void Collator::createCollator() const
{
ASSERT(!m_collator);
UErrorCode status = U_ZERO_ERROR;
{
Locker<Mutex> lock(cachedCollatorMutex());
if (cachedCollator) {
const char* cachedCollatorLocale = ucol_getLocaleByType(cachedCollator, ULOC_REQUESTED_LOCALE, &status);
ASSERT(U_SUCCESS(status));
ASSERT(cachedCollatorLocale);
UColAttributeValue cachedCollatorLowerFirst = ucol_getAttribute(cachedCollator, UCOL_CASE_FIRST, &status);
ASSERT(U_SUCCESS(status));
// FIXME: default locale is never matched, because ucol_getLocaleByType returns the actual one used, not 0.
if (m_locale && 0 == strcmp(cachedCollatorLocale, m_locale)
&& ((UCOL_LOWER_FIRST == cachedCollatorLowerFirst && m_lowerFirst) || (UCOL_UPPER_FIRST == cachedCollatorLowerFirst && !m_lowerFirst))) {
m_collator = cachedCollator;
cachedCollator = 0;
return;
}
}
}
m_collator = ucol_open(m_locale, &status);
if (U_FAILURE(status)) {
status = U_ZERO_ERROR;
m_collator = ucol_open("", &status); // Fallback to Unicode Collation Algorithm.
}
ASSERT(U_SUCCESS(status));
ucol_setAttribute(m_collator, UCOL_CASE_FIRST, m_lowerFirst ? UCOL_LOWER_FIRST : UCOL_UPPER_FIRST, &status);
ASSERT(U_SUCCESS(status));
}
void Target::setTargetString(const UnicodeString *target)
{
if (charBreakIterator != NULL) {
ubrk_close(charBreakIterator);
ucol_closeElements(elements);
}
targetString = target;
if (targetString != NULL) {
UErrorCode status = U_ZERO_ERROR;
targetBuffer = targetString->getBuffer();
targetLength = targetString->length();
elements = ucol_openElements(coll, target->getBuffer(), target->length(), &status);
ucol_forceHanImplicit(elements, &status);
charBreakIterator = ubrk_open(UBRK_CHARACTER, ucol_getLocaleByType(coll, ULOC_VALID_LOCALE, &status),
targetBuffer, targetLength, &status);
} else {
targetBuffer = NULL;
targetLength = 0;
}
}
const Locale RuleBasedCollator::getLocale(ULocDataLocaleType type, UErrorCode &status) const {
const char *result = ucol_getLocaleByType(ucollator, type, &status);
if(result == NULL) {
Locale res("");
res.setToBogus();
return res;
} else {
return Locale(result);
}
}
// Collator.actual_locale {{{
static PyObject *
icu_Collator_actual_locale(icu_Collator *self, void *closure) {
const char *loc = NULL;
UErrorCode status = U_ZERO_ERROR;
loc = ucol_getLocaleByType(self->collator, ULOC_ACTUAL_LOCALE, &status);
if (loc == NULL || U_FAILURE(status)) {
PyErr_SetString(PyExc_Exception, "Failed to get actual locale"); return NULL;
}
return Py_BuildValue("s", loc);
}
static String HHVM_METHOD(Collator, getLocale, int64_t type) {
FETCH_COL(data, this_, "");
data->clearError();
UErrorCode error = U_ZERO_ERROR;
auto loc = ucol_getLocaleByType(data->collator(), (ULocDataLocaleType)type,
&error);
if (U_FAILURE(error)) {
data->setError(error, "Error getting locale by type");
}
return String(loc, CopyString);
}
// Collator.display_name {{{
static PyObject *
icu_Collator_display_name(icu_Collator *self, void *closure) {
const char *loc = NULL;
UErrorCode status = U_ZERO_ERROR;
UChar dname[400];
int32_t sz = 0;
loc = ucol_getLocaleByType(self->collator, ULOC_ACTUAL_LOCALE, &status);
if (loc == NULL) {
PyErr_SetString(PyExc_Exception, "Failed to get actual locale"); return NULL;
}
sz = ucol_getDisplayName(loc, "en", dname, sizeof(dname), &status);
if (U_FAILURE(status)) {PyErr_SetString(PyExc_ValueError, u_errorName(status)); return NULL; }
return icu_to_python(dname, sz);
}
String c_Collator::t_getlocale(int64_t type /* = 0 */) {
if (!m_ucoll) {
raise_error("getlocale called on uninitialized Collator object");
return "";
}
m_errcode.clearError();
UErrorCode error = U_ZERO_ERROR;
String ret(
(char*)ucol_getLocaleByType(m_ucoll, (ULocDataLocaleType)type,
&error),
CopyString);
if (U_FAILURE(error)) {
m_errcode.setError(error, "Error getting locale by type");
return "";
}
return ret;
}
String c_Collator::t_getlocale(int64_t type /* = 0 */) {
if (!m_ucoll) {
raise_warning("getlocale called on uninitialized Collator object");
return "";
}
m_errcode.clear();
String ret(
(char*)ucol_getLocaleByType(m_ucoll, (ULocDataLocaleType)type,
&(m_errcode.code)),
CopyString);
if (U_FAILURE(m_errcode.code)) {
m_errcode.custom_error_message = "Error getting locale by type";
s_intl_error->m_error.code = m_errcode.code;
s_intl_error->m_error.custom_error_message =
m_errcode.custom_error_message;
return "";
}
return ret;
}
String c_Collator::t_getlocale(int64 type /* = 0 */) {
INSTANCE_METHOD_INJECTION_BUILTIN(Collator, Collator::getlocale);
if (!m_ucoll) {
raise_warning("getlocale called on uninitialized Collator object");
return "";
}
m_errcode.clear();
String ret(
(char*)ucol_getLocaleByType(m_ucoll, (ULocDataLocaleType)type,
&(m_errcode.code)),
AttachLiteral);
if (U_FAILURE(m_errcode.code)) {
m_errcode.custom_error_message = "Error getting locale by type";
s_intl_error->m_error.code = m_errcode.code;
s_intl_error->m_error.custom_error_message =
m_errcode.custom_error_message;
return "";
}
return ret;
}
// Collator.display_name {{{
static PyObject *
icu_Collator_display_name(icu_Collator *self, void *closure) {
const char *loc = NULL;
UErrorCode status = U_ZERO_ERROR;
UChar dname[400];
char buf[100];
loc = ucol_getLocaleByType(self->collator, ULOC_ACTUAL_LOCALE, &status);
if (loc == NULL || U_FAILURE(status)) {
PyErr_SetString(PyExc_Exception, "Failed to get actual locale"); return NULL;
}
ucol_getDisplayName(loc, "en", dname, 100, &status);
if (U_FAILURE(status)) return PyErr_NoMemory();
u_strToUTF8(buf, 100, NULL, dname, -1, &status);
if (U_FAILURE(status)) {
PyErr_SetString(PyExc_Exception, "Failed to convert dname to UTF-8"); return NULL;
}
return Py_BuildValue("s", buf);
}
static UBool simpleSearch(UCollator *coll, const UnicodeString &target, int32_t offset, const UnicodeString &pattern, int32_t &matchStart, int32_t &matchEnd)
{
UErrorCode status = U_ZERO_ERROR;
OrderList targetOrders(coll, target, offset);
OrderList patternOrders(coll, pattern);
int32_t targetSize = targetOrders.size() - 1;
int32_t patternSize = patternOrders.size() - 1;
UBreakIterator *charBreakIterator = ubrk_open(UBRK_CHARACTER, ucol_getLocaleByType(coll, ULOC_VALID_LOCALE, &status),
target.getBuffer(), target.length(), &status);
if (patternSize == 0) {
// Searching for an empty pattern always fails
matchStart = matchEnd = -1;
ubrk_close(charBreakIterator);
return FALSE;
}
matchStart = matchEnd = -1;
for(int32_t i = 0; i < targetSize; i += 1) {
if (targetOrders.matchesAt(i, patternOrders)) {
int32_t start = targetOrders.getLowOffset(i);
int32_t maxLimit = targetOrders.getLowOffset(i + patternSize);
int32_t minLimit = targetOrders.getLowOffset(i + patternSize - 1);
// if the low and high offsets of the first CE in
// the match are the same, it means that the match
// starts in the middle of an expansion - all but
// the first CE of the expansion will have the offset
// of the following character.
if (start == targetOrders.getHighOffset(i)) {
continue;
}
// Make sure match starts on a grapheme boundary
if (! ubrk_isBoundary(charBreakIterator, start)) {
continue;
}
// If the low and high offsets of the CE after the match
// are the same, it means that the match ends in the middle
// of an expansion sequence.
if (maxLimit == targetOrders.getHighOffset(i + patternSize) &&
targetOrders.getOrder(i + patternSize) != UCOL_NULLORDER) {
continue;
}
int32_t mend = maxLimit;
// Find the first grapheme break after the character index
// of the last CE in the match. If it's after character index
// that's after the last CE in the match, use that index
// as the end of the match.
if (minLimit < maxLimit) {
// When the last CE's low index is same with its high index, the CE is likely
// a part of expansion. In this case, the index is located just after the
// character corresponding to the CEs compared above. If the index is right
// at the break boundary, move the position to the next boundary will result
// incorrect match length when there are ignorable characters exist between
// the position and the next character produces CE(s). See ticket#8482.
if (minLimit == targetOrders.getHighOffset(i + patternSize - 1) && ubrk_isBoundary(charBreakIterator, minLimit)) {
mend = minLimit;
} else {
int32_t nba = ubrk_following(charBreakIterator, minLimit);
if (nba >= targetOrders.getHighOffset(i + patternSize - 1)) {
mend = nba;
}
}
}
if (mend > maxLimit) {
continue;
}
if (! ubrk_isBoundary(charBreakIterator, mend)) {
continue;
}
matchStart = start;
matchEnd = mend;
ubrk_close(charBreakIterator);
return TRUE;
}
}
ubrk_close(charBreakIterator);
return FALSE;
}
void CollationServiceTest::TestRegister()
{
#if !UCONFIG_NO_SERVICE
// register a singleton
const Locale& FR = Locale::getFrance();
const Locale& US = Locale::getUS();
const Locale US_FOO("en", "US", "FOO");
UErrorCode status = U_ZERO_ERROR;
Collator* frcol = Collator::createInstance(FR, status);
Collator* uscol = Collator::createInstance(US, status);
if(U_FAILURE(status)) {
errcheckln(status, "Failed to create collators with %s", u_errorName(status));
delete frcol;
delete uscol;
return;
}
{ // try override en_US collator
URegistryKey key = Collator::registerInstance(frcol, US, status);
Collator* ncol = Collator::createInstance(US_FOO, status);
if (*frcol != *ncol) {
errln("register of french collator for en_US failed on request for en_US_FOO");
}
// ensure original collator's params not touched
Locale loc = frcol->getLocale(ULOC_REQUESTED_LOCALE, status);
if (loc != FR) {
errln(UnicodeString("fr collator's requested locale changed to ") + loc.getName());
}
loc = frcol->getLocale(ULOC_VALID_LOCALE, status);
if (loc != FR) {
errln(UnicodeString("fr collator's valid locale changed to ") + loc.getName());
}
loc = ncol->getLocale(ULOC_REQUESTED_LOCALE, status);
if (loc != US_FOO) {
errln(UnicodeString("requested locale for en_US_FOO is not en_US_FOO but ") + loc.getName());
}
loc = ncol->getLocale(ULOC_VALID_LOCALE, status);
if (loc != US) {
errln(UnicodeString("valid locale for en_US_FOO is not en_US but ") + loc.getName());
}
loc = ncol->getLocale(ULOC_ACTUAL_LOCALE, status);
if (loc != US) {
errln(UnicodeString("actual locale for en_US_FOO is not en_US but ") + loc.getName());
}
delete ncol; ncol = NULL;
if (!Collator::unregister(key, status)) {
errln("failed to unregister french collator");
}
// !!! frcol pointer is now invalid !!!
ncol = Collator::createInstance(US, status);
if (*uscol != *ncol) {
errln("collator after unregister does not match original");
}
delete ncol; ncol = NULL;
}
// recreate frcol
frcol = Collator::createInstance(FR, status);
LocalUCollatorPointer frFR(ucol_open("fr_FR", &status));
{ // try create collator for new locale
Locale fu_FU_FOO("fu", "FU", "FOO");
Locale fu_FU("fu", "FU", "");
Collator* fucol = Collator::createInstance(fu_FU, status);
URegistryKey key = Collator::registerInstance(frcol, fu_FU, status);
Collator* ncol = Collator::createInstance(fu_FU_FOO, status);
if (*frcol != *ncol) {
errln("register of fr collator for fu_FU failed");
}
UnicodeString locName = fu_FU.getName();
StringEnumeration* localeEnum = Collator::getAvailableLocales();
UBool found = FALSE;
const UnicodeString* locStr, *ls2;
for (locStr = localeEnum->snext(status);
!found && locStr != NULL;
locStr = localeEnum->snext(status)) {
//
if (locName == *locStr) {
found = TRUE;
}
}
StringEnumeration *le2 = NULL;
localeEnum->reset(status);
int32_t i, count;
count = localeEnum->count(status);
for(i = 0; i < count; ++i) {
if(i == count / 2) {
le2 = localeEnum->clone();
if(le2 == NULL || count != le2->count(status)) {
errln("ServiceEnumeration.clone() failed");
break;
}
}
if(i >= count / 2) {
locStr = localeEnum->snext(status);
ls2 = le2->snext(status);
if(*locStr != *ls2) {
errln("ServiceEnumeration.clone() failed for item %d", i);
}
} else {
localeEnum->snext(status);
}
}
delete localeEnum;
delete le2;
if (!found) {
errln("new locale fu_FU not reported as supported locale");
}
UnicodeString displayName;
Collator::getDisplayName(fu_FU, displayName);
/* The locale display pattern for the locale ja, ko, and zh are different. */
const UChar zh_fuFU_Array[] = { 0x0066, 0x0075, 0xff08, 0x0046, 0x0055, 0xff09, 0 };
const UnicodeString zh_fuFU(zh_fuFU_Array);
const Locale& defaultLocale = Locale::getDefault();
if (displayName != "fu (FU)" &&
((defaultLocale == Locale::getKorean() && defaultLocale == Locale::getJapanese()) && displayName == "fu(FU)") &&
((defaultLocale == Locale::getChinese()) && displayName != zh_fuFU)) {
errln(UnicodeString("found ") + displayName + " for fu_FU");
}
Collator::getDisplayName(fu_FU, fu_FU, displayName);
if (displayName != "fu (FU)" &&
((defaultLocale == Locale::getKorean() && defaultLocale == Locale::getJapanese()) && displayName == "fu(FU)") &&
((defaultLocale == Locale::getChinese()) && displayName != zh_fuFU)) {
errln(UnicodeString("found ") + displayName + " for fu_FU");
}
// test ucol_open
LocalUCollatorPointer fufu(ucol_open("fu_FU_FOO", &status));
if (fufu.isNull()) {
errln("could not open fu_FU_FOO with ucol_open");
} else {
if (!ucol_equals(fufu.getAlias(), frFR.getAlias())) {
errln("collator fufu != collator frFR");
}
}
if (!Collator::unregister(key, status)) {
errln("failed to unregister french collator");
}
// !!! note frcoll invalid again, but we're no longer using it
// other collators should still work ok
Locale nloc = ncol->getLocale(ULOC_VALID_LOCALE, status);
if (nloc != fu_FU) {
errln(UnicodeString("asked for nloc valid locale after close and got") + nloc.getName());
}
delete ncol; ncol = NULL;
if (fufu.isValid()) {
const char* nlocstr = ucol_getLocaleByType(fufu.getAlias(), ULOC_VALID_LOCALE, &status);
if (uprv_strcmp(nlocstr, "fu_FU") != 0) {
errln(UnicodeString("asked for uloc valid locale after close and got ") + nlocstr);
}
}
ncol = Collator::createInstance(fu_FU, status);
if (*fucol != *ncol) {
errln("collator after unregister does not match original fu_FU");
}
delete uscol; uscol = NULL;
delete ncol; ncol = NULL;
delete fucol; fucol = NULL;
}
#endif
}
U_CAPI int32_t U_EXPORT2
ucol_getShortDefinitionString(const UCollator *coll,
const char *locale,
char *dst,
int32_t capacity,
UErrorCode *status)
{
if(U_FAILURE(*status)) return 0;
if(coll->delegate != NULL) {
return ((icu::Collator*)coll->delegate)->internalGetShortDefinitionString(locale,dst,capacity,*status);
}
char buffer[internalBufferSize];
uprv_memset(buffer, 0, internalBufferSize*sizeof(char));
int32_t resultSize = 0;
char tempbuff[internalBufferSize];
char locBuff[internalBufferSize];
uprv_memset(buffer, 0, internalBufferSize*sizeof(char));
int32_t elementSize = 0;
UBool isAvailable = 0;
CollatorSpec s;
ucol_sit_initCollatorSpecs(&s);
if(!locale) {
locale = ucol_getLocaleByType(coll, ULOC_VALID_LOCALE, status);
}
elementSize = ucol_getFunctionalEquivalent(locBuff, internalBufferSize, "collation", locale, &isAvailable, status);
if(elementSize) {
// we should probably canonicalize here...
elementSize = uloc_getLanguage(locBuff, tempbuff, internalBufferSize, status);
appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, languageArg);
elementSize = uloc_getCountry(locBuff, tempbuff, internalBufferSize, status);
appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, regionArg);
elementSize = uloc_getScript(locBuff, tempbuff, internalBufferSize, status);
appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, scriptArg);
elementSize = uloc_getVariant(locBuff, tempbuff, internalBufferSize, status);
appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, variantArg);
elementSize = uloc_getKeywordValue(locBuff, "collation", tempbuff, internalBufferSize, status);
appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, keywordArg);
}
int32_t i = 0;
UColAttributeValue attribute = UCOL_DEFAULT;
for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
if(options[i].action == _processCollatorOption) {
attribute = ucol_getAttributeOrDefault(coll, (UColAttribute)options[i].attr, status);
if(attribute != UCOL_DEFAULT) {
char letter = ucol_sit_attributeValueToLetter(attribute, status);
appendShortStringElement(&letter, 1,
buffer, &resultSize, /*capacity*/internalBufferSize, options[i].optionStart);
}
}
}
if(coll->variableTopValueisDefault == FALSE) {
//s.variableTopValue = ucol_getVariableTop(coll, status);
elementSize = T_CString_integerToString(tempbuff, coll->variableTopValue, 16);
appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, variableTopValArg);
}
UParseError parseError;
return ucol_normalizeShortDefinitionString(buffer, dst, capacity, &parseError, status);
}