uint32_t
FCDUIterCollationIterator::handleNextCE32(UChar32 &c, UErrorCode &errorCode) {
for(;;) {
if(state == ITER_CHECK_FWD) {
c = iter.next(&iter);
if(c < 0) {
return Collation::FALLBACK_CE32;
}
if(CollationFCD::hasTccc(c)) {
if(CollationFCD::maybeTibetanCompositeVowel(c) ||
CollationFCD::hasLccc(iter.current(&iter))) {
iter.previous(&iter);
if(!nextSegment(errorCode)) {
c = U_SENTINEL;
return Collation::FALLBACK_CE32;
}
continue;
}
}
break;
} else if(state == ITER_IN_FCD_SEGMENT && pos != limit) {
c = iter.next(&iter);
++pos;
U_ASSERT(c >= 0);
break;
} else if(state >= IN_NORM_ITER_AT_LIMIT && pos != normalized.length()) {
c = normalized[pos++];
break;
} else {
switchToForward();
}
}
return UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c);
}
UChar32
FCDUIterCollationIterator::nextCodePoint(UErrorCode &errorCode) {
UChar32 c;
for(;;) {
if(state == ITER_CHECK_FWD) {
c = iter.next(&iter);
if(c < 0) {
return c;
}
if(CollationFCD::hasTccc(c)) {
if(CollationFCD::maybeTibetanCompositeVowel(c) ||
CollationFCD::hasLccc(iter.current(&iter))) {
iter.previous(&iter);
if(!nextSegment(errorCode)) {
return U_SENTINEL;
}
continue;
}
}
if(U16_IS_LEAD(c)) {
UChar32 trail = iter.next(&iter);
if(U16_IS_TRAIL(trail)) {
return U16_GET_SUPPLEMENTARY(c, trail);
} else if(trail >= 0) {
iter.previous(&iter);
}
}
return c;
} else if(state == ITER_IN_FCD_SEGMENT && pos != limit) {
c = uiter_next32(&iter);
pos += U16_LENGTH(c);
U_ASSERT(c >= 0);
return c;
} else if(state >= IN_NORM_ITER_AT_LIMIT && pos != normalized.length()) {
c = normalized.char32At(pos);
pos += U16_LENGTH(c);
return c;
} else {
switchToForward();
}
}
}
UChar32
FCDUTF8CollationIterator::nextCodePoint(UErrorCode &errorCode) {
UChar32 c;
for(;;) {
if(state == CHECK_FWD) {
if(pos == length || ((c = u8[pos]) == 0 && length < 0)) {
return U_SENTINEL;
}
if(c < 0x80) {
++pos;
return c;
}
U8_NEXT_OR_FFFD(u8, pos, length, c);
if(CollationFCD::hasTccc(c <= 0xffff ? c : U16_LEAD(c)) &&
(CollationFCD::maybeTibetanCompositeVowel(c) ||
(pos != length && nextHasLccc()))) {
// c is not FCD-inert, therefore it is not U+FFFD and it has a valid byte sequence
// and we can use U8_LENGTH() rather than a previous-position variable.
pos -= U8_LENGTH(c);
if(!nextSegment(errorCode)) {
return U_SENTINEL;
}
continue;
}
return c;
} else if(state == IN_FCD_SEGMENT && pos != limit) {
U8_NEXT_OR_FFFD(u8, pos, length, c);
return c;
} else if(state == IN_NORMALIZED && pos != normalized.length()) {
c = normalized.char32At(pos);
pos += U16_LENGTH(c);
return c;
} else {
switchToForward();
}
}
}
uint32_t
FCDUTF8CollationIterator::handleNextCE32(UChar32 &c, UErrorCode &errorCode) {
for(;;) {
if(state == CHECK_FWD) {
// Combination of UTF8CollationIterator::handleNextCE32() with FCD check fastpath.
if(pos == length) {
c = U_SENTINEL;
return Collation::FALLBACK_CE32;
}
c = u8[pos++];
if(c < 0xc0) {
// ASCII 00..7F; trail bytes 80..BF map to error values.
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);
pos += 2;
if(CollationFCD::hasTccc(c) &&
(CollationFCD::maybeTibetanCompositeVowel(c) ||
(pos != length && nextHasLccc()))) {
pos -= 3;
} else {
break; // return CE32(BMP)
}
} else {
// Function call for supplementary code points and error cases.
// Illegal byte sequences yield U+FFFD.
c = utf8_nextCharSafeBody(u8, &pos, length, c, -3);
if(c == 0xfffd) {
return Collation::FFFD_CE32;
} else {
U_ASSERT(c > 0xffff);
if(CollationFCD::hasTccc(U16_LEAD(c)) && pos != length && nextHasLccc()) {
pos -= 4;
} else {
return data->getCE32FromSupplementary(c);
}
}
}
if(!nextSegment(errorCode)) {
c = U_SENTINEL;
return Collation::FALLBACK_CE32;
}
continue;
} else if(state == IN_FCD_SEGMENT && pos != limit) {
return UTF8CollationIterator::handleNextCE32(c, errorCode);
} else if(state == IN_NORMALIZED && pos != normalized.length()) {
c = normalized[pos++];
break;
} else {
switchToForward();
}
}
return UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c);
}
