// catch degenerate case if (newSource.length() == 0) {
pieces = static_cast<UnicodeString**>(uprv_malloc(sizeof(UnicodeString*)));
pieces_lengths = static_cast<int32_t*>(uprv_malloc(1 * sizeof(int32_t)));
pieces_length = 1;
current = static_cast<int32_t*>(uprv_malloc(1 * sizeof(int32_t)));
current_length = 1; if (pieces == nullptr || pieces_lengths == nullptr || current == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; goto CleanPartialInitialization;
}
current[0] = 0;
pieces[0] = new UnicodeString[1];
pieces_lengths[0] = 1; if (pieces[0] == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; goto CleanPartialInitialization;
} return;
}
list = new UnicodeString[source.length()]; if (list == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; goto CleanPartialInitialization;
}
// i should initially be the number of code units at the // start of the string
i = U16_LENGTH(source.char32At(0)); // int32_t i = 1; // find the segments // This code iterates through the source string and // extracts segments that end up on a codepoint that // doesn't start any decompositions. (Analysis is done // on the NFD form - see above). for (; i < source.length(); i += U16_LENGTH(cp)) {
cp = source.char32At(i); if (nfcImpl->isCanonSegmentStarter(cp)) {
source.extract(start, i-start, list[list_length++]); // add up to i
start = i;
}
}
source.extract(start, i-start, list[list_length++]); // add last one
// allocate the arrays, and find the strings that are CE to each segment
pieces = static_cast<UnicodeString**>(uprv_malloc(list_length * sizeof(UnicodeString*)));
pieces_length = list_length;
pieces_lengths = static_cast<int32_t*>(uprv_malloc(list_length * sizeof(int32_t)));
current = static_cast<int32_t*>(uprv_malloc(list_length * sizeof(int32_t)));
current_length = list_length; if (pieces == nullptr || pieces_lengths == nullptr || current == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; goto CleanPartialInitialization;
}
for (i = 0; i < current_length; i++) {
current[i] = 0;
} // for each segment, get all the combinations that can produce // it after NFD normalization for (i = 0; i < pieces_length; ++i) { //if (PROGRESS) printf("SEGMENT\n");
pieces[i] = getEquivalents(list[i], pieces_lengths[i], status);
}
delete[] list; return; // Common section to cleanup all local variables and reset object variables.
CleanPartialInitialization: delete[] list;
cleanPieces();
}
/** *Dumbrecursiveimplementationofpermutation. *TODO:optimize *@paramsourcethestringtofindpermutationsfor *@returntheresultsinaset.
*/ void U_EXPORT2 CanonicalIterator::permute(UnicodeString &source, UBool skipZeros, Hashtable *result, UErrorCode &status, int32_t depth) { if(U_FAILURE(status)) { return;
} // To avoid infinity loop caused by permute, we limit the depth of recursive // call to permute and return U_UNSUPPORTED_ERROR. // We know in some unit test we need at least 4. Set to 8 just in case some // unforseen use cases.
constexpr int32_t kPermuteDepthLimit = 8; if (depth > kPermuteDepthLimit) {
status = U_UNSUPPORTED_ERROR; return;
} //if (PROGRESS) printf("Permute: %s\n", UToS(Tr(source)));
int32_t i = 0;
// optimization: // if zero or one character, just return a set with it // we check for length < 2 to keep from counting code points all the time if (source.length() <= 2 && source.countChar32() <= 1) {
UnicodeString *toPut = new UnicodeString(source); /* test for nullptr */ if (toPut == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return;
}
result->put(source, toPut, status); return;
}
// otherwise iterate through the string, and recursively permute all the other characters
UChar32 cp;
Hashtable subpermute(status); if(U_FAILURE(status)) { return;
}
subpermute.setValueDeleter(uprv_deleteUObject);
for (i = 0; i < source.length(); i += U16_LENGTH(cp)) {
cp = source.char32At(i); const UHashElement *ne = nullptr;
int32_t el = UHASH_FIRST;
UnicodeString subPermuteString = source;
// optimization: // if the character is canonical combining class zero, // don't permute it if (skipZeros && i != 0 && u_getCombiningClass(cp) == 0) { //System.out.println("Skipping " + Utility.hex(UTF16.valueOf(source, i))); continue;
}
subpermute.removeAll();
// see what the permutations of the characters before and after this one are //Hashtable *subpermute = permute(source.substring(0,i) + source.substring(i + UTF16.getCharCount(cp)));
permute(subPermuteString.remove(i, U16_LENGTH(cp)), skipZeros, &subpermute, status, depth+1); /* Test for buffer overflows */ if(U_FAILURE(status)) { return;
} // The upper remove is destructive. The question is do we have to make a copy, or we don't care about the contents // of source at this point.
// prefix this character to all of them
ne = subpermute.nextElement(el); while (ne != nullptr) {
UnicodeString* permRes = static_cast<UnicodeString*>(ne->value.pointer);
UnicodeString *chStr = new UnicodeString(cp); //test for nullptr if (chStr == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return;
}
chStr->append(*permRes); //*((UnicodeString *)(ne->value.pointer)); //if (PROGRESS) printf(" Piece: %s\n", UToS(*chStr));
result->put(*chStr, chStr, status);
ne = subpermute.nextElement(el);
}
} //return result;
}
// privates
// we have a segment, in NFD. Find all the strings that are canonically equivalent to it.
UnicodeString* CanonicalIterator::getEquivalents(const UnicodeString &segment, int32_t &result_len, UErrorCode &status) {
Hashtable result(status);
Hashtable permutations(status);
Hashtable basic(status); if (U_FAILURE(status)) { return nullptr;
}
result.setValueDeleter(uprv_deleteUObject);
permutations.setValueDeleter(uprv_deleteUObject);
basic.setValueDeleter(uprv_deleteUObject);
// TODO: check if operator == is semanticaly the same as attempt.equals(segment) if (attempt==segment) { //if (PROGRESS) printf("Adding Permutation: %s\n", UToS(Tr(*possible))); // TODO: use the hashtable just to catch duplicates - store strings directly (somehow).
result.put(possible, new UnicodeString(possible), status); //add(possible);
} else { //if (PROGRESS) printf("-Skipping Permutation: %s\n", UToS(Tr(*possible)));
}
ne2 = permutations.nextElement(el2);
}
ne = basic.nextElement(el);
}
/* Test for buffer overflows */ if(U_FAILURE(status)) { return nullptr;
} // convert into a String[] to clean up storage //String[] finalResult = new String[result.size()];
UnicodeString *finalResult = nullptr;
int32_t resultCount; if((resultCount = result.count()) != 0) {
finalResult = new UnicodeString[resultCount]; if (finalResult == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return nullptr;
}
} else {
status = U_ILLEGAL_ARGUMENT_ERROR; return nullptr;
} //result.toArray(finalResult);
result_len = 0;
el = UHASH_FIRST;
ne = result.nextElement(el); while(ne != nullptr) {
finalResult[result_len++] = *static_cast<UnicodeString*>(ne->value.pointer);
ne = result.nextElement(el);
}
fillinResult->put(toPut, new UnicodeString(toPut), status);
UnicodeSet starts;
// cycle through all the characters
UChar32 cp; for (int32_t i = 0; i < segLen; i += U16_LENGTH(cp)) { // see if any character is at the start of some decomposition
U16_GET(segment, 0, i, segLen, cp); if (!nfcImpl->getCanonStartSet(cp, starts)) { continue;
} // if so, see which decompositions match
UnicodeSetIterator iter(starts); while (iter.next()) {
UChar32 cp2 = iter.getCodepoint();
Hashtable remainder(status);
remainder.setValueDeleter(uprv_deleteUObject); if (extract(&remainder, cp2, segment, segLen, i, status) == nullptr) { if (U_FAILURE(status)) { return nullptr;
} continue;
}
// there were some matches, so add all the possibilities to the set.
UnicodeString prefix(segment, i);
prefix += cp2;
int32_t el = UHASH_FIRST; const UHashElement *ne = remainder.nextElement(el); while (ne != nullptr) {
UnicodeString item = *static_cast<UnicodeString*>(ne->value.pointer);
UnicodeString *toAdd = new UnicodeString(prefix); /* test for nullptr */ if (toAdd == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return nullptr;
}
*toAdd += item;
fillinResult->put(*toAdd, toAdd, status);
ne = remainder.nextElement(el);
} // ICU-22642 Guards against strings that have so many permutations // that they would otherwise hang the function.
constexpr int32_t kResultLimit = 4096; if (fillinResult->count() > kResultLimit) {
status = U_UNSUPPORTED_ERROR; return nullptr;
}
}
}
/* Test for buffer overflows */ if(U_FAILURE(status)) { return nullptr;
} return fillinResult;
}
UnicodeString temp(comp);
int32_t inputLen=temp.length();
UnicodeString decompString;
nfd->normalize(temp, decompString, status); if (U_FAILURE(status)) { return nullptr;
} if (decompString.isBogus()) {
status = U_MEMORY_ALLOCATION_ERROR; return nullptr;
} const char16_t *decomp=decompString.getBuffer();
int32_t decompLen=decompString.length();
// See if it matches the start of segment (at segmentPos)
UBool ok = false;
UChar32 cp;
int32_t decompPos = 0;
UChar32 decompCp;
U16_NEXT(decomp, decompPos, decompLen, decompCp);
int32_t i = segmentPos; while(i < segLen) {
U16_NEXT(segment, i, segLen, cp);
if (cp == decompCp) { // if equal, eat another cp from decomp
if (decompPos == decompLen) { // done, have all decomp characters!
temp.append(segment+i, segLen-i);
ok = true; break;
}
U16_NEXT(decomp, decompPos, decompLen, decompCp);
} else { //if (PROGRESS) printf(" buffer: %s\n", UToS(Tr(UnicodeString(cp))));
// brute force approach
temp.append(cp);
/* TODO: optimize // since we know that the classes are monotonically increasing, after zero // e.g. 0 5 7 9 0 3 // we can do an optimization // there are only a few cases that work: zero, less, same, greater // if both classes are the same, we fail // if the decomp class < the segment class, we fail
segClass=getClass(cp); if(decompClass<=segClass)returnnull;
*/
}
} if (!ok) return nullptr; // we failed, characters left over
//if (PROGRESS) printf("Matches\n");
if (inputLen == temp.length()) {
fillinResult->put(UnicodeString(), new UnicodeString(), status); return fillinResult; // succeed, but no remainder
}
// brute force approach // check to make sure result is canonically equivalent
UnicodeString trial;
nfd->normalize(temp, trial, status); if(U_FAILURE(status) || trial.compare(segment+segmentPos, segLen - segmentPos) != 0) { return nullptr;
}
¤ Diese beiden folgenden Angebotsgruppen bietet das Unternehmen0.21Angebot
(Wie Sie bei der Firma Beratungs- und Dienstleistungen beauftragen können 2026-08-26)
¤
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.