bug 732443 - handle surrogate pairs correctly in ClusterIterator. r=emk

This commit is contained in:
Jonathan Kew 2012-03-03 09:16:45 +00:00
parent 30ad782016
commit 9e0b2f2f6c

View File

@ -292,10 +292,13 @@ ClusterIterator::Next()
PRUint32 ch = *mPos++;
// Handle conjoining Jamo that make Hangul syllables
if ((ch & ~0xff) == 0x1100 ||
if (NS_IS_HIGH_SURROGATE(ch) && mPos < mLimit &&
NS_IS_LOW_SURROGATE(*mPos)) {
ch = SURROGATE_TO_UCS4(ch, *mPos++);
} else if ((ch & ~0xff) == 0x1100 ||
(ch >= 0xa960 && ch <= 0xa97f) ||
(ch >= 0xac00 && ch <= 0xd7ff)) {
// Handle conjoining Jamo that make Hangul syllables
HSType hangulState = GetHangulSyllableType(ch);
while (mPos < mLimit) {
ch = *mPos;
@ -337,16 +340,19 @@ ClusterIterator::Next()
// Check for surrogate pairs; note that isolated surrogates will just
// be treated as generic (non-cluster-extending) characters here,
// which is fine for cluster-iterating purposes
if (NS_IS_LOW_SURROGATE(ch) &&
NS_IS_HIGH_SURROGATE(*(mPos - 1))) {
ch = SURROGATE_TO_UCS4(*(mPos - 1), *mPos);
mPos++;
if (NS_IS_HIGH_SURROGATE(ch) && mPos < mLimit - 1 &&
NS_IS_LOW_SURROGATE(*(mPos + 1))) {
ch = SURROGATE_TO_UCS4(ch, *(mPos + 1));
}
if (!IsClusterExtender(ch)) {
break;
}
mPos++;
if (!IS_IN_BMP(ch)) {
mPos++;
}
}
}