wuggy/Dactyloidae
wuggy/Dactyloidae
1
0
Fork 0
mirror of https://repo.dactyloidae.xyz/Dactyloidae/UXP.git synced 2026-09-20 07:17:32 +09:00
Code Issues Projects Releases Packages Wiki Activity

Issue #1971 - Part 2: Update ICU source to 63.2.

Browse source
This commit is contained in:
Job Bautista 2022-07-24 21:03:27 +08:00 • committed by roytam1
commit 1e69214382
3160 changed files with 275815 additions and 234203 deletions
Download patch file Download diff file Expand all files Collapse all files

70
intl/icu/source/i18n/utf8collationiterator.cpp
View file

@ -1,4 +1,4 @@
// Copyright (C) 2016 and later: Unicode, Inc. and others.
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
@ -49,26 +49,25 @@ UTF8CollationIterator::handleNextCE32(UChar32 &c, UErrorCode & /*errorCode*/) {
}
// Optimized combination of U8_NEXT_OR_FFFD() and UTRIE2_U8_NEXT32().
c = u8[pos++];
if(c < 0xc0) {
// ASCII 00..7F; trail bytes 80..BF map to error values.
if(U8_IS_SINGLE(c)) {
// ASCII 00..7F
return trie->data32[c];
}
uint8_t t1, t2;
if(c < 0xe0 && pos != length && (t1 = (u8[pos] - 0x80)) <= 0x3f) {
// U+0080..U+07FF; 00..7F map to error values.
if(0xe0 <= c && c < 0xf0 &&
((pos + 1) < length || length < 0) &&
U8_IS_VALID_LEAD3_AND_T1(c, t1 = u8[pos]) &&
(t2 = (u8[pos + 1] - 0x80)) <= 0x3f) {
// U+0800..U+FFFF except surrogates
c = (((c & 0xf) << 12) | ((t1 & 0x3f) << 6) | t2);
pos += 2;
return UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c);
} else if(c < 0xe0 && c >= 0xc2 && pos != length && (t1 = (u8[pos] - 0x80)) <= 0x3f) {
// U+0080..U+07FF
uint32_t ce32 = trie->data32[trie->index[(UTRIE2_UTF8_2B_INDEX_2_OFFSET - 0xc0) + c] + t1];
c = ((c & 0x1f) << 6) | t1;
++pos;
return ce32;
} else if(c <= 0xef &&
((pos + 1) < length || length < 0) &&
(t1 = (u8[pos] - 0x80)) <= 0x3f && (c != 0xe0 || t1 >= 0x20) &&
(t2 = (u8[pos + 1] - 0x80)) <= 0x3f
) {
// U+0800..U+FFFF; caller maps surrogates to error values.
c = (UChar)((c << 12) | (t1 << 6) | t2);
pos += 2;
return UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c);
} else {
// Function call for supplementary code points and error cases.
// Illegal byte sequences yield U+FFFD.
@ -158,28 +157,17 @@ FCDUTF8CollationIterator::handleNextCE32(UChar32 &c, UErrorCode &errorCode) {
return Collation::FALLBACK_CE32;
}
c = u8[pos++];
if(c < 0xc0) {
// ASCII 00..7F; trail bytes 80..BF map to error values.
if(U8_IS_SINGLE(c)) {
// ASCII 00..7F
return trie->data32[c];
}
uint8_t t1, t2;
if(c < 0xe0 && pos != length && (t1 = (u8[pos] - 0x80)) <= 0x3f) {
// U+0080..U+07FF; 00..7F map to error values.
uint32_t ce32 = trie->data32[trie->index[(UTRIE2_UTF8_2B_INDEX_2_OFFSET - 0xc0) + c] + t1];
c = ((c & 0x1f) << 6) | t1;
++pos;
if(CollationFCD::hasTccc(c) && pos != length && nextHasLccc()) {
pos -= 2;
} else {
return ce32;
}
} else if(c <= 0xef &&
((pos + 1) < length || length < 0) &&
(t1 = (u8[pos] - 0x80)) <= 0x3f && (c != 0xe0 || t1 >= 0x20) &&
(t2 = (u8[pos + 1] - 0x80)) <= 0x3f
) {
// U+0800..U+FFFF; caller maps surrogates to error values.
c = (UChar)((c << 12) | (t1 << 6) | t2);
if(0xe0 <= c && c < 0xf0 &&
((pos + 1) < length || length < 0) &&
U8_IS_VALID_LEAD3_AND_T1(c, t1 = u8[pos]) &&
(t2 = (u8[pos + 1] - 0x80)) <= 0x3f) {
// U+0800..U+FFFF except surrogates
c = (((c & 0xf) << 12) | ((t1 & 0x3f) << 6) | t2);
pos += 2;
if(CollationFCD::hasTccc(c) &&
(CollationFCD::maybeTibetanCompositeVowel(c) ||
@ -188,6 +176,16 @@ FCDUTF8CollationIterator::handleNextCE32(UChar32 &c, UErrorCode &errorCode) {
} else {
break; // return CE32(BMP)
}
} else if(c < 0xe0 && c >= 0xc2 && pos != length && (t1 = (u8[pos] - 0x80)) <= 0x3f) {
// U+0080..U+07FF
uint32_t ce32 = trie->data32[trie->index[(UTRIE2_UTF8_2B_INDEX_2_OFFSET - 0xc0) + c] + t1];
c = ((c & 0x1f) << 6) | t1;
++pos;
if(CollationFCD::hasTccc(c) && pos != length && nextHasLccc()) {
pos -= 2;
} else {
return ce32;
}
} else {
// Function call for supplementary code points and error cases.
// Illegal byte sequences yield U+FFFD.
@ -237,7 +235,7 @@ UBool
FCDUTF8CollationIterator::previousHasTccc() const {
U_ASSERT(state == CHECK_BWD && pos != 0);
UChar32 c = u8[pos - 1];
if(c < 0x80) { return FALSE; }
if(U8_IS_SINGLE(c)) { return FALSE; }
int32_t i = pos;
U8_PREV_OR_FFFD(u8, 0, i, c);
if(c > 0xffff) { c = U16_LEAD(c); }
@ -271,7 +269,7 @@ FCDUTF8CollationIterator::nextCodePoint(UErrorCode &errorCode) {
if(pos == length || ((c = u8[pos]) == 0 && length < 0)) {
return U_SENTINEL;
}
if(c < 0x80) {
if(U8_IS_SINGLE(c)) {
++pos;
return c;
}
@ -309,7 +307,7 @@ FCDUTF8CollationIterator::previousCodePoint(UErrorCode &errorCode) {
if(pos == 0) {
return U_SENTINEL;
}
if((c = u8[pos - 1]) < 0x80) {
if(U8_IS_SINGLE(c = u8[pos - 1])) {
--pos;
return c;
}