virtual int32_t next(int32_t n) { return fDelegate->next(n); }
int32_t ULISentenceBreakIterator::next() {
int32_t n = fDelegate->next();
if(n == UBRK_DONE || // at end or
fBackwardsTrie.isNull()) { // .. no backwards table loaded == no exceptions
return n;
}
// OK, do we need to break here?
UErrorCode status = U_ZERO_ERROR;
// refresh text
fText.adoptInstead(fDelegate->getUText(fText.orphan(), status));
//if(debug2) u_printf("str, native len=%d\n", utext_nativeLength(fText.getAlias()));
do { // outer loop runs once per underlying break (from fDelegate).
// loops while 'n' points to an exception.
utext_setNativeIndex(fText.getAlias(), n); // from n..
fBackwardsTrie->reset();
UChar32 uch;
//if(debug2) u_printf(" n@ %d\n", n);
// Assume a space is following the '.' (so we handle the case: "Mr. /Brown")
if((uch=utext_previous32(fText.getAlias()))==(UChar32)0x0020) { // TODO: skip a class of chars here??
// TODO only do this the 1st time?
//if(debug2) u_printf("skipping prev: |%C| \n", (UChar)uch);
} else {
//if(debug2) u_printf("not skipping prev: |%C| \n", (UChar)uch);
uch = utext_next32(fText.getAlias());
//if(debug2) u_printf(" -> : |%C| \n", (UChar)uch);
}
UStringTrieResult r = USTRINGTRIE_INTERMEDIATE_VALUE;
int32_t bestPosn = -1;
int32_t bestValue = -1;
while((uch=utext_previous32(fText.getAlias()))!=U_SENTINEL && // more to consume backwards and..
USTRINGTRIE_HAS_NEXT(r=fBackwardsTrie->nextForCodePoint(uch))) {// more in the trie
if(USTRINGTRIE_HAS_VALUE(r)) { // remember the best match so far
bestPosn = utext_getNativeIndex(fText.getAlias());
bestValue = fBackwardsTrie->getValue();
}
//if(debug2) u_printf("rev< /%C/ cont?%d @%d\n", (UChar)uch, r, utext_getNativeIndex(fText.getAlias()));
}
if(USTRINGTRIE_MATCHES(r)) { // exact match?
//if(debug2) u_printf("rev<?/%C/?end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (UChar)uch, r, bestPosn, bestValue);
bestValue = fBackwardsTrie->getValue();
bestPosn = utext_getNativeIndex(fText.getAlias());
//if(debug2) u_printf("rev<+/%C/+end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (UChar)uch, r, bestPosn, bestValue);
}
if(bestPosn>=0) {
//if(debug2) u_printf("rev< /%C/ end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (UChar)uch, r, bestPosn, bestValue);
//if(USTRINGTRIE_MATCHES(r)) { // matched - so, now what?
//int32_t bestValue = fBackwardsTrie->getValue();
////if(debug2) u_printf("rev< /%C/ matched, skip..%d bestValue=%d\n", (UChar)uch, r, bestValue);
if(bestValue == kMATCH) { // exact match!
//if(debug2) u_printf(" exact backward match\n");
n = fDelegate->next(); // skip this one. Find the next lowerlevel break.
if(n==UBRK_DONE) return n;
continue; // See if the next is another exception.
} else if(bestValue == kPARTIAL
&& fForwardsPartialTrie.isValid()) { // make sure there's a forward trie
//if(debug2) u_printf(" partial backward match\n");
// We matched the "Ph." in "Ph.D." - now we need to run everything through the forwards trie
// to see if it matches something going forward.
fForwardsPartialTrie->reset();
UStringTrieResult rfwd = USTRINGTRIE_INTERMEDIATE_VALUE;
utext_setNativeIndex(fText.getAlias(), bestPosn); // hope that's close ..
//if(debug2) u_printf("Retrying at %d\n", bestPosn);
while((uch=utext_next32(fText.getAlias()))!=U_SENTINEL &&
USTRINGTRIE_HAS_NEXT(rfwd=fForwardsPartialTrie->nextForCodePoint(uch))) {
//if(debug2) u_printf("fwd> /%C/ cont?%d @%d\n", (UChar)uch, rfwd, utext_getNativeIndex(fText.getAlias()));
}
if(USTRINGTRIE_MATCHES(rfwd)) {
//if(debug2) u_printf("fwd> /%C/ == forward match!\n", (UChar)uch);
// only full matches here, nothing to check
// skip the next:
n = fDelegate->next();
if(n==UBRK_DONE) return n;
continue;
} else {
//if(debug2) u_printf("fwd> /%C/ no match.\n", (UChar)uch);
// no match (no exception) -return the 'underlying' break
return n;
}
} else {
return n; // internal error and/or no forwards trie
}
} else {
//if(debug2) u_printf("rev< /%C/ .. no match..%d\n", (UChar)uch, r); // no best match
return n; // No match - so exit. Not an exception.
}
} while(n != UBRK_DONE);
return n;
}