433 lines
11 KiB
C
433 lines
11 KiB
C
|
/* This software is licensed by the MIT License, see LICENSE file */
|
||
|
|
/* Copyright © 2022 Gregory Lirent */
|
||
|
|
|
||
|
|
#include "../../build.h"
|
||
|
|
|
||
|
|
static unsigned int toupper_latin_extended_a(unsigned int uc) { // 0x0100—0x017f
|
||
|
|
|
||
|
|
if ( ( ( (uc >= 0x0100 && uc <= 0x0137) ||
|
||
|
|
(uc >= 0x014a && uc <= 0x0177) ) && (uc&1) ) ||
|
||
|
|
( ( (uc >= 0x0139 && uc <= 0x0148) ||
|
||
|
|
(uc >= 0x0179 && uc <= 0x017e) ) && !(uc&1) )
|
||
|
|
) return uc - 1;
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_latin_extended_b(unsigned int uc) { // 0x0180—0x024f
|
||
|
|
|
||
|
|
#ifndef LATIN_EXTENDED_B_DISABLE
|
||
|
|
if ( ( ( (uc >= 0x01a0 && uc <= 0x01a5) ||
|
||
|
|
(uc >= 0x01de && uc <= 0x01ef) ||
|
||
|
|
(uc >= 0x01f8 && uc <= 0x01ff) ||
|
||
|
|
(uc >= 0x0200 && uc <= 0x021b) ||
|
||
|
|
(uc >= 0x0222 && uc <= 0x0233) ||
|
||
|
|
(uc >= 0x0246 && uc <= 0x024f) ) && (uc&1) ) ||
|
||
|
|
( ( (uc >= 0x0189 && uc <= 0x0192) ||
|
||
|
|
(uc >= 0x01cd && uc <= 0x01dc) ) && !(uc&1) )
|
||
|
|
) return uc - 1;
|
||
|
|
|
||
|
|
|
||
|
|
switch (uc) { default: break;
|
||
|
|
|
||
|
|
case 0x01c6:
|
||
|
|
case 0x01c9:
|
||
|
|
case 0x01cc:
|
||
|
|
case 0x01f3: return uc - 2; // or -1
|
||
|
|
|
||
|
|
case 0x0183:
|
||
|
|
case 0x0185:
|
||
|
|
case 0x0188:
|
||
|
|
case 0x0199:
|
||
|
|
case 0x01a7:
|
||
|
|
case 0x01ad:
|
||
|
|
case 0x01b0:
|
||
|
|
case 0x01b4:
|
||
|
|
case 0x01b6:
|
||
|
|
case 0x01b9:
|
||
|
|
case 0x01bd:
|
||
|
|
case 0x01f5:
|
||
|
|
case 0x021d:
|
||
|
|
case 0x021f:
|
||
|
|
case 0x023c:
|
||
|
|
case 0x023d:
|
||
|
|
case 0x0242: return uc - 1;
|
||
|
|
|
||
|
|
case 0x0180: return 0x0243;
|
||
|
|
case 0x0195: return 0x01f6;
|
||
|
|
case 0x019a: return 0x023d;
|
||
|
|
case 0x019e: return 0x0220;
|
||
|
|
case 0x01dd: return 0x018e;
|
||
|
|
|
||
|
|
#ifndef LATIN_EXTENDED_C_DISABLE
|
||
|
|
case 0x023f: return 0x2c7e;
|
||
|
|
case 0x0240: return 0x2c7f;
|
||
|
|
#endif
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_latin_extended_c(unsigned int uc) { // 0x2c60—0x2c7f
|
||
|
|
#ifndef LATIN_EXTENDED_C_DISABLE
|
||
|
|
if ( (uc >= 0x2c67 && uc <= 0x2c6c) && !(uc&1) )
|
||
|
|
return uc - 1;
|
||
|
|
|
||
|
|
switch (uc) { default: break;
|
||
|
|
|
||
|
|
case 0x2c73:
|
||
|
|
case 0x2c61: return uc - 1;
|
||
|
|
|
||
|
|
#ifndef LATIN_EXTENDED_B_DISABLE
|
||
|
|
case 0x2c65: return 0x023a;
|
||
|
|
#endif
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_latin_extended_d(unsigned int uc) { // 0xa720—0xa7ff
|
||
|
|
#ifndef LATIN_EXTENDED_D_DISABLE
|
||
|
|
if ( ( ( (uc >= 0xa796 && uc <= 0xa7a9) ||
|
||
|
|
(uc >= 0xa7b4 && uc <= 0xa7bf) ||
|
||
|
|
(uc >= 0xa722 && uc <= 0xa72f) ||
|
||
|
|
(uc >= 0xa732 && uc <= 0xa76f) ||
|
||
|
|
(uc >= 0xa77e && uc <= 0xa787) ) && (uc&1) )
|
||
|
|
) return uc - 1;
|
||
|
|
|
||
|
|
switch (uc) { default: break;
|
||
|
|
case 0xa7c3:
|
||
|
|
case 0xa7c8:
|
||
|
|
case 0xa7ca:
|
||
|
|
case 0xa791:
|
||
|
|
case 0xa793:
|
||
|
|
case 0xa77a:
|
||
|
|
case 0xa77c: return uc - 1;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_ipa_extensions(unsigned int uc) { // 0x0250-0x02af
|
||
|
|
#ifndef IPA_EXTENSIONS_DISABLE
|
||
|
|
switch (uc) { default: break;
|
||
|
|
#ifndef LATIN_EXTENDED_B_DISABLE
|
||
|
|
case 0x0259: return 0x01bf;
|
||
|
|
case 0x025b: return 0x0190;
|
||
|
|
case 0x0260: return 0x0193;
|
||
|
|
case 0x0263: return 0x0194;
|
||
|
|
case 0x0269: return 0x0196;
|
||
|
|
case 0x0268: return 0x0197;
|
||
|
|
case 0x026f: return 0x019c;
|
||
|
|
case 0x0272: return 0x019d;
|
||
|
|
case 0x0275: return 0x019f;
|
||
|
|
case 0x0280: return 0x01a6;
|
||
|
|
case 0x0283: return 0x01a9;
|
||
|
|
case 0x0288: return 0x01ae;
|
||
|
|
case 0x028a: return 0x01b1;
|
||
|
|
case 0x028b: return 0x01b2;
|
||
|
|
case 0x0292: return 0x01b7;
|
||
|
|
case 0x02bf: return 0x01f7;
|
||
|
|
case 0x0253: return 0x0181;
|
||
|
|
case 0x0254: return 0x0186;
|
||
|
|
case 0x0289: return 0x0244;
|
||
|
|
case 0x028c: return 0x0245;
|
||
|
|
#endif
|
||
|
|
#ifndef LATIN_EXTENDED_C_DISABLE
|
||
|
|
case 0x0250: return 0x2c6f;
|
||
|
|
case 0x0251: return 0x2c6d;
|
||
|
|
case 0x0252: return 0x2c70;
|
||
|
|
case 0x026b: return 0x2c62;
|
||
|
|
case 0x0271: return 0x2c6e;
|
||
|
|
case 0x027d: return 0x2c64;
|
||
|
|
#endif
|
||
|
|
#ifndef LATIN_EXTENDED_D_DISABLE
|
||
|
|
case 0x025c: return 0xa7ab;
|
||
|
|
case 0x0261: return 0xa7ac;
|
||
|
|
case 0x0265: return 0xa78d;
|
||
|
|
case 0x0266: return 0xa7aa;
|
||
|
|
case 0x026a: return 0xa7ae;
|
||
|
|
case 0x026c: return 0xa7ad;
|
||
|
|
case 0x0287: return 0xa7b1;
|
||
|
|
case 0x029d: return 0xa7b2;
|
||
|
|
case 0x029e: return 0xa7b0;
|
||
|
|
#endif
|
||
|
|
#ifndef LATIN_EXTENDED_ADDITIONAL_DISABLE
|
||
|
|
case 0x00df: return 0x1e9e;
|
||
|
|
#endif
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_latin_extended_additional(unsigned int uc) { // 0x1e00-0x1eff
|
||
|
|
#ifndef LATIN_EXTENDED_ADDITIONAL_DISABLE
|
||
|
|
if ( ( ( (uc >= 0x1e00 && uc <= 0x1e95) ||
|
||
|
|
(uc >= 0x1ea0 && uc <= 0x1eff) ) && (uc&1) )
|
||
|
|
) return uc - 1;
|
||
|
|
|
||
|
|
switch (uc) { default: break;
|
||
|
|
case 0x1e9b: return 0x1e60;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_greek_and_coptic(unsigned int uc) { // 0x0370-0x03ff
|
||
|
|
|
||
|
|
if ((uc >= 0x03b1 && uc <= 0x03c1) || (uc >= 0x03c4 && uc <= 0x03cb)) {
|
||
|
|
return uc - 0x20;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x03d8 && uc <= 0x03ef && (uc&1)) {
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
switch (uc) { default: return uc;
|
||
|
|
case 0x03ad:
|
||
|
|
case 0x03ae:
|
||
|
|
case 0x03af: return uc - 0x25;
|
||
|
|
|
||
|
|
case 0x03cd:
|
||
|
|
case 0x03ce: ++uc;
|
||
|
|
case 0x03cc: return uc - 0x40;
|
||
|
|
|
||
|
|
case 0x037b:
|
||
|
|
case 0x037c:
|
||
|
|
case 0x037d: return uc + 0x82;
|
||
|
|
|
||
|
|
case 0x0371:
|
||
|
|
case 0x0373:
|
||
|
|
case 0x0377:
|
||
|
|
case 0x03f8:
|
||
|
|
case 0x03fb: return uc - 1;
|
||
|
|
|
||
|
|
case 0x03f3: return 0x037f;
|
||
|
|
case 0x03ac: return 0x0386;
|
||
|
|
case 0x03f2: return 0x03f9;
|
||
|
|
case 0x03f0: return 0x039a;
|
||
|
|
case 0x03f1: return 0x03a1;
|
||
|
|
case 0x03f5: return 0x0395;
|
||
|
|
case 0x03c3:
|
||
|
|
case 0x03c2: return 0x03a3;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_greek_extended(unsigned int uc) { // 0x1f00-0x1fff
|
||
|
|
#ifndef GREEK_EXTENDED_DISABLE
|
||
|
|
if (((uc >= 0x1f00 && uc <= 0x1f6f) || (uc >= 0x1f80 && uc <= 0x1faf)) && !(uc&0x08)) {
|
||
|
|
switch (uc) { default: break;
|
||
|
|
case 0x1f16:
|
||
|
|
case 0x1f17:
|
||
|
|
case 0x1f46:
|
||
|
|
case 0x1f47:
|
||
|
|
case 0x1f50:
|
||
|
|
case 0x1f52:
|
||
|
|
case 0x1f54:
|
||
|
|
case 0x1f56: return uc;
|
||
|
|
}
|
||
|
|
return uc | 0x08;
|
||
|
|
}
|
||
|
|
|
||
|
|
switch (uc) { default: break;
|
||
|
|
case 0x1fbe: return 0x0399;
|
||
|
|
|
||
|
|
case 0x1f70:
|
||
|
|
case 0x1f71: return uc + 0x4a;
|
||
|
|
|
||
|
|
case 0x1f72:
|
||
|
|
case 0x1f73:
|
||
|
|
case 0x1f74:
|
||
|
|
case 0x1f75: return uc + 0x56;
|
||
|
|
|
||
|
|
case 0x1f76:
|
||
|
|
case 0x1f77: return uc + 0x64;
|
||
|
|
|
||
|
|
case 0x1f7a:
|
||
|
|
case 0x1f7b: uc |= 0x10;
|
||
|
|
case 0x1f78:
|
||
|
|
case 0x1f79: return uc | 0x80;
|
||
|
|
|
||
|
|
case 0x1fb0:
|
||
|
|
case 0x1fb1:
|
||
|
|
case 0x1fe0:
|
||
|
|
case 0x1fe1:
|
||
|
|
case 0x1fd0:
|
||
|
|
case 0x1fd1: return uc | 0x08;
|
||
|
|
|
||
|
|
case 0x1f7c:
|
||
|
|
case 0x1f7d: return uc + 0x7e;
|
||
|
|
|
||
|
|
case 0x1fe5: uc -= 2;
|
||
|
|
case 0x1fc3:
|
||
|
|
case 0x1ff3: return uc + 0x09;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
static unsigned int toupper_exclusions(unsigned int uc) {
|
||
|
|
|
||
|
|
switch (uc) { default: break;
|
||
|
|
case 0x04cf: return 0x04c0;
|
||
|
|
case 0x00ff: return 0x0178;
|
||
|
|
|
||
|
|
case 0x10fd: return 0x1cbd;
|
||
|
|
case 0x10fe: return 0x1cbe;
|
||
|
|
case 0x10ff: return 0x1cbf;
|
||
|
|
|
||
|
|
#if !defined(LATIN_EXTENDED_E_DISABLE) && !defined(LATIN_EXTENDED_D_DISABLE)
|
||
|
|
case 0xab53: return 0xa7b3;
|
||
|
|
#endif
|
||
|
|
#ifndef LATIN_EXTENDED_ADDITIONAL_DISABLE
|
||
|
|
case 0x00df: return 0x1e9e;
|
||
|
|
#endif
|
||
|
|
#ifndef PHONETIC_EXTENSIONS_DISABLE
|
||
|
|
#ifndef LATIN_EXTENDED_C_DISABLE
|
||
|
|
case 0x1d7d: return 0x2c63;
|
||
|
|
#endif
|
||
|
|
#ifndef LATIN_EXTENDED_D_DISABLE
|
||
|
|
case 0x1d79: return 0xa77d;
|
||
|
|
#endif
|
||
|
|
#endif
|
||
|
|
|
||
|
|
#ifndef CYRILLIC_EXTENDED_C_DISABLE
|
||
|
|
case 0x1c80: return 0x0412;
|
||
|
|
case 0x1c81: return 0x0414;
|
||
|
|
case 0x1c82: return 0x041e;
|
||
|
|
case 0x1c83: return 0x0421;
|
||
|
|
case 0x1c84: return 0x0422;
|
||
|
|
case 0x1c85: return 0x0422;
|
||
|
|
case 0x1c86: return 0x042a;
|
||
|
|
case 0x1c87: return 0x0462;
|
||
|
|
#ifndef CYRILLIC_EXTENDED_B_DISABLE
|
||
|
|
case 0x1c88: return 0xa64a;
|
||
|
|
#endif
|
||
|
|
#endif
|
||
|
|
|
||
|
|
#ifndef COPTIC_DISABLE
|
||
|
|
case 0x2cf3:
|
||
|
|
case 0x2cec:
|
||
|
|
case 0x2cee: return uc - 1;
|
||
|
|
#endif
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x0460 && uc <= 0x0481 && (uc&1)) { // Cyrillic
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x048a && uc <= 0x04bf && (uc&1)) { // Cyrillic
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x04c1 && uc <= 0x04ce && !(uc&1)) { // Cyrillic
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x04d0 && uc <= 0x04ff && (uc&1)) { // Cyrillic
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
#ifndef CYRILLIC_SUPLEMENT_DISABLE
|
||
|
|
if (uc >= 0x0500 && uc <= 0x052f && (uc&1)) { // Cyrillic Supplement
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef CYRILLIC_EXTENDED_B_DISABLE
|
||
|
|
if (uc >= 0xa680 && uc <= 0xa69b && (uc&1)) { // Cyrillic Extended-B
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
if (uc >= 0xa640 && uc <= 0xa66d && (uc&1)) { // Cyrillic Extended-B
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
if (uc >= 0x0561 && uc <= 0x0586) { // Armenian
|
||
|
|
return uc - 0x30;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x10d0 && uc <= 0x10fa) { // Georgian
|
||
|
|
return uc + 0x0bc0;
|
||
|
|
}
|
||
|
|
#ifndef CHEROKEE_DISABLE
|
||
|
|
if (uc >= 0x13f8 && uc <= 0x13fd) { // Cherokee
|
||
|
|
return uc ^ 0x08;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef GLAGOLITIC_DISABLE
|
||
|
|
if (uc >= 0x2c30 && uc <= 0x2c5e) { // Glagolitic
|
||
|
|
return uc - 0x30;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef COPTIC_DISABLE
|
||
|
|
if (uc >= 0x2c80 && uc <= 0x2ce3 && (uc&1)) { // Coptic
|
||
|
|
return uc - 1;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef GEORGIAN_SUPPLEMENT_DISABLE
|
||
|
|
if ((uc >= 0x2d00 && uc <= 0x2d25) || uc == 0x2d27 || uc == 0x2d2d) { // Georgian Supplement
|
||
|
|
return uc - 0x1c60;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef CHEROKEE_DISABLE
|
||
|
|
if (uc >= 0xab70 && uc <= 0xabbf) { // Cherokee Supplement
|
||
|
|
return uc - 0x97d0;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef DESERET_DISABLE
|
||
|
|
if (uc >= 0x10428 && uc <= 0x1044f) { // Deseret
|
||
|
|
return uc - 0x28;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef OSAGE_DISABLE
|
||
|
|
if (uc >= 0x104d8 && uc <= 0x104fb) { // Osage
|
||
|
|
return uc - 0x28;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef OLD_HUNGARIAN_DISABLE
|
||
|
|
if (uc >= 0x10cc0 && uc <= 0x10cf2) { // Old Hungarian
|
||
|
|
return uc ^ 0x40;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
#ifndef WARANG_CITI_DISABLE
|
||
|
|
if (uc >= 0x118c0 && uc <= 0x118df) { // Warang Citi
|
||
|
|
return uc ^ 0x60;
|
||
|
|
}
|
||
|
|
#endif
|
||
|
|
|
||
|
|
return uc;
|
||
|
|
}
|
||
|
|
|
||
|
|
unsigned int toupper_unicode2(unsigned int uc) {
|
||
|
|
|
||
|
|
if (uc >= 0x61 && uc <= 0x7a || (uc >= 0xe0 && uc <= 0xfe && uc != 0xf7)) {
|
||
|
|
return uc ^ 0x20;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x0430 && uc <= 0x044f) { // Cyrillic
|
||
|
|
return uc - 0x20;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x0450 && uc <= 0x045f) {
|
||
|
|
return uc - 0x50;
|
||
|
|
}
|
||
|
|
|
||
|
|
if (uc >= 0x0370 && uc <= 0x03ff) return toupper_greek_and_coptic(uc);
|
||
|
|
if (uc >= 0x1f00 && uc <= 0x1fff) return toupper_greek_extended(uc);
|
||
|
|
|
||
|
|
if (uc >= 0x0100 && uc <= 0x017f) return toupper_latin_extended_a(uc);
|
||
|
|
if (uc >= 0x0180 && uc <= 0x024f) return toupper_latin_extended_b(uc);
|
||
|
|
if (uc >= 0x2c60 && uc <= 0x2c7f) return toupper_latin_extended_c(uc);
|
||
|
|
if (uc >= 0xa720 && uc <= 0xa7ff) return toupper_latin_extended_d(uc);
|
||
|
|
if (uc >= 0x0250 && uc <= 0x02af) return toupper_ipa_extensions(uc);
|
||
|
|
if (uc >= 0x1e00 && uc <= 0x1eff) return toupper_latin_extended_additional(uc);
|
||
|
|
|
||
|
|
return toupper_exclusions(uc);
|
||
|
|
}
|