UBreakIterator* openLineBreakIterator(const AtomicString& locale)
{
bool localeIsEmpty = locale.isEmpty();
UErrorCode openStatus = U_ZERO_ERROR;
UBreakIterator* ubrkIter = ubrk_open(UBRK_LINE, localeIsEmpty ? currentTextBreakLocaleID() : locale.string().utf8().data(), 0, 0, &openStatus);
// locale comes from a web page and it can be invalid, leading ICU
// to fail, in which case we fall back to the default locale.
if (!localeIsEmpty && U_FAILURE(openStatus)) {
openStatus = U_ZERO_ERROR;
ubrkIter = ubrk_open(UBRK_LINE, currentTextBreakLocaleID(), 0, 0, &openStatus);
}
if (U_FAILURE(openStatus)) {
LOG_ERROR("ubrk_open failed with status %d", openStatus);
return nullptr;
}
return ubrkIter;
}
int KWQFindNextWordFromIndex(const QChar *chars, int len, int position, bool forward)
{
int pos = 0;
UErrorCode status = U_ZERO_ERROR;
UBreakIterator *boundary = ubrk_open(UBRK_WORD, (const char*)currentTextBreakLocaleID().c_str(), const_cast<UChar *>(reinterpret_cast<const UChar *>(chars)), len, &status);
if ( boundary && U_SUCCESS(status) ) {
if (forward) {
pos = ubrk_following(boundary, position);
if (pos == UBRK_DONE)
pos = len;
} else {
pos = ubrk_preceding(boundary, position);
if (pos == UBRK_DONE)
pos = 0;
}
ubrk_close(boundary);
}
return pos;
}
void KWQFindSentenceBoundary(const QChar *chars, int len, int position, int *start, int *end)
{
int startPos = 0;
int endPos = 0;
UErrorCode status = U_ZERO_ERROR;
UBreakIterator *boundary = ubrk_open(UBRK_SENTENCE, (const char*)currentTextBreakLocaleID().c_str(), const_cast<UChar *>(reinterpret_cast<const UChar *>(chars)), len, &status);
if ( boundary && U_SUCCESS(status) ) {
startPos = ubrk_preceding(boundary, position);
if (startPos == UBRK_DONE) {
startPos = 0;
}
endPos = ubrk_following(boundary, startPos);
if (endPos == UBRK_DONE)
endPos = len;
ubrk_close(boundary);
}
*start = startPos;
*end = endPos;
}
static TextBreakIterator* setUpIterator(bool& createdIterator, TextBreakIterator*& iterator,
UBreakIteratorType type, const UChar* string, int length)
{
if (!string)
return 0;
if (!createdIterator) {
UErrorCode openStatus = U_ZERO_ERROR;
iterator = reinterpret_cast<TextBreakIterator*>(ubrk_open(type, currentTextBreakLocaleID(), 0, 0, &openStatus));
createdIterator = true;
ASSERT_WITH_MESSAGE(U_SUCCESS(openStatus), "ICU could not open a break iterator: %s (%d)", u_errorName(openStatus), openStatus);
}
if (!iterator)
return 0;
UErrorCode setTextStatus = U_ZERO_ERROR;
ubrk_setText(reinterpret_cast<UBreakIterator*>(iterator), string, length, &setTextStatus);
if (U_FAILURE(setTextStatus))
return 0;
return iterator;
}
static UBreakIterator* initializeIterator(UBreakIteratorType type, const char* locale = currentTextBreakLocaleID())
{
UErrorCode openStatus = U_ZERO_ERROR;
UBreakIterator* iterator = ubrk_open(type, locale, 0, 0, &openStatus);
ASSERT_WITH_MESSAGE(U_SUCCESS(openStatus), "ICU could not open a break iterator: %s (%d)", u_errorName(openStatus), openStatus);
return iterator;
}
