uint32_t RuleBasedCollator::setVariableTop(const UnicodeString varTop, UErrorCode &status) {
checkOwned();
return ucol_setVariableTop(ucollator, varTop.getBuffer(), varTop.length(), &status);
}
U_CAPI UCollator* U_EXPORT2
ucol_openFromShortString( const char *definition,
UBool forceDefaults,
UParseError *parseError,
UErrorCode *status)
{
UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN_FROM_SHORT_STRING);
UTRACE_DATA1(UTRACE_INFO, "short string = \"%s\"", definition);
if(U_FAILURE(*status)) return 0;
UParseError internalParseError;
if(!parseError) {
parseError = &internalParseError;
}
parseError->line = 0;
parseError->offset = 0;
parseError->preContext[0] = 0;
parseError->postContext[0] = 0;
// first we want to pick stuff out of short string.
// we'll end up with an UCA version, locale and a bunch of
// settings
// analyse the string in order to get everything we need.
const char *string = definition;
CollatorSpec s;
ucol_sit_initCollatorSpecs(&s);
string = ucol_sit_readSpecs(&s, definition, parseError, status);
ucol_sit_calculateWholeLocale(&s);
char buffer[internalBufferSize];
uprv_memset(buffer, 0, internalBufferSize);
uloc_canonicalize(s.locale, buffer, internalBufferSize, status);
UCollator *result = ucol_open(buffer, status);
int32_t i = 0;
for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
if(s.options[i] != UCOL_DEFAULT) {
if(forceDefaults || ucol_getAttribute(result, (UColAttribute)i, status) != s.options[i]) {
ucol_setAttribute(result, (UColAttribute)i, s.options[i], status);
}
if(U_FAILURE(*status)) {
parseError->offset = (int32_t)(string - definition);
ucol_close(result);
return NULL;
}
}
}
if(s.variableTopSet) {
if(s.variableTopString[0]) {
ucol_setVariableTop(result, s.variableTopString, s.variableTopStringLen, status);
} else { // we set by value, using 'B'
ucol_restoreVariableTop(result, s.variableTopValue, status);
}
}
if(U_FAILURE(*status)) { // here it can only be a bogus value
ucol_close(result);
result = NULL;
}
UTRACE_EXIT_PTR_STATUS(result, *status);
return result;
}
uint32_t RuleBasedCollator::setVariableTop(const UChar *varTop, int32_t len, UErrorCode &status) {
checkOwned();
return ucol_setVariableTop(ucollator, varTop, len, &status);
}
/*
* The collator returned by this function is owned by the callee and must be
* closed when this method returns with a U_SUCCESS UErrorCode.
*
* On error, the return value is undefined.
*/
UCollator* CloneCollatorWithOptions(const UCollator* pCollator, int32_t options, UErrorCode* pErr)
{
UColAttributeValue strength = ucol_getStrength(pCollator);
bool isIgnoreCase = (options & CompareOptionsIgnoreCase) == CompareOptionsIgnoreCase;
bool isIgnoreNonSpace = (options & CompareOptionsIgnoreNonSpace) == CompareOptionsIgnoreNonSpace;
bool isIgnoreSymbols = (options & CompareOptionsIgnoreSymbols) == CompareOptionsIgnoreSymbols;
if (isIgnoreCase)
{
strength = UCOL_SECONDARY;
}
if (isIgnoreNonSpace)
{
strength = UCOL_PRIMARY;
}
UCollator* pClonedCollator;
std::vector<UChar> customRules = GetCustomRules(options, strength, isIgnoreSymbols);
if (customRules.empty())
{
pClonedCollator = ucol_safeClone(pCollator, nullptr, nullptr, pErr);
}
else
{
int32_t customRuleLength = customRules.size();
int32_t localeRulesLength;
const UChar* localeRules = ucol_getRules(pCollator, &localeRulesLength);
std::vector<UChar> completeRules(localeRulesLength + customRuleLength + 1, '\0');
for (int i = 0; i < localeRulesLength; i++)
{
completeRules[i] = localeRules[i];
}
for (int i = 0; i < customRuleLength; i++)
{
completeRules[localeRulesLength + i] = customRules[i];
}
pClonedCollator = ucol_openRules(completeRules.data(), completeRules.size(), UCOL_DEFAULT, strength, NULL, pErr);
}
if (isIgnoreSymbols)
{
ucol_setAttribute(pClonedCollator, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, pErr);
// by default, ICU alternate shifted handling only ignores punctuation, but
// IgnoreSymbols needs symbols and currency as well, so change the "variable top"
// to include all symbols and currency
#if HAVE_SET_MAX_VARIABLE
ucol_setMaxVariable(pClonedCollator, UCOL_REORDER_CODE_CURRENCY, pErr);
#else
// 0xfdfc is the last currency character before the first digit character
// in http://source.icu-project.org/repos/icu/icu/tags/release-52-1/source/data/unidata/FractionalUCA.txt
const UChar ignoreSymbolsVariableTop[] = { 0xfdfc };
ucol_setVariableTop(pClonedCollator, ignoreSymbolsVariableTop, 1, pErr);
#endif
}
ucol_setAttribute(pClonedCollator, UCOL_STRENGTH, strength, pErr);
// casing differs at the tertiary level.
// if strength is less than tertiary, but we are not ignoring case, then we need to flip CASE_LEVEL On
if (strength < UCOL_TERTIARY && !isIgnoreCase)
{
ucol_setAttribute(pClonedCollator, UCOL_CASE_LEVEL, UCOL_ON, pErr);
}
return pClonedCollator;
}
