETTER E WITH MACRON
    121     { 0x0114, 0x0115 },  // LATIN CAPITAL LETTER E WITH BREVE
    122     { 0x0116, 0x0117 },  // LATIN CAPITAL LETTER E WITH DOT ABOVE
    123     { 0x0118, 0x0119 },  // LATIN CAPITAL LETTER E WITH OGONEK
    124     { 0x011A, 0x011B },  // LATIN CAPITAL LETTER E WITH CARON
    125     { 0x011C, 0x011D },  // LATIN CAPITAL LETTER G WITH CIRCUMFLEX
    126     { 0x011E, 0x011F },  // LATIN CAPITAL LETTER G WITH BREVE
    127     { 0x0120, 0x0121 },  // LATIN CAPITAL LETTER G WITH DOT ABOVE
    128     { 0x0122, 0x0123 },  // LATIN CAPITAL LETTER G WITH CEDILLA
    129     { 0x0124, 0x0125 },  // LATIN CAPITAL LETTER H WITH CIRCUMFLEX
    130     { 0x0126, 0x0127 },  // LATIN CAPITAL LETTER H WITH STROKE
    131     { 0x0128, 0x0129 },  // LATIN CAPITAL LETTER I WITH TILDE
    132     { 0x012A, 0x012B },  // LATIN CAPITAL LETTER I WITH MACRON
    133     { 0x012C, 0x012D },  // LATIN CAPITAL LETTER I WITH BREVE
    134     { 0x012E, 0x012F },  // LATIN CAPITAL LETTER I WITH OGONEK
    135     { 0x0130, 0x0069 },  // LATIN CAPITAL LETTER I WITH DOT ABOVE
    136     { 0x0132, 0x0133 },  // LATIN CAPITAL LIGATURE IJ
    137     { 0x0134, 0x0135 },  // LATIN CAPITAL LETTER J WITH CIRCUMFLEX
    138     { 0x0136, 0x0137 },  // LATIN CAPITAL LETTER K WITH CEDILLA
    139     { 0x0139, 0x013A },  // LATIN CAPITAL LETTER L WITH ACUTE
    140     { 0x013B, 0x013C },  // LATIN CAPITAL LETTER L WITH CEDILLA
    141     { 0x013D, 0x013E },  // LATIN CAPITAL LETTER L WITH CARON
    142     { 0x013F, 0x0140 },  // LATIN CAPITAL LETTER L WITH MIDDLE DOT
    143     { 0x0141, 0x0142 },  // LATIN CAPITAL LETTER L WITH STROKE
    144     { 0x0143, 0x0144 },  // LATIN CAPITAL LETTER N WITH ACUTE
    145     { 0x0145, 0x0146 },  // LATIN CAPITAL LETTER N WITH CEDILLA
    146     { 0x0147, 0x0148 },  // LATIN CAPITAL LETTER N WITH CARON
    147     { 0x014A, 0x014B },  // LATIN CAPITAL LETTER ENG
    148     { 0x014C, 0x014D },  // LATIN CAPITAL LETTER O WITH MACRON
    149     { 0x014E, 0x014F },  // LATIN CAPITAL LETTER O WITH BREVE
    150     { 0x0150, 0x0151 },  // LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
    151     { 0x0152, 0x0153 },  // LATIN CAPITAL LIGATURE OE
    152     { 0x0154, 0x0155 },  // LATIN CAPITAL LETTER R WITH ACUTE
    153     { 0x0156, 0x0157 },  // LATIN CAPITAL LETTER R WITH CEDILLA
    154     { 0x0158, 0x0159 },  // LATIN CAPITAL LETTER R WITH CARON
    155     { 0x015A, 0x015B },  // LATIN CAPITAL LETTER S WITH ACUTE
    156     { 0x015C, 0x015D },  // LATIN CAPITAL LETTER S WITH CIRCUMFLEX
    157     { 0x015E, 0x015F },  // LATIN CAPITAL LETTER S WITH CEDILLA
    158     { 0x0160, 0x0161 },  // LATIN CAPITAL LETTER S WITH CARON
    159     { 0x0162, 0x0163 },  // LATIN CAPITAL LETTER T WITH CEDILLA
    160     { 0x0164, 0x0165 },  // LATIN CAPITAL LETTER T WITH CARON
    161     { 0x0166, 0x0167 },  // LATIN CAPITAL LETTER T WITH STROKE
    162     { 0x0168, 0x0169 },  // LATIN CAPITAL LETTER U WITH TILDE
    163     { 0x016A, 0x016B },  // LATIN CAPITAL LETTER U WITH MACRON
    164     { 0x016C, 0x016D },  // LATIN CAPITAL LETTER U WITH BREVE
    165     { 0x016E, 0x016F },  // LATIN CAPITAL LETTER U WITH RING ABOVE
    166     { 0x0170, 0x0171 },  // LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
    167     { 0x0172, 0x0173 },  // LATIN CAPITAL LETTER U WITH OGONEK
    168     { 0x0174, 0x0175 },  // LATIN CAPITAL LETTER W WITH CIRCUMFLEX
    169     { 0x0176, 0x0177 },  // LATIN CAPITAL LETTER Y WITH CIRCUMFLEX
    170     { 0x0178, 0x00FF },  // LATIN CAPITAL LETTER Y WITH DIAERESIS
    171     { 0x0179, 0x017A },  // LATIN CAPITAL LETTER Z WITH ACUTE
    172     { 0x017B, 0x017C },  // LATIN CAPITAL LETTER Z WITH DOT ABOVE
    173     { 0x017D, 0x017E },  // LATIN CAPITAL LETTER Z WITH CARON
    174     { 0x0181, 0x0253 },  // LATIN CAPITAL LETTER B WITH HOOK
    175     { 0x0182, 0x0183 },  // LATIN CAPITAL LETTER B WITH TOPBAR
    176     { 0x0184, 0x0185 },  // LATIN CAPITAL LETTER TONE SIX
    177     { 0x0186, 0x0254 },  // LATIN CAPITAL LETTER OPEN O
    178     { 0x0187, 0x0188 },  // LATIN CAPITAL LETTER C WITH HOOK
    179     { 0x0189, 0x0256 },  // LATIN CAPITAL LETTER AFRICAN D
    180     { 0x018A, 0x0257 },  // LATIN CAPITAL LETTER D WITH HOOK
    181     { 0x018B, 0x018C },  // LATIN CAPITAL LETTER D WITH TOPBAR
    182     { 0x018E, 0x01DD },  // LATIN CAPITAL LETTER REVERSED E
    183     { 0x018F, 0x0259 },  // LATIN CAPITAL LETTER SCHWA
    184     { 0x0190, 0x025B },  // LATIN CAPITAL LETTER OPEN E
    185     { 0x0191, 0x0192 },  // LATIN CAPITAL LETTER F WITH HOOK
    186     { 0x0193, 0x0260 },  // LATIN CAPITAL LETTER G WITH HOOK
    187     { 0x0194, 0x0263 },  // LATIN CAPITAL LETTER GAMMA
    188     { 0x0196, 0x0269 },  // LATIN CAPITAL LETTER IOTA
    189     { 0x0197, 0x0268 },  // LATIN CAPITAL LETTER I WITH STROKE
    190     { 0x0198, 0x0199 },  // LATIN CAPITAL LETTER K WITH HOOK
    191     { 0x019C, 0x026F },  // LATIN CAPITAL LETTER TURNED M
    192     { 0x019D, 0x0272 },  // LATIN CAPITAL LETTER N WITH LEFT HOOK
    193     { 0x019F, 0x0275 },  // LATIN CAPITAL LETTER O WITH MIDDLE TILDE
    194     { 0x01A0, 0x01A1 },  // LATIN CAPITAL LETTER O WITH HORN
    195     { 0x01A2, 0x01A3 },  // LATIN CAPITAL LETTER OI
    196     { 0x01A4, 0x01A5 },  // LATIN CAPITAL LETTER P WITH HOOK
    197     { 0x01A6, 0x0280 },  // LATIN LETTER YR
    198     { 0x01A7, 0x01A8 },  // LATIN CAPITAL LETTER TONE TWO
    199     { 0x01A9, 0x0283 },  // LATIN CAPITAL LETTER ESH
    200     { 0x01AC, 0x01AD },  // LATIN CAPITAL LETTER T WITH HOOK
    201     { 0x01AE, 0x0288 },  // LATIN CAPITAL LETTER T WITH RETROFLEX HOOK
    202     { 0x01AF, 0x01B0 },  // LATIN CAPITAL LETTER U WITH HORN
    203     { 0x01B1, 0x028A },  // LATIN CAPITAL LETTER UPSILON
    204     { 0x01B2, 0x028B },  // LATIN CAPITAL LETTER V WITH HOOK
    205     { 0x01B3, 0x01B4 },  // LATIN CAPITAL LETTER Y WITH HOOK
    206     { 0x01B5, 0x01B6 },  // LATIN CAPITAL LETTER Z WITH STROKE
    207     { 0x01B7, 0x0292 },  // LATIN CAPITAL LETTER EZH
    208     { 0x01B8, 0x01B9 },  // LATIN CAPITAL LETTER EZH REVERSED
    209     { 0x01BC, 0x01BD },  // LATIN CAPITAL LETTER TONE FIVE
    210     { 0x01C4, 0x01C6 },  // LATIN CAPITAL LETTER DZ WITH CARON
    211     { 0x01C5, 0x01C6 },  // LATIN CAPITAL LETTER D WITH SMALL LETTER Z WITH CARON
    212     { 0x01C7, 0x01C9 },  // LATIN CAPITAL LETTER LJ
    213     { 0x01C8, 0x01C9 },  // LATIN CAPITAL LETTER L WITH SMALL LETTER J
    214     { 0x01CA, 0x01CC },  // LATIN CAPITAL LETTER NJ
    215     { 0x01CB, 0x01CC },  // LATIN CAPITAL LETTER N WITH SMALL LETTER J
    216     { 0x01CD, 0x01CE },  // LATIN CAPITAL LETTER A WITH CARON
    217     { 0x01CF, 0x01D0 },  // LATIN CAPITAL LETTER I WITH CARON
    218     { 0x01D1, 0x01D2 },  // LATIN CAPITAL LETTER O WITH CARON
    219     { 0x01D3, 0x01D4 },  // LATIN CAPITAL LETTER U WITH CARON
    220     { 0x01D5, 0x01D6 },  // LATIN CAPITAL LETTER U WITH DIAERESIS AND MACRON
    221     { 0x01D7, 0x01D8 },  // LATIN CAPITAL LETTER U WITH DIAERESIS AND ACUTE
    222     { 0x01D9, 0x01DA },  // LATIN CAPITAL LETTER U WITH DIAERESIS AND CARON
    223     { 0x01DB, 0x01DC },  // LATIN CAPITAL LETTER U WITH DIAERESIS AND GRAVE
    224     { 0x01DE, 0x01DF },  // LATIN CAPITAL LETTER A WITH DIAERESIS AND MACRON
    225     { 0x01E0, 0x01E1 },  // LATIN CAPITAL LETTER A WITH DOT ABOVE AND MACRON
    226     { 0x01E2, 0x01E3 },  // LATIN CAPITAL LETTER AE WITH MACRON
    227     { 0x01E4, 0x01E5 },  // LATIN CAPITAL LETTER G WITH STROKE
    228     { 0x01E6, 0x01E7 },  // LATIN CAPITAL LETTER G WITH CARON
    229     { 0x01E8, 0x01E9 },  // LATIN CAPITAL LETTER K WITH CARON
    230     { 0x01EA, 0x01EB },  // LATIN CAPITAL LETTER O WITH OGONEK
    231     { 0x01EC, 0x01ED },  // LATIN CAPITAL LETTER O WITH OGONEK AND MACRON
    232     { 0x01EE, 0x01EF },  // LATIN CAPITAL LETTER EZH WITH CARON
    233     { 0x01F1, 0x01F3 },  // LATIN CAPITAL LETTER DZ
    234     { 0x01F2, 0x01F3 },  // LATIN CAPITAL LETTER D WITH SMALL LETTER Z
    235     { 0x01F4, 0x01F5 },  // LATIN CAPITAL LETTER G WITH ACUTE
    236     { 0x01F6, 0x0195 },  // LATIN CAPITAL LETTER HWAIR
    237     { 0x01F7, 0x01BF },  // LATIN CAPITAL LETTER WYNN
    238     { 0x01F8, 0x01F9 },  // LATIN CAPITAL LETTER N WITH GRAVE
    239     { 0x01FA, 0x01FB },  // LATIN CAPITAL LETTER A WITH RING ABOVE AND ACUTE
    240     { 0x01FC, 0x01FD },  // LATIN CAPITAL LETTER AE WITH ACUTE
    241     { 0x01FE, 0x01FF },  // LATIN CAPITAL LETTER O WITH STROKE AND ACUTE
    242     { 0x0200, 0x0201 },  // LATIN CAPITAL LETTER A WITH DOUBLE GRAVE
    243     { 0x0202, 0x0203 },  // LATIN CAPITAL LETTER A WITH INVERTED BREVE
    244     { 0x0204, 0x0205 },  // LATIN CAPITAL LETTER E WITH DOUBLE GRAVE
    245     { 0x0206, 0x0207 },  // LATIN CAPITAL LETTER E WITH INVERTED BREVE
    246     { 0x0208, 0x0209 },  // LATIN CAPITAL LETTER I WITH DOUBLE GRAVE
    247     { 0x020A, 0x020B },  // LATIN CAPITAL LETTER I WITH INVERTED BREVE
    248     { 0x020C, 0x020D },  // LATIN CAPITAL LETTER O WITH DOUBLE GRAVE
    249     { 0x020E, 0x020F },  // LATIN CAPITAL LETTER O WITH INVERTED BREVE
    250     { 0x0210, 0x0211 },  // LATIN CAPITAL LETTER R WITH DOUBLE GRAVE
    251     { 0x0212, 0x0213 },  // LATIN CAPITAL LETTER R WITH INVERTED BREVE
    252     { 0x0214, 0x0215 },  // LATIN CAPITAL LETTER U WITH DOUBLE GRAVE
    253     { 0x0216, 0x0217 },  // LATIN CAPITAL LETTER U WITH INVERTED BREVE
    254     { 0x0218, 0x0219 },  // LATIN CAPITAL LETTER S WITH COMMA BELOW
    255     { 0x021A, 0x021B },  // LATIN CAPITAL LETTER T WITH COMMA BELOW
    256     { 0x021C, 0x021D },  // LATIN CAPITAL LETTER YOGH
    257     { 0x021E, 0x021F },  // LATIN CAPITAL LETTER H WITH CARON
    258     { 0x0220, 0x019E },  // LATIN CAPITAL LETTER N WITH LONG RIGHT LEG
    259     { 0x0222, 0x0223 },  // LATIN CAPITAL LETTER OU
    260     { 0x0224, 0x0225 },  // LATIN CAPITAL LETTER Z WITH HOOK
    261     { 0x0226, 0x0227 },  // LATIN CAPITAL LETTER A WITH DOT ABOVE
    262     { 0x0228, 0x0229 },  // LATIN CAPITAL LETTER E WITH CEDILLA
    263     { 0x022A, 0x022B },  // LATIN CAPITAL LETTER O WITH DIAERESIS AND MACRON
    264     { 0x022C, 0x022D },  // LATIN CAPITAL LETTER O WITH TILDE AND MACRON
    265     { 0x022E, 0x022F },  // LATIN CAPITAL LETTER O WITH DOT ABOVE
    266     { 0x0230, 0x0231 },  // LATIN CAPITAL LETTER O WITH DOT ABOVE AND MACRON
    267     { 0x0232, 0x0233 },  // LATIN CAPITAL LETTER Y WITH MACRON
    268     { 0x023A, 0x2C65 },  // LATIN CAPITAL LETTER A WITH STROKE
    269     { 0x023B, 0x023C },  // LATIN CAPITAL LETTER C WITH STROKE
    270     { 0x023D, 0x019A },  // LATIN CAPITAL LETTER L WITH BAR
    271     { 0x023E, 0x2C66 },  // LATIN CAPITAL LETTER T WITH DIAGONAL STROKE
    272     { 0x0241, 0x0242 },  // LATIN CAPITAL LETTER GLOTTAL STOP
    273     { 0x0243, 0x0180 },  // LATIN CAPITAL LETTER B WITH STROKE
    274     { 0x0244, 0x0289 },  // LATIN CAPITAL LETTER U BAR
    275     { 0x0245, 0x028C },  // LATIN CAPITAL LETTER TURNED V
    276     { 0x0246, 0x0247 },  // LATIN CAPITAL LETTER E WITH STROKE
    277     { 0x0248, 0x0249 },  // LATIN CAPITAL LETTER J WITH STROKE
    278     { 0x024A, 0x024B },  // LATIN CAPITAL LETTER SMALL Q WITH HOOK TAIL
    279     { 0x024C, 0x024D },  // LATIN CAPITAL LETTER R WITH STROKE
    280     { 0x024E, 0x024F },  // LATIN CAPITAL LETTER Y WITH STROKE
    281     { 0x0370, 0x0371 },  // GREEK CAPITAL LETTER HETA
    282     { 0x0372, 0x0373 },  // GREEK CAPITAL LETTER ARCHAIC SAMPI
    283     { 0x0376, 0x0377 },  // GREEK CAPITAL LETTER PAMPHYLIAN DIGAMMA
    284     { 0x0386, 0x03AC },  // GREEK CAPITAL LETTER ALPHA WITH TONOS
    285     { 0x0388, 0x03AD },  // GREEK CAPITAL LETTER EPSILON WITH TONOS
    286     { 0x0389, 0x03AE },  // GREEK CAPITAL LETTER ETA WITH TONOS
    287     { 0x038A, 0x03AF },  // GREEK CAPITAL LETTER IOTA WITH TONOS
    288     { 0x038C, 0x03CC },  // GREEK CAPITAL LETTER OMICRON WITH TONOS
    289     { 0x038E, 0x03CD },  // GREEK CAPITAL LETTER UPSILON WITH TONOS
    290     { 0x038F, 0x03CE },  // GREEK CAPITAL LETTER OMEGA WITH TONOS
    291     { 0x0391, 0x03B1 },  // GREEK CAPITAL LETTER ALPHA
    292     { 0x0392, 0x03B2 },  // GREEK CAPITAL LETTER BETA
    293     { 0x0393, 0x03B3 },  // GREEK CAPITAL LETTER GAMMA
    294     { 0x0394, 0x03B4 },  // GREEK CAPITAL LETTER DELTA
    295     { 0x0395, 0x03B5 },  // GREEK CAPITAL LETTER EPSILON
    296     { 0x0396, 0x03B6 },  // GREEK CAPITAL LETTER ZETA
    297     { 0x0397, 0x03B7 },  // GREEK CAPITAL LETTER ETA
    298     { 0x0398, 0x03B8 },  // GREEK CAPITAL LETTER THETA
    299     { 0x0399, 0x03B9 },  // GREEK CAPITAL LETTER IOTA
    300     { 0x039A, 0x03BA },  // GREEK CAPITAL LETTER KAPPA
    301     { 0x039B, 0x03BB },  // GREEK CAPITAL LETTER LAMDA
    302     { 0x039C, 0x03BC },  // GREEK CAPITAL LETTER MU
    303     { 0x039D, 0x03BD },  // GREEK CAPITAL LETTER NU
    304     { 0x039E, 0x03BE },  // GREEK CAPITAL LETTER XI
    305     { 0x039F, 0x03BF },  // GREEK CAPITAL LETTER OMICRON
    306     { 0x03A0, 0x03C0 },  // GREEK CAPITAL LETTER PI
    307     { 0x03A1, 0x03C1 },  // GREEK CAPITAL LETTER RHO
    308     { 0x03A3, 0x03C3 },  // GREEK CAPITAL LETTER SIGMA
    309     { 0x03A4, 0x03C4 },  // GREEK CAPITAL LETTER TAU
    310     { 0x03A5, 0x03C5 },  // GREEK CAPITAL LETTER UPSILON
    311     { 0x03A6, 0x03C6 },  // GREEK CAPITAL LETTER PHI
    312     { 0x03A7, 0x03C7 },  // GREEK CAPITAL LETTER CHI
    313     { 0x03A8, 0x03C8 },  // GREEK CAPITAL LETTER PSI
    314     { 0x03A9, 0x03C9 },  // GREEK CAPITAL LETTER OMEGA
    315     { 0x03AA, 0x03CA },  // GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
    316     { 0x03AB, 0x03CB },  // GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
    317     { 0x03CF, 0x03D7 },  // GREEK CAPITAL KAI SYMBOL
    318     { 0x03D8, 0x03D9 },  // GREEK LETTER ARCHAIC KOPPA
    319     { 0x03DA, 0x03DB },  // GREEK LETTER STIGMA
    320     { 0x03DC, 0x03DD },  // GREEK LETTER DIGAMMA
    321     { 0x03DE, 0x03DF },  // GREEK LETTER KOPPA
    322     { 0x03E0, 0x03E1 },  // GREEK LETTER SAMPI
    323     { 0x03E2, 0x03E3 },  // COPTIC CAPITAL LETTER SHEI
    324     { 0x03E4, 0x03E5 },  // COPTIC CAPITAL LETTER FEI
    325     { 0x03E6, 0x03E7 },  // COPTIC CAPITAL LETTER KHEI
    326     { 0x03E8, 0x03E9 },  // COPTIC CAPITAL LETTER HORI
    327     { 0x03EA, 0x03EB },  // COPTIC CAPITAL LETTER GANGIA
    328     { 0x03EC, 0x03ED },  // COPTIC CAPITAL LETTER SHIMA
    329     { 0x03EE, 0x03EF },  // COPTIC CAPITAL LETTER DEI
    330     { 0x03F4, 0x03B8 },  // GREEK CAPITAL THETA SYMBOL
    331     { 0x03F7, 0x03F8 },  // GREEK CAPITAL LETTER SHO
    332     { 0x03F9, 0x03F2 },  // GREEK CAPITAL LUNATE SIGMA SYMBOL
    333     { 0x03FA, 0x03FB },  // GREEK CAPITAL LETTER SAN
    334     { 0x03FD, 0x037B },  // GREEK CAPITAL REVERSED LUNATE SIGMA SYMBOL
    335     { 0x03FE, 0x037C },  // GREEK CAPITAL DOTTED LUNATE SIGMA SYMBOL
    336     { 0x03FF, 0x037D },  // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL
    337     { 0x0400, 0x0450 },  // CYRILLIC CAPITAL LETTER IE WITH GRAVE
    338     { 0x0401, 0x0451 },  // CYRILLIC CAPITAL LETTER IO
    339     { 0x0402, 0x0452 },  // CYRILLIC CAPITAL LETTER DJE
    340     { 0x0403, 0x0453 },  // CYRILLIC CAPITAL LETTER GJE
    341     { 0x0404, 0x0454 },  // CYRILLIC CAPITAL LETTER UKRAINIAN IE
    342     { 0x0405, 0x0455 },  // CYRILLIC CAPITAL LETTER DZE
    343     { 0x0406, 0x0456 },  // CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
    344     { 0x0407, 0x0457 },  // CYRILLIC CAPITAL LETTER YI
    345     { 0x0408, 0x0458 },  // CYRILLIC CAPITAL LETTER JE
    346     { 0x0409, 0x0459 },  // CYRILLIC CAPITAL LETTER LJE
    347     { 0x040A, 0x045A },  // CYRILLIC CAPITAL LETTER NJE
    348     { 0x040B, 0x045B },  // CYRILLIC CAPITAL LETTER TSHE
    349     { 0x040C, 0x045C },  // CYRILLIC CAPITAL LETTER KJE
    350     { 0x040D, 0x045D },  // CYRILLIC CAPITAL LETTER I WITH GRAVE
    351     { 0x040E, 0x045E },  // CYRILLIC CAPITAL LETTER SHORT U
    352     { 0x040F, 0x045F },  // CYRILLIC CAPITAL LETTER DZHE
    353     { 0x0410, 0x0430 },  // CYRILLIC CAPITAL LETTER A
    354     { 0x0411, 0x0431 },  // CYRILLIC CAPITAL LETTER BE
    355     { 0x0412, 0x0432 },  // CYRILLIC CAPITAL LETTER VE
    356     { 0x0413, 0x0433 },  // CYRILLIC CAPITAL LETTER GHE
    357     { 0x0414, 0x0434 },  // CYRILLIC CAPITAL LETTER DE
    358     { 0x0415, 0x0435 },  // CYRILLIC CAPITAL LETTER IE
    359     { 0x0416, 0x0436 },  // CYRILLIC CAPITAL LETTER ZHE
    360     { 0x0417, 0x0437 },  // CYRILLIC CAPITAL LETTER ZE
    361     { 0x0418, 0x0438 },  // CYRILLIC CAPITAL LETTER I
    362     { 0x0419, 0x0439 },  // CYRILLIC CAPITAL LETTER SHORT I
    363     { 0x041A, 0x043A },  // CYRILLIC CAPITAL LETTER KA
    364     { 0x041B, 0x043B },  // CYRILLIC CAPITAL LETTER EL
    365     { 0x041C, 0x043C },  // CYRILLIC CAPITAL LETTER EM
    366     { 0x041D, 0x043D },  // CYRILLIC CAPITAL LETTER EN
    367     { 0x041E, 0x043E },  // CYRILLIC CAPITAL LETTER O
    368     { 0x041F, 0x043F },  // CYRILLIC CAPITAL LETTER PE
    369     { 0x0420, 0x0440 },  // CYRILLIC CAPITAL LETTER ER
    370     { 0x0421, 0x0441 },  // CYRILLIC CAPITAL LETTER ES
    371     { 0x0422, 0x0442 },  // CYRILLIC CAPITAL LETTER TE
    372     { 0x0423, 0x0443 },  // CYRILLIC CAPITAL LETTER U
    373     { 0x0424, 0x0444 },  // CYRILLIC CAPITAL LETTER EF
    374     { 0x0425, 0x0445 },  // CYRILLIC CAPITAL LETTER HA
    375     { 0x0426, 0x0446 },  // CYRILLIC CAPITAL LETTER TSE
    376     { 0x0427, 0x0447 },  // CYRILLIC CAPITAL LETTER CHE
    377     { 0x0428, 0x0448 },  // CYRILLIC CAPITAL LETTER SHA
    378     { 0x0429, 0x0449 },  // CYRILLIC CAPITAL LETTER SHCHA
    379     { 0x042A, 0x044A },  // CYRILLIC CAPITAL LETTER HARD SIGN
    380     { 0x042B, 0x044B },  // CYRILLIC CAPITAL LETTER YERU
    381     { 0x042C, 0x044C },  // CYRILLIC CAPITAL LETTER SOFT SIGN
    382     { 0x042D, 0x044D },  // CYRILLIC CAPITAL LETTER E
    383     { 0x042E, 0x044E },  // CYRILLIC CAPITAL LETTER YU
    384     { 0x042F, 0x044F },  // CYRILLIC CAPITAL LETTER YA
    385     { 0x0460, 0x0461 },  // CYRILLIC CAPITAL LETTER OMEGA
    386     { 0x0462, 0x0463 },  // CYRILLIC CAPITAL LETTER YAT
    387     { 0x0464, 0x0465 },  // CYRILLIC CAPITAL LETTER IOTIFIED E
    388     { 0x0466, 0x0467 },  // CYRILLIC CAPITAL LETTER LITTLE YUS
    389     { 0x0468, 0x0469 },  // CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS
    390     { 0x046A, 0x046B },  // CYRILLIC CAPITAL LETTER BIG YUS
    391     { 0x046C, 0x046D },  // CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS
    392     { 0x046E, 0x046F },  // CYRILLIC CAPITAL LETTER KSI
    393     { 0x0470, 0x0471 },  // CYRILLIC CAPITAL LETTER PSI
    394     { 0x0472, 0x0473 },  // CYRILLIC CAPITAL LETTER FITA
    395     { 0x0474, 0x0475 },  // CYRILLIC CAPITAL LETTER IZHITSA
    396     { 0x0476, 0x0477 },  // CYRILLIC CAPITAL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT
    397     { 0x0478, 0x0479 },  // CYRILLIC CAPITAL LETTER UK
    398     { 0x047A, 0x047B },  // CYRILLIC CAPITAL LETTER ROUND OMEGA
    399     { 0x047C, 0x047D },  // CYRILLIC CAPITAL LETTER OMEGA WITH TITLO
    400     { 0x047E, 0x047F },  // CYRILLIC CAPITAL LETTER OT
    401     { 0x0480, 0x0481 },  // CYRILLIC CAPITAL LETTER KOPPA
    402     { 0x048A, 0x048B },  // CYRILLIC CAPITAL LETTER SHORT I WITH TAIL
    403     { 0x048C, 0x048D },  // CYRILLIC CAPITAL LETTER SEMISOFT SIGN
    404     { 0x048E, 0x048F },  // CYRILLIC CAPITAL LETTER ER WITH TICK
    405     { 0x0490, 0x0491 },  // CYRILLIC CAPITAL LETTER GHE WITH UPTURN
    406     { 0x0492, 0x0493 },  // CYRILLIC CAPITAL LETTER GHE WITH STROKE
    407     { 0x0494, 0x0495 },  // CYRILLIC CAPITAL LETTER GHE WITH MIDDLE HOOK
    408     { 0x0496, 0x0497 },  // CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER
    409     { 0x0498, 0x0499 },  // CYRILLIC CAPITAL LETTER ZE WITH DESCENDER
    410     { 0x049A, 0x049B },  // CYRILLIC CAPITAL LETTER KA WITH DESCENDER
    411     { 0x049C, 0x049D },  // CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE
    412     { 0x049E, 0x049F },  // CYRILLIC CAPITAL LETTER KA WITH STROKE
    413     { 0x04A0, 0x04A1 },  // CYRILLIC CAPITAL LETTER BASHKIR KA
    414     { 0x04A2, 0x04A3 },  // CYRILLIC CAPITAL LETTER EN WITH DESCENDER
    415     { 0x04A4, 0x04A5 },  // CYRILLIC CAPITAL LIGATURE EN GHE
    416     { 0x04A6, 0x04A7 },  // CYRILLIC CAPITAL LETTER PE WITH MIDDLE HOOK
    417     { 0x04A8, 0x04A9 },  // CYRILLIC CAPITAL LETTER ABKHASIAN HA
    418     { 0x04AA, 0x04AB },  // CYRILLIC CAPITAL LETTER ES WITH DESCENDER
    419     { 0x04AC, 0x04AD },  // CYRILLIC CAPITAL LETTER TE WITH DESCENDER
    420     { 0x04AE, 0x04AF },  // CYRILLIC CAPITAL LETTER STRAIGHT U
    421     { 0x04B0, 0x04B1 },  // CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE
    422     { 0x04B2, 0x04B3 },  // CYRILLIC CAPITAL LETTER HA WITH DESCENDER
    423     { 0x04B4, 0x04B5 },  // CYRILLIC CAPITAL LIGATURE TE TSE
    424     { 0x04B6, 0x04B7 },  // CYRILLIC CAPITAL LETTER CHE WITH DESCENDER
    425     { 0x04B8, 0x04B9 },  // CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE
    426     { 0x04BA, 0x04BB },  // CYRILLIC CAPITAL LETTER SHHA
    427     { 0x04BC, 0x04BD },  // CYRILLIC CAPITAL LETTER ABKHASIAN CHE
    428     { 0x04BE, 0x04BF },  // CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER
    429     { 0x04C0, 0x04CF },  // CYRILLIC LETTER PALOCHKA
    430     { 0x04C1, 0x04C2 },  // CYRILLIC CAPITAL LETTER ZHE WITH BREVE
    431     { 0x04C3, 0x04C4 },  // CYRILLIC CAPITAL LETTER KA WITH HOOK
    432     { 0x04C5, 0x04C6 },  // CYRILLIC CAPITAL LETTER EL WITH TAIL
    433     { 0x04C7, 0x04C8 },  // CYRILLIC CAPITAL LETTER EN WITH HOOK
    434     { 0x04C9, 0x04CA },  // CYRILLIC CAPITAL LETTER EN WITH TAIL
    435     { 0x04CB, 0x04CC },  // CYRILLIC CAPITAL LETTER KHAKASSIAN CHE
    436     { 0x04CD, 0x04CE },  // CYRILLIC CAPITAL LETTER EM WITH TAIL
    437     { 0x04D0, 0x04D1 },  // CYRILLIC CAPITAL LETTER A WITH BREVE
    438     { 0x04D2, 0x04D3 },  // CYRILLIC CAPITAL LETTER A WITH DIAERESIS
    439     { 0x04D4, 0x04D5 },  // CYRILLIC CAPITAL LIGATURE A IE
    440     { 0x04D6, 0x04D7 },  // CYRILLIC CAPITAL LETTER IE WITH BREVE
    441     { 0x04D8, 0x04D9 },  // CYRILLIC CAPITAL LETTER SCHWA
    442     { 0x04DA, 0x04DB },  // CYRILLIC CAPITAL LETTER SCHWA WITH DIAERESIS
    443     { 0x04DC, 0x04DD },  // CYRILLIC CAPITAL LETTER ZHE WITH DIAERESIS
    444     { 0x04DE, 0x04DF },  // CYRILLIC CAPITAL LETTER ZE WITH DIAERESIS
    445     { 0x04E0, 0x04E1 },  // CYRILLIC CAPITAL LETTER ABKHASIAN DZE
    446     { 0x04E2, 0x04E3 },  // CYRILLIC CAPITAL LETTER I WITH MACRON
    447     { 0x04E4, 0x04E5 },  // CYRILLIC CAPITAL LETTER I WITH DIAERESIS
    448     { 0x04E6, 0x04E7 },  // CYRILLIC CAPITAL LETTER O WITH DIAERESIS
    449     { 0x04E8, 0x04E9 },  // CYRILLIC CAPITAL LETTER BARRED O
    450     { 0x04EA, 0x04EB },  // CYRILLIC CAPITAL LETTER BARRED O WITH DIAERESIS
    451     { 0x04EC, 0x04ED },  // CYRILLIC CAPITAL LETTER E WITH DIAERESIS
    452     { 0x04EE, 0x04EF },  // CYRILLIC CAPITAL LETTER U WITH MACRON
    453     { 0x04F0, 0x04F1 },  // CYRILLIC CAPITAL LETTER U WITH DIAERESIS
    454     { 0x04F2, 0x04F3 },  // CYRILLIC CAPITAL LETTER U WITH DOUBLE ACUTE
    455     { 0x04F4, 0x04F5 },  // CYRILLIC CAPITAL LETTER CHE WITH DIAERESIS
    456     { 0x04F6, 0x04F7 },  // CYRILLIC CAPITAL LETTER GHE WITH DESCENDER
    457     { 0x04F8, 0x04F9 },  // CYRILLIC CAPITAL LETTER YERU WITH DIAERESIS
    458     { 0x04FA, 0x04FB },  // CYRILLIC CAPITAL LETTER GHE WITH STROKE AND HOOK
    459     { 0x04FC, 0x04FD },  // CYRILLIC CAPITAL LETTER HA WITH HOOK
    460     { 0x04FE, 0x04FF },  // CYRILLIC CAPITAL LETTER HA WITH STROKE
    461     { 0x0500, 0x0501 },  // CYRILLIC CAPITAL LETTER KOMI DE
    462     { 0x0502, 0x0503 },  // CYRILLIC CAPITAL LETTER KOMI DJE
    463     { 0x0504, 0x0505 },  // CYRILLIC CAPITAL LETTER KOMI ZJE
    464     { 0x0506, 0x0507 },  // CYRILLIC CAPITAL LETTER KOMI DZJE
    465     { 0x0508, 0x0509 },  // CYRILLIC CAPITAL LETTER KOMI LJE
    466     { 0x050A, 0x050B },  // CYRILLIC CAPITAL LETTER KOMI NJE
    467     { 0x050C, 0x050D },  // CYRILLIC CAPITAL LETTER KOMI SJE
    468     { 0x050E, 0x050F },  // CYRILLIC CAPITAL LETTER KOMI TJE
    469     { 0x0510, 0x0511 },  // CYRILLIC CAPITAL LETTER REVERSED ZE
    470     { 0x0512, 0x0513 },  // CYRILLIC CAPITAL LETTER EL WITH HOOK
    471     { 0x0514, 0x0515 },  // CYRILLIC CAPITAL LETTER LHA
    472     { 0x0516, 0x0517 },  // CYRILLIC CAPITAL LETTER RHA
    473     { 0x0518, 0x0519 },  // CYRILLIC CAPITAL LETTER YAE
    474     { 0x051A, 0x051B },  // CYRILLIC CAPITAL LETTER QA
    475     { 0x051C, 0x051D },  // CYRILLIC CAPITAL LETTER WE
    476     { 0x051E, 0x051F },  // CYRILLIC CAPITAL LETTER ALEUT KA
    477     { 0x0520, 0x0521 },  // CYRILLIC CAPITAL LETTER EL WITH MIDDLE HOOK
    478     { 0x0522, 0x0523 },  // CYRILLIC CAPITAL LETTER EN WITH MIDDLE HOOK
    479     { 0x0524, 0x0525 },  // CYRILLIC CAPITAL LETTER PE WITH DESCENDER
    480     { 0x0526, 0x0527 },  // CYRILLIC CAPITAL LETTER SHHA WITH DESCENDER
    481     { 0x0531, 0x0561 },  // ARMENIAN CAPITAL LETTER AYB
    482     { 0x0532, 0x0562 },  // ARMENIAN CAPITAL LETTER BEN
    483     { 0x0533, 0x0563 },  // ARMENIAN CAPITAL LETTER GIM
    484     { 0x0534, 0x0564 },  // ARMENIAN CAPITAL LETTER DA
    485     { 0x0535, 0x0565 },  // ARMENIAN CAPITAL LETTER ECH
    486     { 0x0536, 0x0566 },  // ARMENIAN CAPITAL LETTER ZA
    487     { 0x0537, 0x0567 },  // ARMENIAN CAPITAL LETTER EH
    488     { 0x0538, 0x0568 },  // ARMENIAN CAPITAL LETTER ET
    489     { 0x0539, 0x0569 },  // ARMENIAN CAPITAL LETTER TO
    490     { 0x053A, 0x056A },  // ARMENIAN CAPITAL LETTER ZHE
    491     { 0x053B, 0x056B },  // ARMENIAN CAPITAL LETTER INI
    492     { 0x053C, 0x056C },  // ARMENIAN CAPITAL LETTER LIWN
    493     { 0x053D, 0x056D },  // ARMENIAN CAPITAL LETTER XEH
    494     { 0x053E, 0x056E },  // ARMENIAN CAPITAL LETTER CA
    495     { 0x053F, 0x056F },  // ARMENIAN CAPITAL LETTER KEN
    496     { 0x0540, 0x0570 },  // ARMENIAN CAPITAL LETTER HO
    497     { 0x0541, 0x0571 },  // ARMENIAN CAPITAL LETTER JA
    498     { 0x0542, 0x0572 },  // ARMENIAN CAPITAL LETTER GHAD
    499     { 0x0543, 0x0573 },  // ARMENIAN CAPITAL LETTER CHEH
    500     { 0x0544, 0x0574 },  // ARMENIAN CAPITAL LETTER MEN
    501     { 0x0545, 0x0575 },  // ARMENIAN CAPITAL LETTER YI
    502     { 0x0546, 0x0576 },  // ARMENIAN CAPITAL LETTER NOW
    503     { 0x0547, 0x0577 },  // ARMENIAN CAPITAL LETTER SHA
    504     { 0x0548, 0x0578 },  // ARMENIAN CAPITAL LETTER VO
    505     { 0x0549, 0x0579 },  // ARMENIAN CAPITAL LETTER CHA
    506     { 0x054A, 0x057A },  // ARMENIAN CAPITAL LETTER PEH
    507     { 0x054B, 0x057B },  // ARMENIAN CAPITAL LETTER JHEH
    508     { 0x054C, 0x057C },  // ARMENIAN CAPITAL LETTER RA
    509     { 0x054D, 0x057D },  // ARMENIAN CAPITAL LETTER SEH
    510     { 0x054E, 0x057E },  // ARMENIAN CAPITAL LETTER VEW
    511     { 0x054F, 0x057F },  // ARMENIAN CAPITAL LETTER TIWN
    512     { 0x0550, 0x0580 },  // ARMENIAN CAPITAL LETTER REH
    513     { 0x0551, 0x0581 },  // ARMENIAN CAPITAL LETTER CO
    514     { 0x0552, 0x0582 },  // ARMENIAN CAPITAL LETTER YIWN
    515     { 0x0553, 0x0583 },  // ARMENIAN CAPITAL LETTER PIWR
    516     { 0x0554, 0x0584 },  // ARMENIAN CAPITAL LETTER KEH
    517     { 0x0555, 0x0585 },  // ARMENIAN CAPITAL LETTER OH
    518     { 0x0556, 0x0586 },  // ARMENIAN CAPITAL LETTER FEH
    519     { 0x10A0, 0x2D00 },  // GEORGIAN CAPITAL LETTER AN
    520     { 0x10A1, 0x2D01 },  // GEORGIAN CAPITAL LETTER BAN
    521     { 0x10A2, 0x2D02 },  // GEORGIAN CAPITAL LETTER GAN
    522     { 0x10A3, 0x2D03 },  // GEORGIAN CAPITAL LETTER DON
    523     { 0x10A4, 0x2D04 },  // GEORGIAN CAPITAL LETTER EN
    524     { 0x10A5, 0x2D05 },  // GEORGIAN CAPITAL LETTER VIN
    525     { 0x10A6, 0x2D06 },  // GEORGIAN CAPITAL LETTER ZEN
    526     { 0x10A7, 0x2D07 },  // GEORGIAN CAPITAL LETTER TAN
    527     { 0x10A8, 0x2D08 },  // GEORGIAN CAPITAL LETTER IN
    528     { 0x10A9, 0x2D09 },  // GEORGIAN CAPITAL LETTER KAN
    529     { 0x10AA, 0x2D0A },  // GEORGIAN CAPITAL LETTER LAS
    530     { 0x10AB, 0x2D0B },  // GEORGIAN CAPITAL LETTER MAN
    531     { 0x10AC, 0x2D0C },  // GEORGIAN CAPITAL LETTER NAR
    532     { 0x10AD, 0x2D0D },  // GEORGIAN CAPITAL LETTER ON
    533     { 0x10AE, 0x2D0E },  // GEORGIAN CAPITAL LETTER PAR
    534     { 0x10AF, 0x2D0F },  // GEORGIAN CAPITAL LETTER ZHAR
    535     { 0x10B0, 0x2D10 },  // GEORGIAN CAPITAL LETTER RAE
    536     { 0x10B1, 0x2D11 },  // GEORGIAN CAPITAL LETTER SAN
    537     { 0x10B2, 0x2D12 },  // GEORGIAN CAPITAL LETTER TAR
    538     { 0x10B3, 0x2D13 },  // GEORGIAN CAPITAL LETTER UN
    539     { 0x10B4, 0x2D14 },  // GEORGIAN CAPITAL LETTER PHAR
    540     { 0x10B5, 0x2D15 },  // GEORGIAN CAPITAL LETTER KHAR
    541     { 0x10B6, 0x2D16 },  // GEORGIAN CAPITAL LETTER GHAN
    542     { 0x10B7, 0x2D17 },  // GEORGIAN CAPITAL LETTER QAR
    543     { 0x10B8, 0x2D18 },  // GEORGIAN CAPITAL LETTER SHIN
    544     { 0x10B9, 0x2D19 },  // GEORGIAN CAPITAL LETTER CHIN
    545     { 0x10BA, 0x2D1A },  // GEORGIAN CAPITAL LETTER CAN
    546     { 0x10BB, 0x2D1B },  // GEORGIAN CAPITAL LETTER JIL
    547     { 0x10BC, 0x2D1C },  // GEORGIAN CAPITAL LETTER CIL
    548     { 0x10BD, 0x2D1D },  // GEORGIAN CAPITAL LETTER CHAR
    549     { 0x10BE, 0x2D1E },  // GEORGIAN CAPITAL LETTER XAN
    550     { 0x10BF, 0x2D1F },  // GEORGIAN CAPITAL LETTER JHAN
    551     { 0x10C0, 0x2D20 },  // GEORGIAN CAPITAL LETTER HAE
    552     { 0x10C1, 0x2D21 },  // GEORGIAN CAPITAL LETTER HE
    553     { 0x10C2, 0x2D22 },  // GEORGIAN CAPITAL LETTER HIE
    554     { 0x10C3, 0x2D23 },  // GEORGIAN CAPITAL LETTER WE
    555     { 0x10C4, 0x2D24 },  // GEORGIAN CAPITAL LETTER HAR
    556     { 0x10C5, 0x2D25 },  // GEORGIAN CAPITAL LETTER HOE
    557     { 0x1E00, 0x1E01 },  // LATIN CAPITAL LETTER A WITH RING BELOW
    558     { 0x1E02, 0x1E03 },  // LATIN CAPITAL LETTER B WITH DOT ABOVE
    559     { 0x1E04, 0x1E05 },  // LATIN CAPITAL LETTER B WITH DOT BELOW
    560     { 0x1E06, 0x1E07 },  // LATIN CAPITAL LETTER B WITH LINE BELOW
    561     { 0x1E08, 0x1E09 },  // LATIN CAPITAL LETTER C WITH CEDILLA AND ACUTE
    562     { 0x1E0A, 0x1E0B },  // LATIN CAPITAL LETTER D WITH DOT ABOVE
    563     { 0x1E0C, 0x1E0D },  // LATIN CAPITAL LETTER D WITH DOT BELOW
    564     { 0x1E0E, 0x1E0F },  // LATIN CAPITAL LETTER D WITH LINE BELOW
    565     { 0x1E10, 0x1E11 },  // LATIN CAPITAL LETTER D WITH CEDILLA
    566     { 0x1E12, 0x1E13 },  // LATIN CAPITAL LETTER D WITH CIRCUMFLEX BELOW
    567     { 0x1E14, 0x1E15 },  // LATIN CAPITAL LETTER E WITH MACRON AND GRAVE
    568     { 0x1E16, 0x1E17 },  // LATIN CAPITAL LETTER E WITH MACRON AND ACUTE
    569     { 0x1E18, 0x1E19 },  // LATIN CAPITAL LETTER E WITH CIRCUMFLEX BELOW
    570     { 0x1E1A, 0x1E1B },  // LATIN CAPITAL LETTER E WITH TILDE BELOW
    571     { 0x1E1C, 0x1E1D },  // LATIN CAPITAL LETTER E WITH CEDILLA AND BREVE
    572     { 0x1E1E, 0x1E1F },  // LATIN CAPITAL LETTER F WITH DOT ABOVE
    573     { 0x1E20, 0x1E21 },  // LATIN CAPITAL LETTER G WITH MACRON
    574     { 0x1E22, 0x1E23 },  // LATIN CAPITAL LETTER H WITH DOT ABOVE
    575     { 0x1E24, 0x1E25 },  // LATIN CAPITAL LETTER H WITH DOT BELOW
    576     { 0x1E26, 0x1E27 },  // LATIN CAPITAL LETTER H WITH DIAERESIS
    577     { 0x1E28, 0x1E29 },  // LATIN CAPITAL LETTER H WITH CEDILLA
    578     { 0x1E2A, 0x1E2B },  // LATIN CAPITAL LETTER H WITH BREVE BELOW
    579     { 0x1E2C, 0x1E2D },  // LATIN CAPITAL LETTER I WITH TILDE BELOW
    580     { 0x1E2E, 0x1E2F },  // LATIN CAPITAL LETTER I WITH DIAERESIS AND ACUTE
    581     { 0x1E30, 0x1E31 },  // LATIN CAPITAL LETTER K WITH ACUTE
    582     { 0x1E32, 0x1E33 },  // LATIN CAPITAL LETTER K WITH DOT BELOW
    583     { 0x1E34, 0x1E35 },  // LATIN CAPITAL LETTER K WITH LINE BELOW
    584     { 0x1E36, 0x1E37 },  // LATIN CAPITAL LETTER L WITH DOT BELOW
    585     { 0x1E38, 0x1E39 },  // LATIN CAPITAL LETTER L WITH DOT BELOW AND MACRON
    586     { 0x1E3A, 0x1E3B },  // LATIN CAPITAL LETTER L WITH LINE BELOW
    587     { 0x1E3C, 0x1E3D },  // LATIN CAPITAL LETTER L WITH CIRCUMFLEX BELOW
    588     { 0x1E3E, 0x1E3F },  // LATIN CAPITAL LETTER M WITH ACUTE
    589     { 0x1E40, 0x1E41 },  // LATIN CAPITAL LETTER M WITH DOT ABOVE
    590     { 0x1E42, 0x1E43 },  // LATIN CAPITAL LETTER M WITH DOT BELOW
    591     { 0x1E44, 0x1E45 },  // LATIN CAPITAL LETTER N WITH DOT ABOVE
    592     { 0x1E46, 0x1E47 },  // LATIN CAPITAL LETTER N WITH DOT BELOW
    593     { 0x1E48, 0x1E49 },  // LATIN CAPITAL LETTER N WITH LINE BELOW
    594     { 0x1E4A, 0x1E4B },  // LATIN CAPITAL LETTER N WITH CIRCUMFLEX BELOW
    595     { 0x1E4C, 0x1E4D },  // LATIN CAPITAL LETTER O WITH TILDE AND ACUTE
    596     { 0x1E4E, 0x1E4F },  // LATIN CAPITAL LETTER O WITH TILDE AND DIAERESIS
    597     { 0x1E50, 0x1E51 },  // LATIN CAPITAL LETTER O WITH MACRON AND GRAVE
    598     { 0x1E52, 0x1E53 },  // LATIN CAPITAL LETTER O WITH MACRON AND ACUTE
    599     { 0x1E54, 0x1E55 },  // LATIN CAPITAL LETTER P WITH ACUTE
    600     { 0x1E56, 0x1E57 },  // LATIN CAPITAL LETTER P WITH DOT ABOVE
    601     { 0x1E58, 0x1E59 },  // LATIN CAPITAL LETTER R WITH DOT ABOVE
    602     { 0x1E5A, 0x1E5B },  // LATIN CAPITAL LETTER R WITH DOT BELOW
    603     { 0x1E5C, 0x1E5D },  // LATIN CAPITAL LETTER R WITH DOT BELOW AND MACRON
    604     { 0x1E5E, 0x1E5F },  // LATIN CAPITAL LETTER R WITH LINE BELOW
    605     { 0x1E60, 0x1E61 },  // LATIN CAPITAL LETTER S WITH DOT ABOVE
    606     { 0x1E62, 0x1E63 },  // LATIN CAPITAL LETTER S WITH DOT BELOW
    607     { 0x1E64, 0x1E65 },  // LATIN CAPITAL LETTER S WITH ACUTE AND DOT ABOVE
    608     { 0x1E66, 0x1E67 },  // LATIN CAPITAL LETTER S WITH CARON AND DOT ABOVE
    609     { 0x1E68, 0x1E69 },  // LATIN CAPITAL LETTER S WITH DOT BELOW AND DOT ABOVE
    610     { 0x1E6A, 0x1E6B },  // LATIN CAPITAL LETTER T WITH DOT ABOVE
    611     { 0x1E6C, 0x1E6D },  // LATIN CAPITAL LETTER T WITH DOT BELOW
    612     { 0x1E6E, 0x1E6F },  // LATIN CAPITAL LETTER T WITH LINE BELOW
    613     { 0x1E70, 0x1E71 },  // LATIN CAPITAL LETTER T WITH CIRCUMFLEX BELOW
    614     { 0x1E72, 0x1E73 },  // LATIN CAPITAL LETTER U WITH DIAERESIS BELOW
    615     { 0x1E74, 0x1E75 },  // LATIN CAPITAL LETTER U WITH TILDE BELOW
    616     { 0x1E76, 0x1E77 },  // LATIN CAPITAL LETTER U WITH CIRCUMFLEX BELOW
    617     { 0x1E78, 0x1E79 },  // LATIN CAPITAL LETTER U WITH TILDE AND ACUTE
    618     { 0x1E7A, 0x1E7B },  // LATIN CAPITAL LETTER U WITH MACRON AND DIAERESIS
    619     { 0x1E7C, 0x1E7D },  // LATIN CAPITAL LETTER V WITH TILDE
    620     { 0x1E7E, 0x1E7F },  // LATIN CAPITAL LETTER V WITH DOT BELOW
    621     { 0x1E80, 0x1E81 },  // LATIN CAPITAL LETTER W WITH GRAVE
    622     { 0x1E82, 0x1E83 },  // LATIN CAPITAL LETTER W WITH ACUTE
    623     { 0x1E84, 0x1E85 },  // LATIN CAPITAL LETTER W WITH DIAERESIS
    624     { 0x1E86, 0x1E87 },  // LATIN CAPITAL LETTER W WITH DOT ABOVE
    625     { 0x1E88, 0x1E89 },  // LATIN CAPITAL LETTER W WITH DOT BELOW
    626     { 0x1E8A, 0x1E8B },  // LATIN CAPITAL LETTER X WITH DOT ABOVE
    627     { 0x1E8C, 0x1E8D },  // LATIN CAPITAL LETTER X WITH DIAERESIS
    628     { 0x1E8E, 0x1E8F },  // LATIN CAPITAL LETTER Y WITH DOT ABOVE
    629     { 0x1E90, 0x1E91 },  // LATIN CAPITAL LETTER Z WITH CIRCUMFLEX
    630     { 0x1E92, 0x1E93 },  // LATIN CAPITAL LETTER Z WITH DOT BELOW
    631     { 0x1E94, 0x1E95 },  // LATIN CAPITAL LETTER Z WITH LINE BELOW
    632     { 0x1E9E, 0x00DF },  // LATIN CAPITAL LETTER SHARP S
    633     { 0x1EA0, 0x1EA1 },  // LATIN CAPITAL LETTER A WITH DOT BELOW
    634     { 0x1EA2, 0x1EA3 },  // LATIN CAPITAL LETTER A WITH HOOK ABOVE
    635     { 0x1EA4, 0x1EA5 },  // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND ACUTE
    636     { 0x1EA6, 0x1EA7 },  // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND GRAVE
    637     { 0x1EA8, 0x1EA9 },  // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND HOOK ABOVE
    638     { 0x1EAA, 0x1EAB },  // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND TILDE
    639     { 0x1EAC, 0x1EAD },  // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND DOT BELOW
    640     { 0x1EAE, 0x1EAF },  // LATIN CAPITAL LETTER A WITH BREVE AND ACUTE
    641     { 0x1EB0, 0x1EB1 },  // LATIN CAPITAL LETTER A WITH BREVE AND GRAVE
    642     { 0x1EB2, 0x1EB3 },  // LATIN CAPITAL LETTER A WITH BREVE AND HOOK ABOVE
    643     { 0x1EB4, 0x1EB5 },  // LATIN CAPITAL LETTER A WITH BREVE AND TILDE
    644     { 0x1EB6, 0x1EB7 },  // LATIN CAPITAL LETTER A WITH BREVE AND DOT BELOW
    645     { 0x1EB8, 0x1EB9 },  // LATIN CAPITAL LETTER E WITH DOT BELOW
    646     { 0x1EBA, 0x1EBB },  // LATIN CAPITAL LETTER E WITH HOOK ABOVE
    647     { 0x1EBC, 0x1EBD },  // LATIN CAPITAL LETTER E WITH TILDE
    648     { 0x1EBE, 0x1EBF },  // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND ACUTE
    649     { 0x1EC0, 0x1EC1 },  // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND GRAVE
    650     { 0x1EC2, 0x1EC3 },  // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND HOOK ABOVE
    651     { 0x1EC4, 0x1EC5 },  // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND TILDE
    652     { 0x1EC6, 0x1EC7 },  // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND DOT BELOW
    653     { 0x1EC8, 0x1EC9 },  // LATIN CAPITAL LETTER I WITH HOOK ABOVE
    654     { 0x1ECA, 0x1ECB },  // LATIN CAPITAL LETTER I WITH DOT BELOW
    655     { 0x1ECC, 0x1ECD },  // LATIN CAPITAL LETTER O WITH DOT BELOW
    656     { 0x1ECE, 0x1ECF },  // LATIN CAPITAL LETTER O WITH HOOK ABOVE
    657     { 0x1ED0, 0x1ED1 },  // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND ACUTE
    658     { 0x1ED2, 0x1ED3 },  // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND GRAVE
    659     { 0x1ED4, 0x1ED5 },  // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND HOOK ABOVE
    660     { 0x1ED6, 0x1ED7 },  // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND TILDE
    661     { 0x1ED8, 0x1ED9 },  // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND DOT BELOW
    662     { 0x1EDA, 0x1EDB },  // LATIN CAPITAL LETTER O WITH HORN AND ACUTE
    663     { 0x1EDC, 0x1EDD },  // LATIN CAPITAL LETTER O WITH HORN AND GRAVE
    664     { 0x1EDE, 0x1EDF },  // LATIN CAPITAL LETTER O WITH HORN AND HOOK ABOVE
    665     { 0x1EE0, 0x1EE1 },  // LATIN CAPITAL LETTER O WITH HORN AND TILDE
    666     { 0x1EE2, 0x1EE3 },  // LATIN CAPITAL LETTER O WITH HORN AND DOT BELOW
    667     { 0x1EE4, 0x1EE5 },  // LATIN CAPITAL LETTER U WITH DOT BELOW
    668     { 0x1EE6, 0x1EE7 },  // LATIN CAPITAL LETTER U WITH HOOK ABOVE
    669     { 0x1EE8, 0x1EE9 },  // LATIN CAPITAL LETTER U WITH HORN AND ACUTE
    670     { 0x1EEA, 0x1EEB },  // LATIN CAPITAL LETTER U WITH HORN AND GRAVE
    671     { 0x1EEC, 0x1EED },  // LATIN CAPITAL LETTER U WITH HORN AND HOOK ABOVE
    672     { 0x1EEE, 0x1EEF },  // LATIN CAPITAL LETTER U WITH HORN AND TILDE
    673     { 0x1EF0, 0x1EF1 },  // LATIN CAPITAL LETTER U WITH HORN AND DOT BELOW
    674     { 0x1EF2, 0x1EF3 },  // LATIN CAPITAL LETTER Y WITH GRAVE
    675     { 0x1EF4, 0x1EF5 },  // LATIN CAPITAL LETTER Y WITH DOT BELOW
    676     { 0x1EF6, 0x1EF7 },  // LATIN CAPITAL LETTER Y WITH HOOK ABOVE
    677     { 0x1EF8, 0x1EF9 },  // LATIN CAPITAL LETTER Y WITH TILDE
    678     { 0x1EFA, 0x1EFB },  // LATIN CAPITAL LETTER MIDDLE-WELSH LL
    679     { 0x1EFC, 0x1EFD },  // LATIN CAPITAL LETTER MIDDLE-WELSH V
    680     { 0x1EFE, 0x1EFF },  // LATIN CAPITAL LETTER Y WITH LOOP
    681     { 0x1F08, 0x1F00 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI
    682     { 0x1F09, 0x1F01 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA
    683     { 0x1F0A, 0x1F02 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA
    684     { 0x1F0B, 0x1F03 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA
    685     { 0x1F0C, 0x1F04 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA
    686     { 0x1F0D, 0x1F05 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA
    687     { 0x1F0E, 0x1F06 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI
    688     { 0x1F0F, 0x1F07 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI
    689     { 0x1F18, 0x1F10 },  // GREEK CAPITAL LETTER EPSILON WITH PSILI
    690     { 0x1F19, 0x1F11 },  // GREEK CAPITAL LETTER EPSILON WITH DASIA
    691     { 0x1F1A, 0x1F12 },  // GREEK CAPITAL LETTER EPSILON WITH PSILI AND VARIA
    692     { 0x1F1B, 0x1F13 },  // GREEK CAPITAL LETTER EPSILON WITH DASIA AND VARIA
    693     { 0x1F1C, 0x1F14 },  // GREEK CAPITAL LETTER EPSILON WITH PSILI AND OXIA
    694     { 0x1F1D, 0x1F15 },  // GREEK CAPITAL LETTER EPSILON WITH DASIA AND OXIA
    695     { 0x1F28, 0x1F20 },  // GREEK CAPITAL LETTER ETA WITH PSILI
    696     { 0x1F29, 0x1F21 },  // GREEK CAPITAL LETTER ETA WITH DASIA
    697     { 0x1F2A, 0x1F22 },  // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA
    698     { 0x1F2B, 0x1F23 },  // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA
    699     { 0x1F2C, 0x1F24 },  // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA
    700     { 0x1F2D, 0x1F25 },  // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA
    701     { 0x1F2E, 0x1F26 },  // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI
    702     { 0x1F2F, 0x1F27 },  // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI
    703     { 0x1F38, 0x1F30 },  // GREEK CAPITAL LETTER IOTA WITH PSILI
    704     { 0x1F39, 0x1F31 },  // GREEK CAPITAL LETTER IOTA WITH DASIA
    705     { 0x1F3A, 0x1F32 },  // GREEK CAPITAL LETTER IOTA WITH PSILI AND VARIA
    706     { 0x1F3B, 0x1F33 },  // GREEK CAPITAL LETTER IOTA WITH DASIA AND VARIA
    707     { 0x1F3C, 0x1F34 },  // GREEK CAPITAL LETTER IOTA WITH PSILI AND OXIA
    708     { 0x1F3D, 0x1F35 },  // GREEK CAPITAL LETTER IOTA WITH DASIA AND OXIA
    709     { 0x1F3E, 0x1F36 },  // GREEK CAPITAL LETTER IOTA WITH PSILI AND PERISPOMENI
    710     { 0x1F3F, 0x1F37 },  // GREEK CAPITAL LETTER IOTA WITH DASIA AND PERISPOMENI
    711     { 0x1F48, 0x1F40 },  // GREEK CAPITAL LETTER OMICRON WITH PSILI
    712     { 0x1F49, 0x1F41 },  // GREEK CAPITAL LETTER OMICRON WITH DASIA
    713     { 0x1F4A, 0x1F42 },  // GREEK CAPITAL LETTER OMICRON WITH PSILI AND VARIA
    714     { 0x1F4B, 0x1F43 },  // GREEK CAPITAL LETTER OMICRON WITH DASIA AND VARIA
    715     { 0x1F4C, 0x1F44 },  // GREEK CAPITAL LETTER OMICRON WITH PSILI AND OXIA
    716     { 0x1F4D, 0x1F45 },  // GREEK CAPITAL LETTER OMICRON WITH DASIA AND OXIA
    717     { 0x1F59, 0x1F51 },  // GREEK CAPITAL LETTER UPSILON WITH DASIA
    718     { 0x1F5B, 0x1F53 },  // GREEK CAPITAL LETTER UPSILON WITH DASIA AND VARIA
    719     { 0x1F5D, 0x1F55 },  // GREEK CAPITAL LETTER UPSILON WITH DASIA AND OXIA
    720     { 0x1F5F, 0x1F57 },  // GREEK CAPITAL LETTER UPSILON WITH DASIA AND PERISPOMENI
    721     { 0x1F68, 0x1F60 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI
    722     { 0x1F69, 0x1F61 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA
    723     { 0x1F6A, 0x1F62 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA
    724     { 0x1F6B, 0x1F63 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA
    725     { 0x1F6C, 0x1F64 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA
    726     { 0x1F6D, 0x1F65 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA
    727     { 0x1F6E, 0x1F66 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI
    728     { 0x1F6F, 0x1F67 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI
    729     { 0x1F88, 0x1F80 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PROSGEGRAMMENI
    730     { 0x1F89, 0x1F81 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PROSGEGRAMMENI
    731     { 0x1F8A, 0x1F82 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA AND PROSGEGRAMMENI
    732     { 0x1F8B, 0x1F83 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA AND PROSGEGRAMMENI
    733     { 0x1F8C, 0x1F84 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA AND PROSGEGRAMMENI
    734     { 0x1F8D, 0x1F85 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA AND PROSGEGRAMMENI
    735     { 0x1F8E, 0x1F86 },  // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
    736     { 0x1F8F, 0x1F87 },  // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
    737     { 0x1F98, 0x1F90 },  // GREEK CAPITAL LETTER ETA WITH PSILI AND PROSGEGRAMMENI
    738     { 0x1F99, 0x1F91 },  // GREEK CAPITAL LETTER ETA WITH DASIA AND PROSGEGRAMMENI
    739     { 0x1F9A, 0x1F92 },  // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA AND PROSGEGRAMMENI
    740     { 0x1F9B, 0x1F93 },  // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA AND PROSGEGRAMMENI
    741     { 0x1F9C, 0x1F94 },  // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA AND PROSGEGRAMMENI
    742     { 0x1F9D, 0x1F95 },  // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA AND PROSGEGRAMMENI
    743     { 0x1F9E, 0x1F96 },  // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
    744     { 0x1F9F, 0x1F97 },  // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
    745     { 0x1FA8, 0x1FA0 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PROSGEGRAMMENI
    746     { 0x1FA9, 0x1FA1 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PROSGEGRAMMENI
    747     { 0x1FAA, 0x1FA2 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA AND PROSGEGRAMMENI
    748     { 0x1FAB, 0x1FA3 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA AND PROSGEGRAMMENI
    749     { 0x1FAC, 0x1FA4 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA AND PROSGEGRAMMENI
    750     { 0x1FAD, 0x1FA5 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA AND PROSGEGRAMMENI
    751     { 0x1FAE, 0x1FA6 },  // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
    752     { 0x1FAF, 0x1FA7 },  // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
    753     { 0x1FB8, 0x1FB0 },  // GREEK CAPITAL LETTER ALPHA WITH VRACHY
    754     { 0x1FB9, 0x1FB1 },  // GREEK CAPITAL LETTER ALPHA WITH MACRON
    755     { 0x1FBA, 0x1F70 },  // GREEK CAPITAL LETTER ALPHA WITH VARIA
    756     { 0x1FBB, 0x1F71 },  // GREEK CAPITAL LETTER ALPHA WITH OXIA
    757     { 0x1FBC, 0x1FB3 },  // GREEK CAPITAL LETTER ALPHA WITH PROSGEGRAMMENI
    758     { 0x1FC8, 0x1F72 },  // GREEK CAPITAL LETTER EPSILON WITH VARIA
    759     { 0x1FC9, 0x1F73 },  // GREEK CAPITAL LETTER EPSILON WITH OXIA
    760     { 0x1FCA, 0x1F74 },  // GREEK CAPITAL LETTER ETA WITH VARIA
    761     { 0x1FCB, 0x1F75 },  // GREEK CAPITAL LETTER ETA WITH OXIA
    762     { 0x1FCC, 0x1FC3 },  // GREEK CAPITAL LETTER ETA WITH PROSGEGRAMMENI
    763     { 0x1FD8, 0x1FD0 },  // GREEK CAPITAL LETTER IOTA WITH VRACHY
    764     { 0x1FD9, 0x1FD1 },  // GREEK CAPITAL LETTER IOTA WITH MACRON
    765     { 0x1FDA, 0x1F76 },  // GREEK CAPITAL LETTER IOTA WITH VARIA
    766     { 0x1FDB, 0x1F77 },  // GREEK CAPITAL LETTER IOTA WITH OXIA
    767     { 0x1FE8, 0x1FE0 },  // GREEK CAPITAL LETTER UPSILON WITH VRACHY
    768     { 0x1FE9, 0x1FE1 },  // GREEK CAPITAL LETTER UPSILON WITH MACRON
    769     { 0x1FEA, 0x1F7A },  // GREEK CAPITAL LETTER UPSILON WITH VARIA
    770     { 0x1FEB, 0x1F7B },  // GREEK CAPITAL LETTER UPSILON WITH OXIA
    771     { 0x1FEC, 0x1FE5 },  // GREEK CAPITAL LETTER RHO WITH DASIA
    772     { 0x1FF8, 0x1F78 },  // GREEK CAPITAL LETTER OMICRON WITH VARIA
    773     { 0x1FF9, 0x1F79 },  // GREEK CAPITAL LETTER OMICRON WITH OXIA
    774     { 0x1FFA, 0x1F7C },  // GREEK CAPITAL LETTER OMEGA WITH VARIA
    775     { 0x1FFB, 0x1F7D },  // GREEK CAPITAL LETTER OMEGA WITH OXIA
    776     { 0x1FFC, 0x1FF3 },  // GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI
    777     { 0x2126, 0x03C9 },  // OHM SIGN
    778     { 0x212A, 0x006B },  // KELVIN SIGN
    779     { 0x212B, 0x00E5 },  // ANGSTROM SIGN
    780     { 0x2132, 0x214E },  // TURNED CAPITAL F
    781     { 0x2160, 0x2170 },  // ROMAN NUMERAL ONE
    782     { 0x2161, 0x2171 },  // ROMAN NUMERAL TWO
    783     { 0x2162, 0x2172 },  // ROMAN NUMERAL THREE
    784     { 0x2163, 0x2173 },  // ROMAN NUMERAL FOUR
    785     { 0x2164, 0x2174 },  // ROMAN NUMERAL FIVE
    786     { 0x2165, 0x2175 },  // ROMAN NUMERAL SIX
    787     { 0x2166, 0x2176 },  // ROMAN NUMERAL SEVEN
    788     { 0x2167, 0x2177 },  // ROMAN NUMERAL EIGHT
    789     { 0x2168, 0x2178 },  // ROMAN NUMERAL NINE
    790     { 0x2169, 0x2179 },  // ROMAN NUMERAL TEN
    791     { 0x216A, 0x217A },  // ROMAN NUMERAL ELEVEN
    792     { 0x216B, 0x217B },  // ROMAN NUMERAL TWELVE
    793     { 0x216C, 0x217C },  // ROMAN NUMERAL FIFTY
    794     { 0x216D, 0x217D },  // ROMAN NUMERAL ONE HUNDRED
    795     { 0x216E, 0x217E },  // ROMAN NUMERAL FIVE HUNDRED
    796     { 0x216F, 0x217F },  // ROMAN NUMERAL ONE THOUSAND
    797     { 0x2183, 0x2184 },  // ROMAN NUMERAL REVERSED ONE HUNDRED
    798     { 0x24B6, 0x24D0 },  // CIRCLED LATIN CAPITAL LETTER A
    799     { 0x24B7, 0x24D1 },  // CIRCLED LATIN CAPITAL LETTER B
    800     { 0x24B8, 0x24D2 },  // CIRCLED LATIN CAPITAL LETTER C
    801     { 0x24B9, 0x24D3 },  // CIRCLED LATIN CAPITAL LETTER D
    802     { 0x24BA, 0x24D4 },  // CIRCLED LATIN CAPITAL LETTER E
    803     { 0x24BB, 0x24D5 },  // CIRCLED LATIN CAPITAL LETTER F
    804     { 0x24BC, 0x24D6 },  // CIRCLED LATIN CAPITAL LETTER G
    805     { 0x24BD, 0x24D7 },  // CIRCLED LATIN CAPITAL LETTER H
    806     { 0x24BE, 0x24D8 },  // CIRCLED LATIN CAPITAL LETTER I
    807     { 0x24BF, 0x24D9 },  // CIRCLED LATIN CAPITAL LETTER J
    808     { 0x24C0, 0x24DA },  // CIRCLED LATIN CAPITAL LETTER K
    809     { 0x24C1, 0x24DB },  // CIRCLED LATIN CAPITAL LETTER L
    810     { 0x24C2, 0x24DC },  // CIRCLED LATIN CAPITAL LETTER M
    811     { 0x24C3, 0x24DD },  // CIRCLED LATIN CAPITAL LETTER N
    812     { 0x24C4, 0x24DE },  // CIRCLED LATIN CAPITAL LETTER O
    813     { 0x24C5, 0x24DF },  // CIRCLED LATIN CAPITAL LETTER P
    814     { 0x24C6, 0x24E0 },  // CIRCLED LATIN CAPITAL LETTER Q
    815     { 0x24C7, 0x24E1 },  // CIRCLED LATIN CAPITAL LETTER R
    816     { 0x24C8, 0x24E2 },  // CIRCLED LATIN CAPITAL LETTER S
    817     { 0x24C9, 0x24E3 },  // CIRCLED LATIN CAPITAL LETTER T
    818     { 0x24CA, 0x24E4 },  // CIRCLED LATIN CAPITAL LETTER U
    819     { 0x24CB, 0x24E5 },  // CIRCLED LATIN CAPITAL LETTER V
    820     { 0x24CC, 0x24E6 },  // CIRCLED LATIN CAPITAL LETTER W
    821     { 0x24CD, 0x24E7 },  // CIRCLED LATIN CAPITAL LETTER X
    822     { 0x24CE, 0x24E8 },  // CIRCLED LATIN CAPITAL LETTER Y
    823     { 0x24CF, 0x24E9 },  // CIRCLED LATIN CAPITAL LETTER Z
    824     { 0x2C00, 0x2C30 },  // GLAGOLITIC CAPITAL LETTER AZU
    825     { 0x2C01, 0x2C31 },  // GLAGOLITIC CAPITAL LETTER BUKY
    826     { 0x2C02, 0x2C32 },  // GLAGOLITIC CAPITAL LETTER VEDE
    827     { 0x2C03, 0x2C33 },  // GLAGOLITIC CAPITAL LETTER GLAGOLI
    828     { 0x2C04, 0x2C34 },  // GLAGOLITIC CAPITAL LETTER DOBRO
    829     { 0x2C05, 0x2C35 },  // GLAGOLITIC CAPITAL LETTER YESTU
    830     { 0x2C06, 0x2C36 },  // GLAGOLITIC CAPITAL LETTER ZHIVETE
    831     { 0x2C07, 0x2C37 },  // GLAGOLITIC CAPITAL LETTER DZELO
    832     { 0x2C08, 0x2C38 },  // GLAGOLITIC CAPITAL LETTER ZEMLJA
    833     { 0x2C09, 0x2C39 },  // GLAGOLITIC CAPITAL LETTER IZHE
    834     { 0x2C0A, 0x2C3A },  // GLAGOLITIC CAPITAL LETTER INITIAL IZHE
    835     { 0x2C0B, 0x2C3B },  // GLAGOLITIC CAPITAL LETTER I
    836     { 0x2C0C, 0x2C3C },  // GLAGOLITIC CAPITAL LETTER DJERVI
    837     { 0x2C0D, 0x2C3D },  // GLAGOLITIC CAPITAL LETTER KAKO
    838     { 0x2C0E, 0x2C3E },  // GLAGOLITIC CAPITAL LETTER LJUDIJE
    839     { 0x2C0F, 0x2C3F },  // GLAGOLITIC CAPITAL LETTER MYSLITE
    840     { 0x2C10, 0x2C40 },  // GLAGOLITIC CAPITAL LETTER NASHI
    841     { 0x2C11, 0x2C41 },  // GLAGOLITIC CAPITAL LETTER ONU
    842     { 0x2C12, 0x2C42 },  // GLAGOLITIC CAPITAL LETTER POKOJI
    843     { 0x2C13, 0x2C43 },  // GLAGOLITIC CAPITAL LETTER RITSI
    844     { 0x2C14, 0x2C44 },  // GLAGOLITIC CAPITAL LETTER SLOVO
    845     { 0x2C15, 0x2C45 },  // GLAGOLITIC CAPITAL LETTER TVRIDO
    846     { 0x2C16, 0x2C46 },  // GLAGOLITIC CAPITAL LETTER UKU
    847     { 0x2C17, 0x2C47 },  // GLAGOLITIC CAPITAL LETTER FRITU
    848     { 0x2C18, 0x2C48 },  // GLAGOLITIC CAPITAL LETTER HERU
    849     { 0x2C19, 0x2C49 },  // GLAGOLITIC CAPITAL LETTER OTU
    850     { 0x2C1A, 0x2C4A },  // GLAGOLITIC CAPITAL LETTER PE
    851     { 0x2C1B, 0x2C4B },  // GLAGOLITIC CAPITAL LETTER SHTA
    852     { 0x2C1C, 0x2C4C },  // GLAGOLITIC CAPITAL LETTER TSI
    853     { 0x2C1D, 0x2C4D },  // GLAGOLITIC CAPITAL LETTER CHRIVI
    854     { 0x2C1E, 0x2C4E },  // GLAGOLITIC CAPITAL LETTER SHA
    855     { 0x2C1F, 0x2C4F },  // GLAGOLITIC CAPITAL LETTER YERU
    856     { 0x2C20, 0x2C50 },  // GLAGOLITIC CAPITAL LETTER YERI
    857     { 0x2C21, 0x2C51 },  // GLAGOLITIC CAPITAL LETTER YATI
    858     { 0x2C22, 0x2C52 },  // GLAGOLITIC CAPITAL LETTER SPIDERY HA
    859     { 0x2C23, 0x2C53 },  // GLAGOLITIC CAPITAL LETTER YU
    860     { 0x2C24, 0x2C54 },  // GLAGOLITIC CAPITAL LETTER SMALL YUS
    861     { 0x2C25, 0x2C55 },  // GLAGOLITIC CAPITAL LETTER SMALL YUS WITH TAIL
    862     { 0x2C26, 0x2C56 },  // GLAGOLITIC CAPITAL LETTER YO
    863     { 0x2C27, 0x2C57 },  // GLAGOLITIC CAPITAL LETTER IOTATED SMALL YUS
    864     { 0x2C28, 0x2C58 },  // GLAGOLITIC CAPITAL LETTER BIG YUS
    865     { 0x2C29, 0x2C59 },  // GLAGOLITIC CAPITAL LETTER IOTATED BIG YUS
    866     { 0x2C2A, 0x2C5A },  // GLAGOLITIC CAPITAL LETTER FITA
    867     { 0x2C2B, 0x2C5B },  // GLAGOLITIC CAPITAL LETTER IZHITSA
    868     { 0x2C2C, 0x2C5C },  // GLAGOLITIC CAPITAL LETTER SHTAPIC
    869     { 0x2C2D, 0x2C5D },  // GLAGOLITIC CAPITAL LETTER TROKUTASTI A
    870     { 0x2C2E, 0x2C5E },  // GLAGOLITIC CAPITAL LETTER LATINATE MYSLITE
    871     { 0x2C60, 0x2C61 },  // LATIN CAPITAL LETTER L WITH DOUBLE BAR
    872     { 0x2C62, 0x026B },  // LATIN CAPITAL LETTER L WITH MIDDLE TILDE
    873     { 0x2C63, 0x1D7D },  // LATIN CAPITAL LETTER P WITH STROKE
    874     { 0x2C64, 0x027D },  // LATIN CAPITAL LETTER R WITH TAIL
    875     { 0x2C67, 0x2C68 },  // LATIN CAPITAL LETTER H WITH DESCENDER
    876     { 0x2C69, 0x2C6A },  // LATIN CAPITAL LETTER K WITH DESCENDER
    877     { 0x2C6B, 0x2C6C },  // LATIN CAPITAL LETTER Z WITH DESCENDER
    878     { 0x2C6D, 0x0251 },  // LATIN CAPITAL LETTER ALPHA
    879     { 0x2C6E, 0x0271 },  // LATIN CAPITAL LETTER M WITH HOOK
    880     { 0x2C6F, 0x0250 },  // LATIN CAPITAL LETTER TURNED A
    881     { 0x2C70, 0x0252 },  // LATIN CAPITAL LETTER TURNED ALPHA
    882     { 0x2C72, 0x2C73 },  // LATIN CAPITAL LETTER W WITH HOOK
    883     { 0x2C75, 0x2C76 },  // LATIN CAPITAL LETTER HALF H
    884     { 0x2C7E, 0x023F },  // LATIN CAPITAL LETTER S WITH SWASH TAIL
    885     { 0x2C7F, 0x0240 },  // LATIN CAPITAL LETTER Z WITH SWASH TAIL
    886     { 0x2C80, 0x2C81 },  // COPTIC CAPITAL LETTER ALFA
    887     { 0x2C82, 0x2C83 },  // COPTIC CAPITAL LETTER VIDA
    888     { 0x2C84, 0x2C85 },  // COPTIC CAPITAL LETTER GAMMA
    889     { 0x2C86, 0x2C87 },  // COPTIC CAPITAL LETTER DALDA
    890     { 0x2C88, 0x2C89 },  // COPTIC CAPITAL LETTER EIE
    891     { 0x2C8A, 0x2C8B },  // COPTIC CAPITAL LETTER SOU
    892     { 0x2C8C, 0x2C8D },  // COPTIC CAPITAL LETTER ZATA
    893     { 0x2C8E, 0x2C8F },  // COPTIC CAPITAL LETTER HATE
    894     { 0x2C90, 0x2C91 },  // COPTIC CAPITAL LETTER THETHE
    895     { 0x2C92, 0x2C93 },  // COPTIC CAPITAL LETTER IAUDA
    896     { 0x2C94, 0x2C95 },  // COPTIC CAPITAL LETTER KAPA
    897     { 0x2C96, 0x2C97 },  // COPTIC CAPITAL LETTER LAULA
    898     { 0x2C98, 0x2C99 },  // COPTIC CAPITAL LETTER MI
    899     { 0x2C9A, 0x2C9B },  // COPTIC CAPITAL LETTER NI
    900     { 0x2C9C, 0x2C9D },  // COPTIC CAPITAL LETTER KSI
    901     { 0x2C9E, 0x2C9F },  // COPTIC CAPITAL LETTER O
    902     { 0x2CA0, 0x2CA1 },  // COPTIC CAPITAL LETTER PI
    903     { 0x2CA2, 0x2CA3 },  // COPTIC CAPITAL LETTER RO
    904     { 0x2CA4, 0x2CA5 },  // COPTIC CAPITAL LETTER SIMA
    905     { 0x2CA6, 0x2CA7 },  // COPTIC CAPITAL LETTER TAU
    906     { 0x2CA8, 0x2CA9 },  // COPTIC CAPITAL LETTER UA
    907     { 0x2CAA, 0x2CAB },  // COPTIC CAPITAL LETTER FI
    908     { 0x2CAC, 0x2CAD },  // COPTIC CAPITAL LETTER KHI
    909     { 0x2CAE, 0x2CAF },  // COPTIC CAPITAL LETTER PSI
    910     { 0x2CB0, 0x2CB1 },  // COPTIC CAPITAL LETTER OOU
    911     { 0x2CB2, 0x2CB3 },  // COPTIC CAPITAL LETTER DIALECT-P ALEF
    912     { 0x2CB4, 0x2CB5 },  // COPTIC CAPITAL LETTER OLD COPTIC AIN
    913     { 0x2CB6, 0x2CB7 },  // COPTIC CAPITAL LETTER CRYPTOGRAMMIC EIE
    914     { 0x2CB8, 0x2CB9 },  // COPTIC CAPITAL LETTER DIALECT-P KAPA
    915     { 0x2CBA, 0x2CBB },  // COPTIC CAPITAL LETTER DIALECT-P NI
    916     { 0x2CBC, 0x2CBD },  // COPTIC CAPITAL LETTER CRYPTOGRAMMIC NI
    917     { 0x2CBE, 0x2CBF },  // COPTIC CAPITAL LETTER OLD COPTIC OOU
    918     { 0x2CC0, 0x2CC1 },  // COPTIC CAPITAL LETTER SAMPI
    919     { 0x2CC2, 0x2CC3 },  // COPTIC CAPITAL LETTER CROSSED SHEI
    920     { 0x2CC4, 0x2CC5 },  // COPTIC CAPITAL LETTER OLD COPTIC SHEI
    921     { 0x2CC6, 0x2CC7 },  // COPTIC CAPITAL LETTER OLD COPTIC ESH
    922     { 0x2CC8, 0x2CC9 },  // COPTIC CAPITAL LETTER AKHMIMIC KHEI
    923     { 0x2CCA, 0x2CCB },  // COPTIC CAPITAL LETTER DIALECT-P HORI
    924     { 0x2CCC, 0x2CCD },  // COPTIC CAPITAL LETTER OLD COPTIC HORI
    925     { 0x2CCE, 0x2CCF },  // COPTIC CAPITAL LETTER OLD COPTIC HA
    926     { 0x2CD0, 0x2CD1 },  // COPTIC CAPITAL LETTER L-SHAPED HA
    927     { 0x2CD2, 0x2CD3 },  // COPTIC CAPITAL LETTER OLD COPTIC HEI
    928     { 0x2CD4, 0x2CD5 },  // COPTIC CAPITAL LETTER OLD COPTIC HAT
    929     { 0x2CD6, 0x2CD7 },  // COPTIC CAPITAL LETTER OLD COPTIC GANGIA
    930     { 0x2CD8, 0x2CD9 },  // COPTIC CAPITAL LETTER OLD COPTIC DJA
    931     { 0x2CDA, 0x2CDB },  // COPTIC CAPITAL LETTER OLD COPTIC SHIMA
    932     { 0x2CDC, 0x2CDD },  // COPTIC CAPITAL LETTER OLD NUBIAN SHIMA
    933     { 0x2CDE, 0x2CDF },  // COPTIC CAPITAL LETTER OLD NUBIAN NGI
    934     { 0x2CE0, 0x2CE1 },  // COPTIC CAPITAL LETTER OLD NUBIAN NYI
    935     { 0x2CE2, 0x2CE3 },  // COPTIC CAPITAL LETTER OLD NUBIAN WAU
    936     { 0x2CEB, 0x2CEC },  // COPTIC CAPITAL LETTER CRYPTOGRAMMIC SHEI
    937     { 0x2CED, 0x2CEE },  // COPTIC CAPITAL LETTER CRYPTOGRAMMIC GANGIA
    938     { 0xA640, 0xA641 },  // CYRILLIC CAPITAL LETTER ZEMLYA
    939     { 0xA642, 0xA643 },  // CYRILLIC CAPITAL LETTER DZELO
    940     { 0xA644, 0xA645 },  // CYRILLIC CAPITAL LETTER REVERSED DZE
    941     { 0xA646, 0xA647 },  // CYRILLIC CAPITAL LETTER IOTA
    942     { 0xA648, 0xA649 },  // CYRILLIC CAPITAL LETTER DJERV
    943     { 0xA64A, 0xA64B },  // CYRILLIC CAPITAL LETTER MONOGRAPH UK
    944     { 0xA64C, 0xA64D },  // CYRILLIC CAPITAL LETTER BROAD OMEGA
    945     { 0xA64E, 0xA64F },  // CYRILLIC CAPITAL LETTER NEUTRAL YER
    946     { 0xA650, 0xA651 },  // CYRILLIC CAPITAL LETTER YERU WITH BACK YER
    947     { 0xA652, 0xA653 },  // CYRILLIC CAPITAL LETTER IOTIFIED YAT
    948     { 0xA654, 0xA655 },  // CYRILLIC CAPITAL LETTER REVERSED YU
    949     { 0xA656, 0xA657 },  // CYRILLIC CAPITAL LETTER IOTIFIED A
    950     { 0xA658, 0xA659 },  // CYRILLIC CAPITAL LETTER CLOSED LITTLE YUS
    951     { 0xA65A, 0xA65B },  // CYRILLIC CAPITAL LETTER BLENDED YUS
    952     { 0xA65C, 0xA65D },  // CYRILLIC CAPITAL LETTER IOTIFIED CLOSED LITTLE YUS
    953     { 0xA65E, 0xA65F },  // CYRILLIC CAPITAL LETTER YN
    954     { 0xA660, 0xA661 },  // CYRILLIC CAPITAL LETTER REVERSED TSE
    955     { 0xA662, 0xA663 },  // CYRILLIC CAPITAL LETTER SOFT DE
    956     { 0xA664, 0xA665 },  // CYRILLIC CAPITAL LETTER SOFT EL
    957     { 0xA666, 0xA667 },  // CYRILLIC CAPITAL LETTER SOFT EM
    958     { 0xA668, 0xA669 },  // CYRILLIC CAPITAL LETTER MONOCULAR O
    959     { 0xA66A, 0xA66B },  // CYRILLIC CAPITAL LETTER BINOCULAR O
    960     { 0xA66C, 0xA66D },  // CYRILLIC CAPITAL LETTER DOUBLE MONOCULAR O
    961     { 0xA680, 0xA681 },  // CYRILLIC CAPITAL LETTER DWE
    962     { 0xA682, 0xA683 },  // CYRILLIC CAPITAL LETTER DZWE
    963     { 0xA684, 0xA685 },  // CYRILLIC CAPITAL LETTER ZHWE
    964     { 0xA686, 0xA687 },  // CYRILLIC CAPITAL LETTER CCHE
    965     { 0xA688, 0xA689 },  // CYRILLIC CAPITAL LETTER DZZE
    966     { 0xA68A, 0xA68B },  // CYRILLIC CAPITAL LETTER TE WITH MIDDLE HOOK
    967     { 0xA68C, 0xA68D },  // CYRILLIC CAPITAL LETTER TWE
    968     { 0xA68E, 0xA68F },  // CYRILLIC CAPITAL LETTER TSWE
    969     { 0xA690, 0xA691 },  // CYRILLIC CAPITAL LETTER TSSE
    970     { 0xA692, 0xA693 },  // CYRILLIC CAPITAL LETTER TCHE
    971     { 0xA694, 0xA695 },  // CYRILLIC CAPITAL LETTER HWE
    972     { 0xA696, 0xA697 },  // CYRILLIC CAPITAL LETTER SHWE
    973     { 0xA722, 0xA723 },  // LATIN CAPITAL LETTER EGYPTOLOGICAL ALEF
    974     { 0xA724, 0xA725 },  // LATIN CAPITAL LETTER EGYPTOLOGICAL AIN
    975     { 0xA726, 0xA727 },  // LATIN CAPITAL LETTER HENG
    976     { 0xA728, 0xA729 },  // LATIN CAPITAL LETTER TZ
    977     { 0xA72A, 0xA72B },  // LATIN CAPITAL LETTER TRESILLO
    978     { 0xA72C, 0xA72D },  // LATIN CAPITAL LETTER CUATRILLO
    979     { 0xA72E, 0xA72F },  // LATIN CAPITAL LETTER CUATRILLO WITH COMMA
    980     { 0xA732, 0xA733 },  // LATIN CAPITAL LETTER AA
    981     { 0xA734, 0xA735 },  // LATIN CAPITAL LETTER AO
    982     { 0xA736, 0xA737 },  // LATIN CAPITAL LETTER AU
    983     { 0xA738, 0xA739 },  // LATIN CAPITAL LETTER AV
    984     { 0xA73A, 0xA73B },  // LATIN CAPITAL LETTER AV WITH HORIZONTAL BAR
    985     { 0xA73C, 0xA73D },  // LATIN CAPITAL LETTER AY
    986     { 0xA73E, 0xA73F },  // LATIN CAPITAL LETTER REVERSED C WITH DOT
    987     { 0xA740, 0xA741 },  // LATIN CAPITAL LETTER K WITH STROKE
    988     { 0xA742, 0xA743 },  // LATIN CAPITAL LETTER K WITH DIAGONAL STROKE
    989     { 0xA744, 0xA745 },  // LATIN CAPITAL LETTER K WITH STROKE AND DIAGONAL STROKE
    990     { 0xA746, 0xA747 },  // LATIN CAPITAL LETTER BROKEN L
    991     { 0xA748, 0xA749 },  // LATIN CAPITAL LETTER L WITH HIGH STROKE
    992     { 0xA74A, 0xA74B },  // LATIN CAPITAL LETTER O WITH LONG STROKE OVERLAY
    993     { 0xA74C, 0xA74D },  // LATIN CAPITAL LETTER O WITH LOOP
    994     { 0xA74E, 0xA74F },  // LATIN CAPITAL LETTER OO
    995     { 0xA750, 0xA751 },  // LATIN CAPITAL LETTER P WITH STROKE THROUGH DESCENDER
    996     { 0xA752, 0xA753 },  // LATIN CAPITAL LETTER P WITH FLOURISH
    997     { 0xA754, 0xA755 },  // LATIN CAPITAL LETTER P WITH SQUIRREL TAIL
    998     { 0xA756, 0xA757 },  // LATIN CAPITAL LETTER Q WITH STROKE THROUGH DESCENDER
    999     { 0xA758, 0xA759 },  // LATIN CAPITAL LETTER Q WITH DIAGONAL STROKE
   1000     { 0xA75A, 0xA75B },  // LATIN CAPITAL LETTER R ROTUNDA
   1001     { 0xA75C, 0xA75D },  // LATIN CAPITAL LETTER RUM ROTUNDA
   1002     { 0xA75E, 0xA75F },  // LATIN CAPITAL LETTER V WITH DIAGONAL STROKE
   1003     { 0xA760, 0xA761 },  // LATIN CAPITAL LETTER VY
   1004     { 0xA762, 0xA763 },  // LATIN CAPITAL LETTER VISIGOTHIC Z
   1005     { 0xA764, 0xA765 },  // LATIN CAPITAL LETTER THORN WITH STROKE
   1006     { 0xA766, 0xA767 },  // LATIN CAPITAL LETTER THORN WITH STROKE THROUGH DESCENDER
   1007     { 0xA768, 0xA769 },  // LATIN CAPITAL LETTER VEND
   1008     { 0xA76A, 0xA76B },  // LATIN CAPITAL LETTER ET
   1009     { 0xA76C, 0xA76D },  // LATIN CAPITAL LETTER IS
   1010     { 0xA76E, 0xA76F },  // LATIN CAPITAL LETTER CON
   1011     { 0xA779, 0xA77A },  // LATIN CAPITAL LETTER INSULAR D
   1012     { 0xA77B, 0xA77C },  // LATIN CAPITAL LETTER INSULAR F
   1013     { 0xA77D, 0x1D79 },  // LATIN CAPITAL LETTER INSULAR G
   1014     { 0xA77E, 0xA77F },  // LATIN CAPITAL LETTER TURNED INSULAR G
   1015     { 0xA780, 0xA781 },  // LATIN CAPITAL LETTER TURNED L
   1016     { 0xA782, 0xA783 },  // LATIN CAPITAL LETTER INSULAR R
   1017     { 0xA784, 0xA785 },  // LATIN CAPITAL LETTER INSULAR S
   1018     { 0xA786, 0xA787 },  // LATIN CAPITAL LETTER INSULAR T
   1019     { 0xA78B, 0xA78C },  // LATIN CAPITAL LETTER SALTILLO
   1020     { 0xA78D, 0x0265 },  // LATIN CAPITAL LETTER TURNED H
   1021     { 0xA790, 0xA791 },  // LATIN CAPITAL LETTER N WITH DESCENDER
   1022     { 0xA7A0, 0xA7A1 },  // LATIN CAPITAL LETTER G WITH OBLIQUE STROKE
   1023     { 0xA7A2, 0xA7A3 },  // LATIN CAPITAL LETTER K WITH OBLIQUE STROKE
   1024     { 0xA7A4, 0xA7A5 },  // LATIN CAPITAL LETTER N WITH OBLIQUE STROKE
   1025     { 0xA7A6, 0xA7A7 },  // LATIN CAPITAL LETTER R WITH OBLIQUE STROKE
   1026     { 0xA7A8, 0xA7A9 },  // LATIN CAPITAL LETTER S WITH OBLIQUE STROKE
   1027     { 0xFF21, 0xFF41 },  // FULLWIDTH LATIN CAPITAL LETTER A
   1028     { 0xFF22, 0xFF42 },  // FULLWIDTH LATIN CAPITAL LETTER B
   1029     { 0xFF23, 0xFF43 },  // FULLWIDTH LATIN CAPITAL LETTER C
   1030     { 0xFF24, 0xFF44 },  // FULLWIDTH LATIN CAPITAL LETTER D
   1031     { 0xFF25, 0xFF45 },  // FULLWIDTH LATIN CAPITAL LETTER E
   1032     { 0xFF26, 0xFF46 },  // FULLWIDTH LATIN CAPITAL LETTER F
   1033     { 0xFF27, 0xFF47 },  // FULLWIDTH LATIN CAPITAL LETTER G
   1034     { 0xFF28, 0xFF48 },  // FULLWIDTH LATIN CAPITAL LETTER H
   1035     { 0xFF29, 0xFF49 },  // FULLWIDTH LATIN CAPITAL LETTER I
   1036     { 0xFF2A, 0xFF4A },  // FULLWIDTH LATIN CAPITAL LETTER J
   1037     { 0xFF2B, 0xFF4B },  // FULLWIDTH LATIN CAPITAL LETTER K
   1038     { 0xFF2C, 0xFF4C },  // FULLWIDTH LATIN CAPITAL LETTER L
   1039     { 0xFF2D, 0xFF4D },  // FULLWIDTH LATIN CAPITAL LETTER M
   1040     { 0xFF2E, 0xFF4E },  // FULLWIDTH LATIN CAPITAL LETTER N
   1041     { 0xFF2F, 0xFF4F },  // FULLWIDTH LATIN CAPITAL LETTER O
   1042     { 0xFF30, 0xFF50 },  // FULLWIDTH LATIN CAPITAL LETTER P
   1043     { 0xFF31, 0xFF51 },  // FULLWIDTH LATIN CAPITAL LETTER Q
   1044     { 0xFF32, 0xFF52 },  // FULLWIDTH LATIN CAPITAL LETTER R
   1045     { 0xFF33, 0xFF53 },  // FULLWIDTH LATIN CAPITAL LETTER S
   1046     { 0xFF34, 0xFF54 },  // FULLWIDTH LATIN CAPITAL LETTER T
   1047     { 0xFF35, 0xFF55 },  // FULLWIDTH LATIN CAPITAL LETTER U
   1048     { 0xFF36, 0xFF56 },  // FULLWIDTH LATIN CAPITAL LETTER V
   1049     { 0xFF37, 0xFF57 },  // FULLWIDTH LATIN CAPITAL LETTER W
   1050     { 0xFF38, 0xFF58 },  // FULLWIDTH LATIN CAPITAL LETTER X
   1051     { 0xFF39, 0xFF59 },  // FULLWIDTH LATIN CAPITAL LETTER Y
   1052     { 0xFF3A, 0xFF5A }   // FULLWIDTH LATIN CAPITAL LETTER Z
   1053 };
   1054 
   1055 static int compare_pair_capital(const void *a, const void *b) {
   1056     return static_cast<int>(*static_cast<const unsigned short *>(a))
   1057             - static_cast<int>((static_cast<const struct LatinCapitalSmallPair *>(b))->capital);
   1058 }
   1059 
   1060 /* static */ int CharUtils::latin_tolower(const int c) {
   1061     struct LatinCapitalSmallPair *p =
   1062             static_cast<struct LatinCapitalSmallPair *>(bsearch(&c, SORTED_CHAR_MAP,
   1063                     NELEMS(SORTED_CHAR_MAP), sizeof(SORTED_CHAR_MAP[0]), compare_pair_capital));
   1064     return p ? static_cast<int>(p->small) : c;
   1065 }
   1066 
   1067 /*
   1068  * Table mapping most combined Latin, Greek, and Cyrillic characters
   1069  * to their base characters.  If c is in range, CharUtils::BASE_CHARS[c] == c
   1070  * if c is not a combined character, or the base character if it
   1071  * is combined.
   1072  *
   1073  * Generated with:
   1074  *   cat UnicodeData.txt | perl -e 'while (<>) { @foo = split(/;/); $foo[5] =~ s/<.*> //; \
   1075  *   $base[hex($foo[0])] = hex($foo[5]);} \
   1076  *   for ($i = 0; $i < 0x500; $i += 8) { printf("/" . "* U+%04X *" . "/ ", $i); \
   1077  *   for ($j = $i; $j < $i + 8; $j++) { \
   1078  *   printf("0x%04X, ", $base[$j] ? $base[$j] : $j)}; print "\n"; }'
   1079  */
   1080 /* static */ const unsigned short CharUtils::BASE_CHARS[CharUtils::BASE_CHARS_SIZE] = {
   1081     /* U+0000 */ 0x0000, 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006, 0x0007,
   1082     /* U+0008 */ 0x0008, 0x0009, 0x000A, 0x000B, 0x000C, 0x000D, 0x000E, 0x000F,
   1083     /* U+0010 */ 0x0010, 0x0011, 0x0012, 0x0013, 0x0014, 0x0015, 0x0016, 0x0017,
   1084     /* U+0018 */ 0x0018, 0x0019, 0x001A, 0x001B, 0x001C, 0x001D, 0x001E, 0x001F,
   1085     /* U+0020 */ 0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027,
   1086     /* U+0028 */ 0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F,
   1087     /* U+0030 */ 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037,
   1088     /* U+0038 */ 0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F,
   1089     /* U+0040 */ 0x0040, 0x0041, 0x0042, 0x0043, 0x0044, 0x0045, 0x0046, 0x0047,
   1090     /* U+0048 */ 0x0048, 0x0049, 0x004A, 0x004B, 0x004C, 0x004D, 0x004E, 0x004F,
   1091     /* U+0050 */ 0x0050, 0x0051, 0x0052, 0x0053, 0x0054, 0x0055, 0x0056, 0x0057,
   1092     /* U+0058 */ 0x0058, 0x0059, 0x005A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F,
   1093     /* U+0060 */ 0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
   1094     /* U+0068 */ 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
   1095     /* U+0070 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
   1096     /* U+0078 */ 0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007F,
   1097     /* U+0080 */ 0x0080, 0x0081, 0x0082, 0x0083, 0x0084, 0x0085, 0x0086, 0x0087,
   1098     /* U+0088 */ 0x0088, 0x0089, 0x008A, 0x008B, 0x008C, 0x008D, 0x008E, 0x008F,
   1099     /* U+0090 */ 0x0090, 0x0091, 0x0092, 0x0093, 0x0094, 0x0095, 0x0096, 0x0097,
   1100     /* U+0098 */ 0x0098, 0x0099, 0x009A, 0x009B, 0x009C, 0x009D, 0x009E, 0x009F,
   1101     /* U+00A0 */ 0x0020, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
   1102     /* U+00A8 */ 0x0020, 0x00A9, 0x0061, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x0020,
   1103     /* U+00B0 */ 0x00B0, 0x00B1, 0x0032, 0x0033, 0x0020, 0x03BC, 0x00B6, 0x00B7,
   1104     /* U+00B8 */ 0x0020, 0x0031, 0x006F, 0x00BB, 0x0031, 0x0031, 0x0033, 0x00BF,
   1105     /* U+00C0 */ 0x0041, 0x0041, 0x0041, 0x0041, 0x0041, 0x0041, 0x00C6, 0x0043,
   1106     /* U+00C8 */ 0x0045, 0x0045, 0x0045, 0x0045, 0x0049, 0x0049, 0x0049, 0x0049,
   1107     /* U+00D0 */ 0x00D0, 0x004E, 0x004F, 0x004F, 0x004F, 0x004F, 0x004F, 0x00D7,
   1108     /* U+00D8 */ 0x004F, 0x0055, 0x0055, 0x0055, 0x0055, 0x0059, 0x00DE, 0x0073,
   1109         // U+00D8: Manually changed from 00D8 to 004F
   1110           // TODO: Check if it's really acceptable to consider  a diacritical variant of O
   1111         // U+00DF: Manually changed from 00DF to 0073
   1112     /* U+00E0 */ 0x0061, 0x0061, 0x0061, 0x0061, 0x0061, 0x0061, 0x00E6, 0x0063,
   1113     /* U+00E8 */ 0x0065, 0x0065, 0x0065, 0x0065, 0x0069, 0x0069, 0x0069, 0x0069,
   1114     /* U+00F0 */ 0x00F0, 0x006E, 0x006F, 0x006F, 0x006F, 0x006F, 0x006F, 0x00F7,
   1115     /* U+00F8 */ 0x006F, 0x0075, 0x0075, 0x0075, 0x0075, 0x0079, 0x00FE, 0x0079,
   1116         // U+00F8: Manually changed from 00F8 to 006F
   1117           // TODO: Check if it's really acceptable to consider  a diacritical variant of o
   1118     /* U+0100 */ 0x0041, 0x0061, 0x0041, 0x0061, 0x0041, 0x0061, 0x0043, 0x0063,
   1119     /* U+0108 */ 0x0043, 0x0063, 0x0043, 0x0063, 0x0043, 0x0063, 0x0044, 0x0064,
   1120     /* U+0110 */ 0x0046, 0x0064, 0x0045, 0x0065, 0x0045, 0x0065, 0x0045, 0x0065,
   1121         // U+0110: Manually changed from 0110 to 0046
   1122         // U+0111: Manually changed from 0111 to 0064
   1123     /* U+0118 */ 0x0045, 0x0065, 0x0045, 0x0065, 0x0047, 0x0067, 0x0047, 0x0067,
   1124     /* U+0120 */ 0x0047, 0x0067, 0x0047, 0x0067, 0x0048, 0x0068, 0x0126, 0x0127,
   1125     /* U+0128 */ 0x0049, 0x0069, 0x0049, 0x0069, 0x0049, 0x0069, 0x0049, 0x0069,
   1126         // U+0131: Manually changed from 0131 to 0049
   1127     /* U+0130 */ 0x0049, 0x0049, 0x0049, 0x0069, 0x004A, 0x006A, 0x004B, 0x006B,
   1128     /* U+0138 */ 0x0138, 0x004C, 0x006C, 0x004C, 0x006C, 0x004C, 0x006C, 0x004C,
   1129     /* U+0140 */ 0x006C, 0x004C, 0x006C, 0x004E, 0x006E, 0x004E, 0x006E, 0x004E,
   1130         // U+0141: Manually changed from 0141 to 004C
   1131         // U+0142: Manually changed from 0142 to 006C
   1132     /* U+0148 */ 0x006E, 0x02BC, 0x014A, 0x014B, 0x004F, 0x006F, 0x004F, 0x006F,
   1133     /* U+0150 */ 0x004F, 0x006F, 0x0152, 0x0153, 0x0052, 0x0072, 0x0052, 0x0072,
   1134     /* U+0158 */ 0x0052, 0x0072, 0x0053, 0x0073, 0x0053, 0x0073, 0x0053, 0x0073,
   1135     /* U+0160 */ 0x0053, 0x0073, 0x0054, 0x0074, 0x0054, 0x0074, 0x0166, 0x0167,
   1136     /* U+0168 */ 0x0055, 0x0075, 0x0055, 0x0075, 0x0055, 0x0075, 0x0055, 0x0075,
   1137     /* U+0170 */ 0x0055, 0x0075, 0x0055, 0x0075, 0x0057, 0x0077, 0x0059, 0x0079,
   1138     /* U+0178 */ 0x0059, 0x005A, 0x007A, 0x005A, 0x007A, 0x005A, 0x007A, 0x0073,
   1139     /* U+0180 */ 0x0180, 0x0181, 0x0182, 0x0183, 0x0184, 0x0185, 0x0186, 0x0187,
   1140           // TODO: A lot of letters are their own base code points, but for
   1141           // some (e.g. U+0180) it doesn't seem right. Ideally each code point should
   1142           // be checked individually with all languages it's used in.
   1143     /* U+0188 */ 0x0188, 0x0189, 0x018A, 0x018B, 0x018C, 0x018D, 0x018E, 0x018F,
   1144     /* U+0190 */ 0x0190, 0x0191, 0x0192, 0x0193, 0x0194, 0x0195, 0x0196, 0x0197,
   1145     /* U+0198 */ 0x0198, 0x0199, 0x019A, 0x019B, 0x019C, 0x019D, 0x019E, 0x019F,
   1146     /* U+01A0 */ 0x004F, 0x006F, 0x01A2, 0x01A3, 0x01A4, 0x01A5, 0x01A6, 0x01A7,
   1147     /* U+01A8 */ 0x01A8, 0x01A9, 0x01AA, 0x01AB, 0x01AC, 0x01AD, 0x01AE, 0x0055,
   1148     /* U+01B0 */ 0x0075, 0x01B1, 0x01B2, 0x01B3, 0x01B4, 0x01B5, 0x01B6, 0x01B7,
   1149     /* U+01B8 */ 0x01B8, 0x01B9, 0x01BA, 0x01BB, 0x01BC, 0x01BD, 0x01BE, 0x01BF,
   1150     /* U+01C0 */ 0x01C0, 0x01C1, 0x01C2, 0x01C3, 0x0044, 0x0044, 0x0064, 0x004C,
   1151     /* U+01C8 */ 0x004C, 0x006C, 0x004E, 0x004E, 0x006E, 0x0041, 0x0061, 0x0049,
   1152     /* U+01D0 */ 0x0069, 0x004F, 0x006F, 0x0055, 0x0075, 0x0055, 0x0075, 0x0055,
   1153         // U+01D5: Manually changed from 00DC to 0055
   1154         // U+01D6: Manually changed from 00FC to 0075
   1155         // U+01D7: Manually changed from 00DC to 0055
   1156     /* U+01D8 */ 0x0075, 0x0055, 0x0075, 0x0055, 0x0075, 0x01DD, 0x0041, 0x0061,
   1157         // U+01D8: Manually changed from 00FC to 0075
   1158         // U+01D9: Manually changed from 00DC to 0055
   1159         // U+01DA: Manually changed from 00FC to 0075
   1160         // U+01DB: Manually changed from 00DC to 0055
   1161         // U+01DC: Manually changed from 00FC to 0075
   1162         // U+01DE: Manually changed from 00C4 to 0041
   1163         // U+01DF: Manually changed from 00E4 to 0061
   1164     /* U+01E0 */ 0x0041, 0x0061, 0x00C6, 0x00E6, 0x01E4, 0x01E5, 0x0047, 0x0067,
   1165         // U+01E0: Manually changed from 0226 to 0041
   1166         // U+01E1: Manually changed from 0227 to 0061
   1167     /* U+01E8 */ 0x004B, 0x006B, 0x004F, 0x006F, 0x004F, 0x006F, 0x01B7, 0x0292,
   1168         // U+01EC: Manually changed from 01EA to 004F
   1169         // U+01ED: Manually changed from 01EB to 006F
   1170     /* U+01F0 */ 0x006A, 0x0044, 0x0044, 0x0064, 0x0047, 0x0067, 0x01F6, 0x01F7,
   1171     /* U+01F8 */ 0x004E, 0x006E, 0x0041, 0x0061, 0x00C6, 0x00E6, 0x004F, 0x006F,
   1172         // U+01FA: Manually changed from 00C5 to 0041
   1173         // U+01FB: Manually changed from 00E5 to 0061
   1174         // U+01FE: Manually changed from 00D8 to 004F
   1175           // TODO: Check if it's really acceptable to consider  a diacritical variant of O
   1176         // U+01FF: Manually changed from 00F8 to 006F
   1177           // TODO: Check if it's really acceptable to consider  a diacritical variant of o
   1178     /* U+0200 */ 0x0041, 0x0061, 0x0041, 0x0061, 0x0045, 0x0065, 0x0045, 0x0065,
   1179     /* U+0208 */ 0x0049, 0x0069, 0x0049, 0x0069, 0x004F, 0x006F, 0x004F, 0x006F,
   1180     /* U+0210 */ 0x0052, 0x0072, 0x0052, 0x0072, 0x0055, 0x0075, 0x0055, 0x0075,
   1181     /* U+0218 */ 0x0053, 0x0073, 0x0054, 0x0074, 0x021C, 0x021D, 0x0048, 0x0068,
   1182     /* U+0220 */ 0x0220, 0x0221, 0x0222, 0x0223, 0x0224, 0x0225, 0x0041, 0x0061,
   1183     /* U+0228 */ 0x0045, 0x0065, 0x004F, 0x006F, 0x004F, 0x006F, 0x004F, 0x006F,
   1184         // U+022A: Manually changed from 00D6 to 004F
   1185         // U+022B: Manually changed from 00F6 to 006F
   1186         // U+022C: Manually changed from 00D5 to 004F
   1187         // U+022D: Manually changed from 00F5 to 006F
   1188     /* U+0230 */ 0x004F, 0x006F, 0x0059, 0x0079, 0x0234, 0x0235, 0x0236, 0x0237,
   1189         // U+0230: Manually changed from 022E to 004F
   1190         // U+0231: Manually changed from 022F to 006F
   1191     /* U+0238 */ 0x0238, 0x0239, 0x023A, 0x023B, 0x023C, 0x023D, 0x023E, 0x023F,
   1192     /* U+0240 */ 0x0240, 0x0241, 0x0242, 0x0243, 0x0244, 0x0245, 0x0246, 0x0247,
   1193     /* U+0248 */ 0x0248, 0x0249, 0x024A, 0x024B, 0x024C, 0x024D, 0x024E, 0x024F,
   1194     /* U+0250 */ 0x0250, 0x0251, 0x0252, 0x0253, 0x0254, 0x0255, 0x0256, 0x0257,
   1195     /* U+0258 */ 0x0258, 0x0259, 0x025A, 0x025B, 0x025C, 0x025D, 0x025E, 0x025F,
   1196     /* U+0260 */ 0x0260, 0x0261, 0x0262, 0x0263, 0x0264, 0x0265, 0x0266, 0x0267,
   1197     /* U+0268 */ 0x0268, 0x0269, 0x026A, 0x026B, 0x026C, 0x026D, 0x026E, 0x026F,
   1198     /* U+0270 */ 0x0270, 0x0271, 0x0272, 0x0273, 0x0274, 0x0275, 0x0276, 0x0277,
   1199     /* U+0278 */ 0x0278, 0x0279, 0x027A, 0x027B, 0x027C, 0x027D, 0x027E, 0x027F,
   1200     /* U+0280 */ 0x0280, 0x0281, 0x0282, 0x0283, 0x0284, 0x0285, 0x0286, 0x0287,
   1201     /* U+0288 */ 0x0288, 0x0289, 0x028A, 0x028B, 0x028C, 0x028D, 0x028E, 0x028F,
   1202     /* U+0290 */ 0x0290, 0x0291, 0x0292, 0x0293, 0x0294, 0x0295, 0x0296, 0x0297,
   1203     /* U+0298 */ 0x0298, 0x0299, 0x029A, 0x029B, 0x029C, 0x029D, 0x029E, 0x029F,
   1204     /* U+02A0 */ 0x02A0, 0x02A1, 0x02A2, 0x02A3, 0x02A4, 0x02A5, 0x02A6, 0x02A7,
   1205     /* U+02A8 */ 0x02A8, 0x02A9, 0x02AA, 0x02AB, 0x02AC, 0x02AD, 0x02AE, 0x02AF,
   1206     /* U+02B0 */ 0x0068, 0x0266, 0x006A, 0x0072, 0x0279, 0x027B, 0x0281, 0x0077,
   1207     /* U+02B8 */ 0x0079, 0x02B9, 0x02BA, 0x02BB, 0x02BC, 0x02BD, 0x02BE, 0x02BF,
   1208     /* U+02C0 */ 0x02C0, 0x02C1, 0x02C2, 0x02C3, 0x02C4, 0x02C5, 0x02C6, 0x02C7,
   1209     /* U+02C8 */ 0x02C8, 0x02C9, 0x02CA, 0x02CB, 0x02CC, 0x02CD, 0x02CE, 0x02CF,
   1210     /* U+02D0 */ 0x02D0, 0x02D1, 0x02D2, 0x02D3, 0x02D4, 0x02D5, 0x02D6, 0x02D7,
   1211     /* U+02D8 */ 0x0020, 0x0020, 0x0020, 0x0020, 0x0020, 0x0020, 0x02DE, 0x02DF,
   1212     /* U+02E0 */ 0x0263, 0x006C, 0x0073, 0x0078, 0x0295, 0x02E5, 0x02E6, 0x02E7,
   1213     /* U+02E8 */ 0x02E8, 0x02E9, 0x02EA, 0x02EB, 0x02EC, 0x02ED, 0x02EE, 0x02EF,
   1214     /* U+02F0 */ 0x02F0, 0x02F1, 0x02F2, 0x02F3, 0x02F4, 0x02F5, 0x02F6, 0x02F7,
   1215     /* U+02F8 */ 0x02F8, 0x02F9, 0x02FA, 0x02FB, 0x02FC, 0x02FD, 0x02FE, 0x02FF,
   1216     /* U+0300 */ 0x0300, 0x0301, 0x0302, 0x0303, 0x0304, 0x0305, 0x0306, 0x0307,
   1217     /* U+0308 */ 0x0308, 0x0309, 0x030A, 0x030B, 0x030C, 0x030D, 0x030E, 0x030F,
   1218     /* U+0310 */ 0x0310, 0x0311, 0x0312, 0x0313, 0x0314, 0x0315, 0x0316, 0x0317,
   1219     /* U+0318 */ 0x0318, 0x0319, 0x031A, 0x031B, 0x031C, 0x031D, 0x031E, 0x031F,
   1220     /* U+0320 */ 0x0320, 0x0321, 0x0322, 0x0323, 0x0324, 0x0325, 0x0326, 0x0327,
   1221     /* U+0328 */ 0x0328, 0x0329, 0x032A, 0x032B, 0x032C, 0x032D, 0x032E, 0x032F,
   1222     /* U+0330 */ 0x0330, 0x0331, 0x0332, 0x0333, 0x0334, 0x0335, 0x0336, 0x0337,
   1223     /* U+0338 */ 0x0338, 0x0339, 0x033A, 0x033B, 0x033C, 0x033D, 0x033E, 0x033F,
   1224     /* U+0340 */ 0x0300, 0x0301, 0x0342, 0x0313, 0x0308, 0x0345, 0x0346, 0x0347,
   1225     /* U+0348 */ 0x0348, 0x0349, 0x034A, 0x034B, 0x034C, 0x034D, 0x034E, 0x034F,
   1226     /* U+0350 */ 0x0350, 0x0351, 0x0352, 0x0353, 0x0354, 0x0355, 0x0356, 0x0357,
   1227     /* U+0358 */ 0x0358, 0x0359, 0x035A, 0x035B, 0x035C, 0x035D, 0x035E, 0x035F,
   1228     /* U+0360 */ 0x0360, 0x0361, 0x0362, 0x0363, 0x0364, 0x0365, 0x0366, 0x0367,
   1229     /* U+0368 */ 0x0368, 0x0369, 0x036A, 0x036B, 0x036C, 0x036D, 0x036E, 0x036F,
   1230     /* U+0370 */ 0x0370, 0x0371, 0x0372, 0x0373, 0x02B9, 0x0375, 0x0376, 0x0377,
   1231     /* U+0378 */ 0x0378, 0x0379, 0x0020, 0x037B, 0x037C, 0x037D, 0x003B, 0x037F,
   1232     /* U+0380 */ 0x0380, 0x0381, 0x0382, 0x0383, 0x0020, 0x00A8, 0x0391, 0x00B7,
   1233     /* U+0388 */ 0x0395, 0x0397, 0x0399, 0x038B, 0x039F, 0x038D, 0x03A5, 0x03A9,
   1234     /* U+0390 */ 0x03CA, 0x0391, 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397,
   1235     /* U+0398 */ 0x0398, 0x0399, 0x039A, 0x039B, 0x039C, 0x039D, 0x039E, 0x039F,
   1236     /* U+03A0 */ 0x03A0, 0x03A1, 0x03A2, 0x03A3, 0x03A4, 0x03A5, 0x03A6, 0x03A7,
   1237     /* U+03A8 */ 0x03A8, 0x03A9, 0x0399, 0x03A5, 0x03B1, 0x03B5, 0x03B7, 0x03B9,
   1238     /* U+03B0 */ 0x03CB, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
   1239     /* U+03B8 */ 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
   1240     /* U+03C0 */ 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
   1241     /* U+03C8 */ 0x03C8, 0x03C9, 0x03B9, 0x03C5, 0x03BF, 0x03C5, 0x03C9, 0x03CF,
   1242     /* U+03D0 */ 0x03B2, 0x03B8, 0x03A5, 0x03D2, 0x03D2, 0x03C6, 0x03C0, 0x03D7,
   1243     /* U+03D8 */ 0x03D8, 0x03D9, 0x03DA, 0x03DB, 0x03DC, 0x03DD, 0x03DE, 0x03DF,
   1244     /* U+03E0 */ 0x03E0, 0x03E1, 0x03E2, 0x03E3, 0x03E4, 0x03E5, 0x03E6, 0x03E7,
   1245     /* U+03E8 */ 0x03E8, 0x03E9, 0x03EA, 0x03EB, 0x03EC, 0x03ED, 0x03EE, 0x03EF,
   1246     /* U+03F0 */ 0x03BA, 0x03C1, 0x03C2, 0x03F3, 0x0398, 0x03B5, 0x03F6, 0x03F7,
   1247     /* U+03F8 */ 0x03F8, 0x03A3, 0x03FA, 0x03FB, 0x03FC, 0x03FD, 0x03FE, 0x03FF,
   1248     /* U+0400 */ 0x0415, 0x0415, 0x0402, 0x0413, 0x0404, 0x0405, 0x0406, 0x0406,
   1249     /* U+0408 */ 0x0408, 0x0409, 0x040A, 0x040B, 0x041A, 0x0418, 0x0423, 0x040F,
   1250     /* U+0410 */ 0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417,
   1251     /* U+0418 */ 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, 0x041F,
   1252         // U+0419: Manually changed from 0418 to 0419
   1253     /* U+0420 */ 0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427,
   1254     /* U+0428 */ 0x0428, 0x0429, 0x042C, 0x042B, 0x042C, 0x042D, 0x042E, 0x042F,
   1255         // U+042A: Manually changed from 042A to 042C
   1256     /* U+0430 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
   1257     /* U+0438 */ 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
   1258         // U+0439: Manually changed from 0438 to 0439
   1259     /* U+0440 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
   1260     /* U+0448 */ 0x0448, 0x0449, 0x044C, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
   1261         // U+044A: Manually changed from 044A to 044C
   1262     /* U+0450 */ 0x0435, 0x0435, 0x0452, 0x0433, 0x0454, 0x0455, 0x0456, 0x0456,
   1263     /* U+0458 */ 0x0458, 0x0459, 0x045A, 0x045B, 0x043A, 0x0438, 0x0443, 0x045F,
   1264     /* U+0460 */ 0x0460, 0x0461, 0x0462, 0x0463, 0x0464, 0x0465, 0x0466, 0x0467,
   1265     /* U+0468 */ 0x0468, 0x0469, 0x046A, 0x046B, 0x046C, 0x046D, 0x046E, 0x046F,
   1266     /* U+0470 */ 0x0470, 0x0471, 0x0472, 0x0473, 0x0474, 0x0475, 0x0474, 0x0475,
   1267     /* U+0478 */ 0x0478, 0x0479, 0x047A, 0x047B, 0x047C, 0x047D, 0x047E, 0x047F,
   1268     /* U+0480 */ 0x0480, 0x0481, 0x0482, 0x0483, 0x0484, 0x0485, 0x0486, 0x0487,
   1269     /* U+0488 */ 0x0488, 0x0489, 0x048A, 0x048B, 0x048C, 0x048D, 0x048E, 0x048F,
   1270     /* U+0490 */ 0x0490, 0x0491, 0x0492, 0x0493, 0x0494, 0x0495, 0x0496, 0x0497,
   1271     /* U+0498 */ 0x0498, 0x0499, 0x049A, 0x049B, 0x049C, 0x049D, 0x049E, 0x049F,
   1272     /* U+04A0 */ 0x04A0, 0x04A1, 0x04A2, 0x04A3, 0x04A4, 0x04A5, 0x04A6, 0x04A7,
   1273     /* U+04A8 */ 0x04A8, 0x04A9, 0x04AA, 0x04AB, 0x04AC, 0x04AD, 0x04AE, 0x04AF,
   1274     /* U+04B0 */ 0x04B0, 0x04B1, 0x04B2, 0x04B3, 0x04B4, 0x04B5, 0x04B6, 0x04B7,
   1275     /* U+04B8 */ 0x04B8, 0x04B9, 0x04BA, 0x04BB, 0x04BC, 0x04BD, 0x04BE, 0x04BF,
   1276     /* U+04C0 */ 0x04C0, 0x0416, 0x0436, 0x04C3, 0x04C4, 0x04C5, 0x04C6, 0x04C7,
   1277     /* U+04C8 */ 0x04C8, 0x04C9, 0x04CA, 0x04CB, 0x04CC, 0x04CD, 0x04CE, 0x04CF,
   1278     /* U+04D0 */ 0x0410, 0x0430, 0x0410, 0x0430, 0x04D4, 0x04D5, 0x0415, 0x0435,
   1279     /* U+04D8 */ 0x04D8, 0x04D9, 0x04D8, 0x04D9, 0x0416, 0x0436, 0x0417, 0x0437,
   1280     /* U+04E0 */ 0x04E0, 0x04E1, 0x0418, 0x0438, 0x0418, 0x0438, 0x041E, 0x043E,
   1281     /* U+04E8 */ 0x04E8, 0x04E9, 0x04E8, 0x04E9, 0x042D, 0x044D, 0x0423, 0x0443,
   1282     /* U+04F0 */ 0x0423, 0x0443, 0x0423, 0x0443, 0x0427, 0x0447, 0x04F6, 0x04F7,
   1283     /* U+04F8 */ 0x042B, 0x044B, 0x04FA, 0x04FB, 0x04FC, 0x04FD, 0x04FE, 0x04FF,
   1284 };
   1285 
   1286 /* static */ const std::vector<int> CharUtils::EMPTY_STRING(1 /* size */, '\0' /* value */);
   1287 } // namespace latinime
   1288