nyancrimew
/
LatinIME
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge "Update SORTED_CHAR_MAP"

main
Ken Wakasa 2013-04-16 10:19:41 +00:00 committed by Android (Google) Code Review
parent ff766f8c4b 2dfaf7a4ac
commit e01e22e1ea
2 changed files with 183 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

190
native/jni/src/char_utils.cpp
View File

@ -45,18 +45,16 @@ struct LatinCapitalSmallPair {
extern "C" int main() { extern "C" int main() {
for (unsigned short c = 0; c < 0xFFFF; c++) { for (unsigned short c = 0; c < 0xFFFF; c++) {
const unsigned short baseC = c < NELEMS(BASE_CHARS) ? BASE_CHARS[c] : c; if (c <= 0x7F) continue;
if (baseC <= 0x7F) continue; const unsigned short icu4cLowerC = u_tolower(c);
const unsigned short icu4cLowerBaseC = u_tolower(baseC); const unsigned short myLowerC = latin_tolower(c);
const unsigned short myLowerBaseC = latin_tolower(baseC); if (c != icu4cLowerC) {
if (baseC != icu4cLowerBaseC) {
#ifdef CONFIRMING_CHAR_UTILS #ifdef CONFIRMING_CHAR_UTILS
if (icu4cLowerBaseC != myLowerBaseC) { if (icu4cLowerC != myLowerC) {
fprintf(stderr, "icu4cLowerBaseC != myLowerBaseC, 0x%04X, 0x%04X\n", fprintf(stderr, "icu4cLowerC != myLowerC, 0x%04X, 0x%04X\n", icu4cLowerC, myLowerC);
icu4cLowerBaseC, myLowerBaseC);
} }
#else // CONFIRMING_CHAR_UTILS #else // CONFIRMING_CHAR_UTILS
printf("0x%04X, 0x%04X\n", baseC, icu4cLowerBaseC); printf("0x%04X, 0x%04X\n", c, icu4cLowerC);
#endif // CONFIRMING_CHAR_UTILS #endif // CONFIRMING_CHAR_UTILS
} }
} }
@ -77,14 +75,99 @@ extern "C" int main() {
* $ * $
*/ */
static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = { static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x00C0, 0x00E0 }, // LATIN CAPITAL LETTER A WITH GRAVE
{ 0x00C1, 0x00E1 }, // LATIN CAPITAL LETTER A WITH ACUTE
{ 0x00C2, 0x00E2 }, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX
{ 0x00C3, 0x00E3 }, // LATIN CAPITAL LETTER A WITH TILDE
{ 0x00C4, 0x00E4 }, // LATIN CAPITAL LETTER A WITH DIAERESIS
{ 0x00C5, 0x00E5 }, // LATIN CAPITAL LETTER A WITH RING ABOVE
{ 0x00C6, 0x00E6 }, // LATIN CAPITAL LETTER AE { 0x00C6, 0x00E6 }, // LATIN CAPITAL LETTER AE
{ 0x00C7, 0x00E7 }, // LATIN CAPITAL LETTER C WITH CEDILLA
{ 0x00C8, 0x00E8 }, // LATIN CAPITAL LETTER E WITH GRAVE
{ 0x00C9, 0x00E9 }, // LATIN CAPITAL LETTER E WITH ACUTE
{ 0x00CA, 0x00EA }, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX
{ 0x00CB, 0x00EB }, // LATIN CAPITAL LETTER E WITH DIAERESIS
{ 0x00CC, 0x00EC }, // LATIN CAPITAL LETTER I WITH GRAVE
{ 0x00CD, 0x00ED }, // LATIN CAPITAL LETTER I WITH ACUTE
{ 0x00CE, 0x00EE }, // LATIN CAPITAL LETTER I WITH CIRCUMFLEX
{ 0x00CF, 0x00EF }, // LATIN CAPITAL LETTER I WITH DIAERESIS
{ 0x00D0, 0x00F0 }, // LATIN CAPITAL LETTER ETH { 0x00D0, 0x00F0 }, // LATIN CAPITAL LETTER ETH
{ 0x00D1, 0x00F1 }, // LATIN CAPITAL LETTER N WITH TILDE
{ 0x00D2, 0x00F2 }, // LATIN CAPITAL LETTER O WITH GRAVE
{ 0x00D3, 0x00F3 }, // LATIN CAPITAL LETTER O WITH ACUTE
{ 0x00D4, 0x00F4 }, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX
{ 0x00D5, 0x00F5 }, // LATIN CAPITAL LETTER O WITH TILDE
{ 0x00D6, 0x00F6 }, // LATIN CAPITAL LETTER O WITH DIAERESIS
{ 0x00D8, 0x00F8 }, // LATIN CAPITAL LETTER O WITH STROKE
{ 0x00D9, 0x00F9 }, // LATIN CAPITAL LETTER U WITH GRAVE
{ 0x00DA, 0x00FA }, // LATIN CAPITAL LETTER U WITH ACUTE
{ 0x00DB, 0x00FB }, // LATIN CAPITAL LETTER U WITH CIRCUMFLEX
{ 0x00DC, 0x00FC }, // LATIN CAPITAL LETTER U WITH DIAERESIS
{ 0x00DD, 0x00FD }, // LATIN CAPITAL LETTER Y WITH ACUTE
{ 0x00DE, 0x00FE }, // LATIN CAPITAL LETTER THORN { 0x00DE, 0x00FE }, // LATIN CAPITAL LETTER THORN
{ 0x0100, 0x0101 }, // LATIN CAPITAL LETTER A WITH MACRON
{ 0x0102, 0x0103 }, // LATIN CAPITAL LETTER A WITH BREVE
{ 0x0104, 0x0105 }, // LATIN CAPITAL LETTER A WITH OGONEK
{ 0x0106, 0x0107 }, // LATIN CAPITAL LETTER C WITH ACUTE
{ 0x0108, 0x0109 }, // LATIN CAPITAL LETTER C WITH CIRCUMFLEX
{ 0x010A, 0x010B }, // LATIN CAPITAL LETTER C WITH DOT ABOVE
{ 0x010C, 0x010D }, // LATIN CAPITAL LETTER C WITH CARON
{ 0x010E, 0x010F }, // LATIN CAPITAL LETTER D WITH CARON
{ 0x0110, 0x0111 }, // LATIN CAPITAL LETTER D WITH STROKE { 0x0110, 0x0111 }, // LATIN CAPITAL LETTER D WITH STROKE
{ 0x0112, 0x0113 }, // LATIN CAPITAL LETTER E WITH MACRON
{ 0x0114, 0x0115 }, // LATIN CAPITAL LETTER E WITH BREVE
{ 0x0116, 0x0117 }, // LATIN CAPITAL LETTER E WITH DOT ABOVE
{ 0x0118, 0x0119 }, // LATIN CAPITAL LETTER E WITH OGONEK
{ 0x011A, 0x011B }, // LATIN CAPITAL LETTER E WITH CARON
{ 0x011C, 0x011D }, // LATIN CAPITAL LETTER G WITH CIRCUMFLEX
{ 0x011E, 0x011F }, // LATIN CAPITAL LETTER G WITH BREVE
{ 0x0120, 0x0121 }, // LATIN CAPITAL LETTER G WITH DOT ABOVE
{ 0x0122, 0x0123 }, // LATIN CAPITAL LETTER G WITH CEDILLA
{ 0x0124, 0x0125 }, // LATIN CAPITAL LETTER H WITH CIRCUMFLEX
{ 0x0126, 0x0127 }, // LATIN CAPITAL LETTER H WITH STROKE { 0x0126, 0x0127 }, // LATIN CAPITAL LETTER H WITH STROKE
{ 0x0128, 0x0129 }, // LATIN CAPITAL LETTER I WITH TILDE
{ 0x012A, 0x012B }, // LATIN CAPITAL LETTER I WITH MACRON
{ 0x012C, 0x012D }, // LATIN CAPITAL LETTER I WITH BREVE
{ 0x012E, 0x012F }, // LATIN CAPITAL LETTER I WITH OGONEK
{ 0x0130, 0x0069 }, // LATIN CAPITAL LETTER I WITH DOT ABOVE
{ 0x0132, 0x0133 }, // LATIN CAPITAL LIGATURE IJ
{ 0x0134, 0x0135 }, // LATIN CAPITAL LETTER J WITH CIRCUMFLEX
{ 0x0136, 0x0137 }, // LATIN CAPITAL LETTER K WITH CEDILLA
{ 0x0139, 0x013A }, // LATIN CAPITAL LETTER L WITH ACUTE
{ 0x013B, 0x013C }, // LATIN CAPITAL LETTER L WITH CEDILLA
{ 0x013D, 0x013E }, // LATIN CAPITAL LETTER L WITH CARON
{ 0x013F, 0x0140 }, // LATIN CAPITAL LETTER L WITH MIDDLE DOT
{ 0x0141, 0x0142 }, // LATIN CAPITAL LETTER L WITH STROKE
{ 0x0143, 0x0144 }, // LATIN CAPITAL LETTER N WITH ACUTE
{ 0x0145, 0x0146 }, // LATIN CAPITAL LETTER N WITH CEDILLA
{ 0x0147, 0x0148 }, // LATIN CAPITAL LETTER N WITH CARON
{ 0x014A, 0x014B }, // LATIN CAPITAL LETTER ENG { 0x014A, 0x014B }, // LATIN CAPITAL LETTER ENG
{ 0x014C, 0x014D }, // LATIN CAPITAL LETTER O WITH MACRON
{ 0x014E, 0x014F }, // LATIN CAPITAL LETTER O WITH BREVE
{ 0x0150, 0x0151 }, // LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
{ 0x0152, 0x0153 }, // LATIN CAPITAL LIGATURE OE { 0x0152, 0x0153 }, // LATIN CAPITAL LIGATURE OE
{ 0x0154, 0x0155 }, // LATIN CAPITAL LETTER R WITH ACUTE
{ 0x0156, 0x0157 }, // LATIN CAPITAL LETTER R WITH CEDILLA
{ 0x0158, 0x0159 }, // LATIN CAPITAL LETTER R WITH CARON
{ 0x015A, 0x015B }, // LATIN CAPITAL LETTER S WITH ACUTE
{ 0x015C, 0x015D }, // LATIN CAPITAL LETTER S WITH CIRCUMFLEX
{ 0x015E, 0x015F }, // LATIN CAPITAL LETTER S WITH CEDILLA
{ 0x0160, 0x0161 }, // LATIN CAPITAL LETTER S WITH CARON
{ 0x0162, 0x0163 }, // LATIN CAPITAL LETTER T WITH CEDILLA
{ 0x0164, 0x0165 }, // LATIN CAPITAL LETTER T WITH CARON
{ 0x0166, 0x0167 }, // LATIN CAPITAL LETTER T WITH STROKE { 0x0166, 0x0167 }, // LATIN CAPITAL LETTER T WITH STROKE
{ 0x0168, 0x0169 }, // LATIN CAPITAL LETTER U WITH TILDE
{ 0x016A, 0x016B }, // LATIN CAPITAL LETTER U WITH MACRON
{ 0x016C, 0x016D }, // LATIN CAPITAL LETTER U WITH BREVE
{ 0x016E, 0x016F }, // LATIN CAPITAL LETTER U WITH RING ABOVE
{ 0x0170, 0x0171 }, // LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
{ 0x0172, 0x0173 }, // LATIN CAPITAL LETTER U WITH OGONEK
{ 0x0174, 0x0175 }, // LATIN CAPITAL LETTER W WITH CIRCUMFLEX
{ 0x0176, 0x0177 }, // LATIN CAPITAL LETTER Y WITH CIRCUMFLEX
{ 0x0178, 0x00FF }, // LATIN CAPITAL LETTER Y WITH DIAERESIS
{ 0x0179, 0x017A }, // LATIN CAPITAL LETTER Z WITH ACUTE
{ 0x017B, 0x017C }, // LATIN CAPITAL LETTER Z WITH DOT ABOVE
{ 0x017D, 0x017E }, // LATIN CAPITAL LETTER Z WITH CARON
{ 0x0181, 0x0253 }, // LATIN CAPITAL LETTER B WITH HOOK { 0x0181, 0x0253 }, // LATIN CAPITAL LETTER B WITH HOOK
{ 0x0182, 0x0183 }, // LATIN CAPITAL LETTER B WITH TOPBAR { 0x0182, 0x0183 }, // LATIN CAPITAL LETTER B WITH TOPBAR
{ 0x0184, 0x0185 }, // LATIN CAPITAL LETTER TONE SIX { 0x0184, 0x0185 }, // LATIN CAPITAL LETTER TONE SIX
@ -105,6 +188,7 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x019C, 0x026F }, // LATIN CAPITAL LETTER TURNED M { 0x019C, 0x026F }, // LATIN CAPITAL LETTER TURNED M
{ 0x019D, 0x0272 }, // LATIN CAPITAL LETTER N WITH LEFT HOOK { 0x019D, 0x0272 }, // LATIN CAPITAL LETTER N WITH LEFT HOOK
{ 0x019F, 0x0275 }, // LATIN CAPITAL LETTER O WITH MIDDLE TILDE { 0x019F, 0x0275 }, // LATIN CAPITAL LETTER O WITH MIDDLE TILDE
{ 0x01A0, 0x01A1 }, // LATIN CAPITAL LETTER O WITH HORN
{ 0x01A2, 0x01A3 }, // LATIN CAPITAL LETTER OI { 0x01A2, 0x01A3 }, // LATIN CAPITAL LETTER OI
{ 0x01A4, 0x01A5 }, // LATIN CAPITAL LETTER P WITH HOOK { 0x01A4, 0x01A5 }, // LATIN CAPITAL LETTER P WITH HOOK
{ 0x01A6, 0x0280 }, // LATIN LETTER YR { 0x01A6, 0x0280 }, // LATIN LETTER YR
@ -112,6 +196,7 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x01A9, 0x0283 }, // LATIN CAPITAL LETTER ESH { 0x01A9, 0x0283 }, // LATIN CAPITAL LETTER ESH
{ 0x01AC, 0x01AD }, // LATIN CAPITAL LETTER T WITH HOOK { 0x01AC, 0x01AD }, // LATIN CAPITAL LETTER T WITH HOOK
{ 0x01AE, 0x0288 }, // LATIN CAPITAL LETTER T WITH RETROFLEX HOOK { 0x01AE, 0x0288 }, // LATIN CAPITAL LETTER T WITH RETROFLEX HOOK
{ 0x01AF, 0x01B0 }, // LATIN CAPITAL LETTER U WITH HORN
{ 0x01B1, 0x028A }, // LATIN CAPITAL LETTER UPSILON { 0x01B1, 0x028A }, // LATIN CAPITAL LETTER UPSILON
{ 0x01B2, 0x028B }, // LATIN CAPITAL LETTER V WITH HOOK { 0x01B2, 0x028B }, // LATIN CAPITAL LETTER V WITH HOOK
{ 0x01B3, 0x01B4 }, // LATIN CAPITAL LETTER Y WITH HOOK { 0x01B3, 0x01B4 }, // LATIN CAPITAL LETTER Y WITH HOOK
@ -119,13 +204,64 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x01B7, 0x0292 }, // LATIN CAPITAL LETTER EZH { 0x01B7, 0x0292 }, // LATIN CAPITAL LETTER EZH
{ 0x01B8, 0x01B9 }, // LATIN CAPITAL LETTER EZH REVERSED { 0x01B8, 0x01B9 }, // LATIN CAPITAL LETTER EZH REVERSED
{ 0x01BC, 0x01BD }, // LATIN CAPITAL LETTER TONE FIVE { 0x01BC, 0x01BD }, // LATIN CAPITAL LETTER TONE FIVE
{ 0x01C4, 0x01C6 }, // LATIN CAPITAL LETTER DZ WITH CARON
{ 0x01C5, 0x01C6 }, // LATIN CAPITAL LETTER D WITH SMALL LETTER Z WITH CARON
{ 0x01C7, 0x01C9 }, // LATIN CAPITAL LETTER LJ
{ 0x01C8, 0x01C9 }, // LATIN CAPITAL LETTER L WITH SMALL LETTER J
{ 0x01CA, 0x01CC }, // LATIN CAPITAL LETTER NJ
{ 0x01CB, 0x01CC }, // LATIN CAPITAL LETTER N WITH SMALL LETTER J
{ 0x01CD, 0x01CE }, // LATIN CAPITAL LETTER A WITH CARON
{ 0x01CF, 0x01D0 }, // LATIN CAPITAL LETTER I WITH CARON
{ 0x01D1, 0x01D2 }, // LATIN CAPITAL LETTER O WITH CARON
{ 0x01D3, 0x01D4 }, // LATIN CAPITAL LETTER U WITH CARON
{ 0x01D5, 0x01D6 }, // LATIN CAPITAL LETTER U WITH DIAERESIS AND MACRON
{ 0x01D7, 0x01D8 }, // LATIN CAPITAL LETTER U WITH DIAERESIS AND ACUTE
{ 0x01D9, 0x01DA }, // LATIN CAPITAL LETTER U WITH DIAERESIS AND CARON
{ 0x01DB, 0x01DC }, // LATIN CAPITAL LETTER U WITH DIAERESIS AND GRAVE
{ 0x01DE, 0x01DF }, // LATIN CAPITAL LETTER A WITH DIAERESIS AND MACRON
{ 0x01E0, 0x01E1 }, // LATIN CAPITAL LETTER A WITH DOT ABOVE AND MACRON
{ 0x01E2, 0x01E3 }, // LATIN CAPITAL LETTER AE WITH MACRON
{ 0x01E4, 0x01E5 }, // LATIN CAPITAL LETTER G WITH STROKE { 0x01E4, 0x01E5 }, // LATIN CAPITAL LETTER G WITH STROKE
{ 0x01E6, 0x01E7 }, // LATIN CAPITAL LETTER G WITH CARON
{ 0x01E8, 0x01E9 }, // LATIN CAPITAL LETTER K WITH CARON
{ 0x01EA, 0x01EB }, // LATIN CAPITAL LETTER O WITH OGONEK
{ 0x01EC, 0x01ED }, // LATIN CAPITAL LETTER O WITH OGONEK AND MACRON
{ 0x01EE, 0x01EF }, // LATIN CAPITAL LETTER EZH WITH CARON
{ 0x01F1, 0x01F3 }, // LATIN CAPITAL LETTER DZ
{ 0x01F2, 0x01F3 }, // LATIN CAPITAL LETTER D WITH SMALL LETTER Z
{ 0x01F4, 0x01F5 }, // LATIN CAPITAL LETTER G WITH ACUTE
{ 0x01F6, 0x0195 }, // LATIN CAPITAL LETTER HWAIR { 0x01F6, 0x0195 }, // LATIN CAPITAL LETTER HWAIR
{ 0x01F7, 0x01BF }, // LATIN CAPITAL LETTER WYNN { 0x01F7, 0x01BF }, // LATIN CAPITAL LETTER WYNN
{ 0x01F8, 0x01F9 }, // LATIN CAPITAL LETTER N WITH GRAVE
{ 0x01FA, 0x01FB }, // LATIN CAPITAL LETTER A WITH RING ABOVE AND ACUTE
{ 0x01FC, 0x01FD }, // LATIN CAPITAL LETTER AE WITH ACUTE
{ 0x01FE, 0x01FF }, // LATIN CAPITAL LETTER O WITH STROKE AND ACUTE
{ 0x0200, 0x0201 }, // LATIN CAPITAL LETTER A WITH DOUBLE GRAVE
{ 0x0202, 0x0203 }, // LATIN CAPITAL LETTER A WITH INVERTED BREVE
{ 0x0204, 0x0205 }, // LATIN CAPITAL LETTER E WITH DOUBLE GRAVE
{ 0x0206, 0x0207 }, // LATIN CAPITAL LETTER E WITH INVERTED BREVE
{ 0x0208, 0x0209 }, // LATIN CAPITAL LETTER I WITH DOUBLE GRAVE
{ 0x020A, 0x020B }, // LATIN CAPITAL LETTER I WITH INVERTED BREVE
{ 0x020C, 0x020D }, // LATIN CAPITAL LETTER O WITH DOUBLE GRAVE
{ 0x020E, 0x020F }, // LATIN CAPITAL LETTER O WITH INVERTED BREVE
{ 0x0210, 0x0211 }, // LATIN CAPITAL LETTER R WITH DOUBLE GRAVE
{ 0x0212, 0x0213 }, // LATIN CAPITAL LETTER R WITH INVERTED BREVE
{ 0x0214, 0x0215 }, // LATIN CAPITAL LETTER U WITH DOUBLE GRAVE
{ 0x0216, 0x0217 }, // LATIN CAPITAL LETTER U WITH INVERTED BREVE
{ 0x0218, 0x0219 }, // LATIN CAPITAL LETTER S WITH COMMA BELOW
{ 0x021A, 0x021B }, // LATIN CAPITAL LETTER T WITH COMMA BELOW
{ 0x021C, 0x021D }, // LATIN CAPITAL LETTER YOGH { 0x021C, 0x021D }, // LATIN CAPITAL LETTER YOGH
{ 0x021E, 0x021F }, // LATIN CAPITAL LETTER H WITH CARON
{ 0x0220, 0x019E }, // LATIN CAPITAL LETTER N WITH LONG RIGHT LEG { 0x0220, 0x019E }, // LATIN CAPITAL LETTER N WITH LONG RIGHT LEG
{ 0x0222, 0x0223 }, // LATIN CAPITAL LETTER OU { 0x0222, 0x0223 }, // LATIN CAPITAL LETTER OU
{ 0x0224, 0x0225 }, // LATIN CAPITAL LETTER Z WITH HOOK { 0x0224, 0x0225 }, // LATIN CAPITAL LETTER Z WITH HOOK
{ 0x0226, 0x0227 }, // LATIN CAPITAL LETTER A WITH DOT ABOVE
{ 0x0228, 0x0229 }, // LATIN CAPITAL LETTER E WITH CEDILLA
{ 0x022A, 0x022B }, // LATIN CAPITAL LETTER O WITH DIAERESIS AND MACRON
{ 0x022C, 0x022D }, // LATIN CAPITAL LETTER O WITH TILDE AND MACRON
{ 0x022E, 0x022F }, // LATIN CAPITAL LETTER O WITH DOT ABOVE
{ 0x0230, 0x0231 }, // LATIN CAPITAL LETTER O WITH DOT ABOVE AND MACRON
{ 0x0232, 0x0233 }, // LATIN CAPITAL LETTER Y WITH MACRON
{ 0x023A, 0x2C65 }, // LATIN CAPITAL LETTER A WITH STROKE { 0x023A, 0x2C65 }, // LATIN CAPITAL LETTER A WITH STROKE
{ 0x023B, 0x023C }, // LATIN CAPITAL LETTER C WITH STROKE { 0x023B, 0x023C }, // LATIN CAPITAL LETTER C WITH STROKE
{ 0x023D, 0x019A }, // LATIN CAPITAL LETTER L WITH BAR { 0x023D, 0x019A }, // LATIN CAPITAL LETTER L WITH BAR
@ -142,6 +278,13 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x0370, 0x0371 }, // GREEK CAPITAL LETTER HETA { 0x0370, 0x0371 }, // GREEK CAPITAL LETTER HETA
{ 0x0372, 0x0373 }, // GREEK CAPITAL LETTER ARCHAIC SAMPI { 0x0372, 0x0373 }, // GREEK CAPITAL LETTER ARCHAIC SAMPI
{ 0x0376, 0x0377 }, // GREEK CAPITAL LETTER PAMPHYLIAN DIGAMMA { 0x0376, 0x0377 }, // GREEK CAPITAL LETTER PAMPHYLIAN DIGAMMA
{ 0x0386, 0x03AC }, // GREEK CAPITAL LETTER ALPHA WITH TONOS
{ 0x0388, 0x03AD }, // GREEK CAPITAL LETTER EPSILON WITH TONOS
{ 0x0389, 0x03AE }, // GREEK CAPITAL LETTER ETA WITH TONOS
{ 0x038A, 0x03AF }, // GREEK CAPITAL LETTER IOTA WITH TONOS
{ 0x038C, 0x03CC }, // GREEK CAPITAL LETTER OMICRON WITH TONOS
{ 0x038E, 0x03CD }, // GREEK CAPITAL LETTER UPSILON WITH TONOS
{ 0x038F, 0x03CE }, // GREEK CAPITAL LETTER OMEGA WITH TONOS
{ 0x0391, 0x03B1 }, // GREEK CAPITAL LETTER ALPHA { 0x0391, 0x03B1 }, // GREEK CAPITAL LETTER ALPHA
{ 0x0392, 0x03B2 }, // GREEK CAPITAL LETTER BETA { 0x0392, 0x03B2 }, // GREEK CAPITAL LETTER BETA
{ 0x0393, 0x03B3 }, // GREEK CAPITAL LETTER GAMMA { 0x0393, 0x03B3 }, // GREEK CAPITAL LETTER GAMMA
@ -166,6 +309,8 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x03A7, 0x03C7 }, // GREEK CAPITAL LETTER CHI { 0x03A7, 0x03C7 }, // GREEK CAPITAL LETTER CHI
{ 0x03A8, 0x03C8 }, // GREEK CAPITAL LETTER PSI { 0x03A8, 0x03C8 }, // GREEK CAPITAL LETTER PSI
{ 0x03A9, 0x03C9 }, // GREEK CAPITAL LETTER OMEGA { 0x03A9, 0x03C9 }, // GREEK CAPITAL LETTER OMEGA
{ 0x03AA, 0x03CA }, // GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
{ 0x03AB, 0x03CB }, // GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
{ 0x03CF, 0x03D7 }, // GREEK CAPITAL KAI SYMBOL { 0x03CF, 0x03D7 }, // GREEK CAPITAL KAI SYMBOL
{ 0x03D8, 0x03D9 }, // GREEK LETTER ARCHAIC KOPPA { 0x03D8, 0x03D9 }, // GREEK LETTER ARCHAIC KOPPA
{ 0x03DA, 0x03DB }, // GREEK LETTER STIGMA { 0x03DA, 0x03DB }, // GREEK LETTER STIGMA
@ -179,19 +324,28 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x03EA, 0x03EB }, // COPTIC CAPITAL LETTER GANGIA { 0x03EA, 0x03EB }, // COPTIC CAPITAL LETTER GANGIA
{ 0x03EC, 0x03ED }, // COPTIC CAPITAL LETTER SHIMA { 0x03EC, 0x03ED }, // COPTIC CAPITAL LETTER SHIMA
{ 0x03EE, 0x03EF }, // COPTIC CAPITAL LETTER DEI { 0x03EE, 0x03EF }, // COPTIC CAPITAL LETTER DEI
{ 0x03F4, 0x03B8 }, // GREEK CAPITAL THETA SYMBOL
{ 0x03F7, 0x03F8 }, // GREEK CAPITAL LETTER SHO { 0x03F7, 0x03F8 }, // GREEK CAPITAL LETTER SHO
{ 0x03F9, 0x03F2 }, // GREEK CAPITAL LUNATE SIGMA SYMBOL
{ 0x03FA, 0x03FB }, // GREEK CAPITAL LETTER SAN { 0x03FA, 0x03FB }, // GREEK CAPITAL LETTER SAN
{ 0x03FD, 0x037B }, // GREEK CAPITAL REVERSED LUNATE SIGMA SYMBOL { 0x03FD, 0x037B }, // GREEK CAPITAL REVERSED LUNATE SIGMA SYMBOL
{ 0x03FE, 0x037C }, // GREEK CAPITAL DOTTED LUNATE SIGMA SYMBOL { 0x03FE, 0x037C }, // GREEK CAPITAL DOTTED LUNATE SIGMA SYMBOL
{ 0x03FF, 0x037D }, // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL { 0x03FF, 0x037D }, // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL
{ 0x0400, 0x0450 }, // CYRILLIC CAPITAL LETTER IE WITH GRAVE
{ 0x0401, 0x0451 }, // CYRILLIC CAPITAL LETTER IO
{ 0x0402, 0x0452 }, // CYRILLIC CAPITAL LETTER DJE { 0x0402, 0x0452 }, // CYRILLIC CAPITAL LETTER DJE
{ 0x0403, 0x0453 }, // CYRILLIC CAPITAL LETTER GJE
{ 0x0404, 0x0454 }, // CYRILLIC CAPITAL LETTER UKRAINIAN IE { 0x0404, 0x0454 }, // CYRILLIC CAPITAL LETTER UKRAINIAN IE
{ 0x0405, 0x0455 }, // CYRILLIC CAPITAL LETTER DZE { 0x0405, 0x0455 }, // CYRILLIC CAPITAL LETTER DZE
{ 0x0406, 0x0456 }, // CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I { 0x0406, 0x0456 }, // CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
{ 0x0407, 0x0457 }, // CYRILLIC CAPITAL LETTER YI
{ 0x0408, 0x0458 }, // CYRILLIC CAPITAL LETTER JE { 0x0408, 0x0458 }, // CYRILLIC CAPITAL LETTER JE
{ 0x0409, 0x0459 }, // CYRILLIC CAPITAL LETTER LJE { 0x0409, 0x0459 }, // CYRILLIC CAPITAL LETTER LJE
{ 0x040A, 0x045A }, // CYRILLIC CAPITAL LETTER NJE { 0x040A, 0x045A }, // CYRILLIC CAPITAL LETTER NJE
{ 0x040B, 0x045B }, // CYRILLIC CAPITAL LETTER TSHE { 0x040B, 0x045B }, // CYRILLIC CAPITAL LETTER TSHE
{ 0x040C, 0x045C }, // CYRILLIC CAPITAL LETTER KJE
{ 0x040D, 0x045D }, // CYRILLIC CAPITAL LETTER I WITH GRAVE
{ 0x040E, 0x045E }, // CYRILLIC CAPITAL LETTER SHORT U
{ 0x040F, 0x045F }, // CYRILLIC CAPITAL LETTER DZHE { 0x040F, 0x045F }, // CYRILLIC CAPITAL LETTER DZHE
{ 0x0410, 0x0430 }, // CYRILLIC CAPITAL LETTER A { 0x0410, 0x0430 }, // CYRILLIC CAPITAL LETTER A
{ 0x0411, 0x0431 }, // CYRILLIC CAPITAL LETTER BE { 0x0411, 0x0431 }, // CYRILLIC CAPITAL LETTER BE
@ -236,6 +390,7 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x0470, 0x0471 }, // CYRILLIC CAPITAL LETTER PSI { 0x0470, 0x0471 }, // CYRILLIC CAPITAL LETTER PSI
{ 0x0472, 0x0473 }, // CYRILLIC CAPITAL LETTER FITA { 0x0472, 0x0473 }, // CYRILLIC CAPITAL LETTER FITA
{ 0x0474, 0x0475 }, // CYRILLIC CAPITAL LETTER IZHITSA { 0x0474, 0x0475 }, // CYRILLIC CAPITAL LETTER IZHITSA
{ 0x0476, 0x0477 }, // CYRILLIC CAPITAL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT
{ 0x0478, 0x0479 }, // CYRILLIC CAPITAL LETTER UK { 0x0478, 0x0479 }, // CYRILLIC CAPITAL LETTER UK
{ 0x047A, 0x047B }, // CYRILLIC CAPITAL LETTER ROUND OMEGA { 0x047A, 0x047B }, // CYRILLIC CAPITAL LETTER ROUND OMEGA
{ 0x047C, 0x047D }, // CYRILLIC CAPITAL LETTER OMEGA WITH TITLO { 0x047C, 0x047D }, // CYRILLIC CAPITAL LETTER OMEGA WITH TITLO
@ -269,17 +424,34 @@ static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
{ 0x04BC, 0x04BD }, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE { 0x04BC, 0x04BD }, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE
{ 0x04BE, 0x04BF }, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER { 0x04BE, 0x04BF }, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER
{ 0x04C0, 0x04CF }, // CYRILLIC LETTER PALOCHKA { 0x04C0, 0x04CF }, // CYRILLIC LETTER PALOCHKA
{ 0x04C1, 0x04C2 }, // CYRILLIC CAPITAL LETTER ZHE WITH BREVE
{ 0x04C3, 0x04C4 }, // CYRILLIC CAPITAL LETTER KA WITH HOOK { 0x04C3, 0x04C4 }, // CYRILLIC CAPITAL LETTER KA WITH HOOK
{ 0x04C5, 0x04C6 }, // CYRILLIC CAPITAL LETTER EL WITH TAIL { 0x04C5, 0x04C6 }, // CYRILLIC CAPITAL LETTER EL WITH TAIL
{ 0x04C7, 0x04C8 }, // CYRILLIC CAPITAL LETTER EN WITH HOOK { 0x04C7, 0x04C8 }, // CYRILLIC CAPITAL LETTER EN WITH HOOK
{ 0x04C9, 0x04CA }, // CYRILLIC CAPITAL LETTER EN WITH TAIL { 0x04C9, 0x04CA }, // CYRILLIC CAPITAL LETTER EN WITH TAIL
{ 0x04CB, 0x04CC }, // CYRILLIC CAPITAL LETTER KHAKASSIAN CHE { 0x04CB, 0x04CC }, // CYRILLIC CAPITAL LETTER KHAKASSIAN CHE
{ 0x04CD, 0x04CE }, // CYRILLIC CAPITAL LETTER EM WITH TAIL { 0x04CD, 0x04CE }, // CYRILLIC CAPITAL LETTER EM WITH TAIL
{ 0x04D0, 0x04D1 }, // CYRILLIC CAPITAL LETTER A WITH BREVE
{ 0x04D2, 0x04D3 }, // CYRILLIC CAPITAL LETTER A WITH DIAERESIS
{ 0x04D4, 0x04D5 }, // CYRILLIC CAPITAL LIGATURE A IE { 0x04D4, 0x04D5 }, // CYRILLIC CAPITAL LIGATURE A IE
{ 0x04D6, 0x04D7 }, // CYRILLIC CAPITAL LETTER IE WITH BREVE
{ 0x04D8, 0x04D9 }, // CYRILLIC CAPITAL LETTER SCHWA { 0x04D8, 0x04D9 }, // CYRILLIC CAPITAL LETTER SCHWA
{ 0x04DA, 0x04DB }, // CYRILLIC CAPITAL LETTER SCHWA WITH DIAERESIS
{ 0x04DC, 0x04DD }, // CYRILLIC CAPITAL LETTER ZHE WITH DIAERESIS
{ 0x04DE, 0x04DF }, // CYRILLIC CAPITAL LETTER ZE WITH DIAERESIS
{ 0x04E0, 0x04E1 }, // CYRILLIC CAPITAL LETTER ABKHASIAN DZE { 0x04E0, 0x04E1 }, // CYRILLIC CAPITAL LETTER ABKHASIAN DZE
{ 0x04E2, 0x04E3 }, // CYRILLIC CAPITAL LETTER I WITH MACRON
{ 0x04E4, 0x04E5 }, // CYRILLIC CAPITAL LETTER I WITH DIAERESIS
{ 0x04E6, 0x04E7 }, // CYRILLIC CAPITAL LETTER O WITH DIAERESIS
{ 0x04E8, 0x04E9 }, // CYRILLIC CAPITAL LETTER BARRED O { 0x04E8, 0x04E9 }, // CYRILLIC CAPITAL LETTER BARRED O
{ 0x04EA, 0x04EB }, // CYRILLIC CAPITAL LETTER BARRED O WITH DIAERESIS
{ 0x04EC, 0x04ED }, // CYRILLIC CAPITAL LETTER E WITH DIAERESIS
{ 0x04EE, 0x04EF }, // CYRILLIC CAPITAL LETTER U WITH MACRON
{ 0x04F0, 0x04F1 }, // CYRILLIC CAPITAL LETTER U WITH DIAERESIS
{ 0x04F2, 0x04F3 }, // CYRILLIC CAPITAL LETTER U WITH DOUBLE ACUTE
{ 0x04F4, 0x04F5 }, // CYRILLIC CAPITAL LETTER CHE WITH DIAERESIS
{ 0x04F6, 0x04F7 }, // CYRILLIC CAPITAL LETTER GHE WITH DESCENDER { 0x04F6, 0x04F7 }, // CYRILLIC CAPITAL LETTER GHE WITH DESCENDER
{ 0x04F8, 0x04F9 }, // CYRILLIC CAPITAL LETTER YERU WITH DIAERESIS
{ 0x04FA, 0x04FB }, // CYRILLIC CAPITAL LETTER GHE WITH STROKE AND HOOK { 0x04FA, 0x04FB }, // CYRILLIC CAPITAL LETTER GHE WITH STROKE AND HOOK
{ 0x04FC, 0x04FD }, // CYRILLIC CAPITAL LETTER HA WITH HOOK { 0x04FC, 0x04FD }, // CYRILLIC CAPITAL LETTER HA WITH HOOK
{ 0x04FE, 0x04FF }, // CYRILLIC CAPITAL LETTER HA WITH STROKE { 0x04FE, 0x04FF }, // CYRILLIC CAPITAL LETTER HA WITH STROKE

3
native/jni/src/char_utils.h
View File

@ -58,7 +58,8 @@ inline static int toBaseCodePoint(int c) {
AK_FORCE_INLINE static int toLowerCase(const int c) { AK_FORCE_INLINE static int toLowerCase(const int c) {
if (isAsciiUpper(c)) { if (isAsciiUpper(c)) {
return toAsciiLower(c); return toAsciiLower(c);
} else if (isAscii(c)) { }
if (isAscii(c)) {
return c; return c;
} }
return static_cast<int>(latin_tolower(static_cast<unsigned short>(c))); return static_cast<int>(latin_tolower(static_cast<unsigned short>(c)));