DictionaryMatcher *
ICULanguageBreakFactory::loadDictionaryMatcherFor(UScriptCode script, int32_t /* brkType */) {
UErrorCode status = U_ZERO_ERROR;
// open root from brkitr tree.
UResourceBundle *b = ures_open(U_ICUDATA_BRKITR, "", &status);
b = ures_getByKeyWithFallback(b, "dictionaries", b, &status);
int32_t dictnlength = 0;
const UChar *dictfname =
ures_getStringByKeyWithFallback(b, uscript_getShortName(script), &dictnlength, &status);
if (U_FAILURE(status)) {
ures_close(b);
return NULL;
}
CharString dictnbuf;
CharString ext;
const UChar *extStart = u_memrchr(dictfname, 0x002e, dictnlength); // last dot
if (extStart != NULL) {
int32_t len = (int32_t)(extStart - dictfname);
ext.appendInvariantChars(UnicodeString(FALSE, extStart + 1, dictnlength - len - 1), status);
dictnlength = len;
}
dictnbuf.appendInvariantChars(UnicodeString(FALSE, dictfname, dictnlength), status);
ures_close(b);
UDataMemory *file = udata_open(U_ICUDATA_BRKITR, ext.data(), dictnbuf.data(), &status);
if (U_SUCCESS(status)) {
// build trie
const uint8_t *data = (const uint8_t *)udata_getMemory(file);
const int32_t *indexes = (const int32_t *)data;
const int32_t offset = indexes[DictionaryData::IX_STRING_TRIE_OFFSET];
const int32_t trieType = indexes[DictionaryData::IX_TRIE_TYPE] & DictionaryData::TRIE_TYPE_MASK;
DictionaryMatcher *m = NULL;
if (trieType == DictionaryData::TRIE_TYPE_BYTES) {
const int32_t transform = indexes[DictionaryData::IX_TRANSFORM];
const char *characters = (const char *)(data + offset);
m = new BytesDictionaryMatcher(characters, transform, file);
}
else if (trieType == DictionaryData::TRIE_TYPE_UCHARS) {
const UChar *characters = (const UChar *)(data + offset);
m = new UCharsDictionaryMatcher(characters, file);
}
if (m == NULL) {
// no matcher exists to take ownership - either we are an invalid
// type or memory allocation failed
udata_close(file);
}
return m;
} else if (dictfname != NULL) {
// we don't have a dictionary matcher.
// returning NULL here will cause us to fail to find a dictionary break engine, as expected
status = U_ZERO_ERROR;
return NULL;
}
return NULL;
}
PluralMapBase::Category
PluralMapBase::toCategory(const UnicodeString &pluralForm) {
CharString cCategory;
UErrorCode status = U_ZERO_ERROR;
cCategory.appendInvariantChars(pluralForm, status);
return U_FAILURE(status) ? NONE : toCategory(cCategory.data());
}
const Hashtable*
LocaleUtility::getAvailableLocaleNames(const UnicodeString& bundleID)
{
// LocaleUtility_cache is a hash-of-hashes. The top-level keys
// are path strings ('bundleID') passed to
// ures_openAvailableLocales. The top-level values are
// second-level hashes. The second-level keys are result strings
// from ures_openAvailableLocales. The second-level values are
// garbage ((void*)1 or other random pointer).
UErrorCode status = U_ZERO_ERROR;
umtx_initOnce(LocaleUtilityInitOnce, locale_utility_init, status);
Hashtable *cache = LocaleUtility_cache;
if (cache == NULL) {
// Catastrophic failure.
return NULL;
}
Hashtable* htp;
umtx_lock(NULL);
htp = (Hashtable*) cache->get(bundleID);
umtx_unlock(NULL);
if (htp == NULL) {
htp = new Hashtable(status);
if (htp && U_SUCCESS(status)) {
CharString cbundleID;
cbundleID.appendInvariantChars(bundleID, status);
const char* path = cbundleID.isEmpty() ? NULL : cbundleID.data();
UEnumeration *uenum = ures_openAvailableLocales(path, &status);
for (;;) {
const UChar* id = uenum_unext(uenum, NULL, &status);
if (id == NULL) {
break;
}
htp->put(UnicodeString(id), (void*)htp, status);
}
uenum_close(uenum);
if (U_FAILURE(status)) {
delete htp;
return NULL;
}
umtx_lock(NULL);
Hashtable *t = static_cast<Hashtable *>(cache->get(bundleID));
if (t != NULL) {
// Another thread raced through this code, creating the cache entry first.
// Discard ours and return theirs.
umtx_unlock(NULL);
delete htp;
htp = t;
} else {
cache->put(bundleID, (void*)htp, status);
umtx_unlock(NULL);
}
}
}
return htp;
}
void
TimeUnitFormat::checkConsistency(UTimeUnitFormatStyle style, const char* key, UErrorCode& err) {
if (U_FAILURE(err)) {
return;
}
// there should be patterns for each plural rule in each time unit.
// For each time unit,
// for each plural rule, following is unit pattern fall-back rule:
// ( for example: "one" hour )
// look for its unit pattern in its locale tree.
// if pattern is not found in its own locale, such as de_DE,
// look for the pattern in its parent, such as de,
// keep looking till found or till root.
// if the pattern is not found in root either,
// fallback to plural count "other",
// look for the pattern of "other" in the locale tree:
// "de_DE" to "de" to "root".
// If not found, fall back to value of
// static variable DEFAULT_PATTERN_FOR_xxx, such as "{0} h".
//
// Following is consistency check to create pattern for each
// plural rule in each time unit using above fall-back rule.
//
StringEnumeration* keywords = getPluralRules().getKeywords(err);
if (U_SUCCESS(err)) {
const UnicodeString* pluralCount;
while ((pluralCount = keywords->snext(err)) != NULL) {
if ( U_SUCCESS(err) ) {
for (int32_t i = 0; i < TimeUnit::UTIMEUNIT_FIELD_COUNT; ++i) {
// for each time unit,
// get all the patterns for each plural rule in this locale.
Hashtable* countToPatterns = fTimeUnitToCountToPatterns[i];
if ( countToPatterns == NULL ) {
countToPatterns = initHash(err);
if (U_FAILURE(err)) {
delete countToPatterns;
return;
}
fTimeUnitToCountToPatterns[i] = countToPatterns;
}
MessageFormat** formatters = (MessageFormat**)countToPatterns->get(*pluralCount);
if( formatters == NULL || formatters[style] == NULL ) {
// look through parents
const char* localeName = getLocaleID(err);
CharString pluralCountChars;
pluralCountChars.appendInvariantChars(*pluralCount, err);
searchInLocaleChain(style, key, localeName,
(TimeUnit::UTimeUnitFields)i,
*pluralCount, pluralCountChars.data(),
countToPatterns, err);
}
}
}
}
}
delete keywords;
}
static void strToLocale(
const UnicodeString &str, void *localePtr, UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
CharString localeStr;
localeStr.appendInvariantChars(str, status);
*static_cast<Locale *>(localePtr) = Locale(localeStr.data());
}
static void strToDouble(
const UnicodeString &str, void *doublePtr, UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
CharString buffer;
buffer.appendInvariantChars(str, status);
if (U_FAILURE(status)) {
return;
}
*static_cast<double *>(doublePtr) = atof(buffer.data());
}
// getCDFUnitFallback returns a pointer to the prefix-suffix pair for a given
// variant and log10 value within table. If the given variant doesn't exist, it
// falls back to the OTHER variant. Therefore, this method will always return
// some non-NULL value.
static const CDFUnit* getCDFUnitFallback(const UHashtable* table, const UnicodeString& variant, int32_t log10Value) {
CharString cvariant;
UErrorCode status = U_ZERO_ERROR;
const CDFUnit *cdfUnit = NULL;
cvariant.appendInvariantChars(variant, status);
if (!U_FAILURE(status)) {
cdfUnit = (const CDFUnit*) uhash_get(table, cvariant.data());
}
if (cdfUnit == NULL) {
cdfUnit = (const CDFUnit*) uhash_get(table, gOther);
}
return &cdfUnit[log10Value];
}
void
CurrencyAffixInfo::set(
const char *locale,
const PluralRules *rules,
const UChar *currency,
UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
fIsDefault = FALSE;
if (currency == NULL) {
fSymbol.setTo(gDefaultSymbols, 1);
fISO.setTo(gDefaultSymbols, 2);
fLong.remove();
fLong.append(gDefaultSymbols, 3);
fIsDefault = TRUE;
return;
}
int32_t len;
UBool unusedIsChoice;
const UChar *symbol = ucurr_getName(
currency, locale, UCURR_SYMBOL_NAME, &unusedIsChoice,
&len, &status);
if (U_FAILURE(status)) {
return;
}
fSymbol.setTo(symbol, len);
fISO.setTo(currency, u_strlen(currency));
fLong.remove();
StringEnumeration* keywords = rules->getKeywords(status);
if (U_FAILURE(status)) {
return;
}
const UnicodeString* pluralCount;
while ((pluralCount = keywords->snext(status)) != NULL) {
CharString pCount;
pCount.appendInvariantChars(*pluralCount, status);
const UChar *pluralName = ucurr_getPluralName(
currency, locale, &unusedIsChoice, pCount.data(),
&len, &status);
fLong.setVariant(pCount.data(), UnicodeString(pluralName, len), status);
}
delete keywords;
}
static int32_t toEnum(
const Numberformattesttuple_EnumConversion *table,
int32_t tableLength,
const UnicodeString &str,
UErrorCode &status) {
if (U_FAILURE(status)) {
return 0;
}
CharString cstr;
cstr.appendInvariantChars(str, status);
if (U_FAILURE(status)) {
return 0;
}
for (int32_t i = 0; i < tableLength; ++i) {
if (uprv_strcmp(cstr.data(), table[i].str) == 0) {
return table[i].value;
}
}
status = U_ILLEGAL_ARGUMENT_ERROR;
return 0;
}
ENumberFormatTestTupleField
NumberFormatTestTuple::getFieldByName(
const UnicodeString &name) {
CharString buffer;
UErrorCode status = U_ZERO_ERROR;
buffer.appendInvariantChars(name, status);
if (U_FAILURE(status)) {
return kNumberFormatTestTupleFieldCount;
}
int32_t result = -1;
for (int32_t i = 0; i < UPRV_LENGTHOF(gFieldData); ++i) {
if (uprv_strcmp(gFieldData[i].name, buffer.data()) == 0) {
result = i;
break;
}
}
if (result == -1) {
return kNumberFormatTestTupleFieldCount;
}
return (ENumberFormatTestTupleField) result;
}
// Helpers.
static int32_t parseSetNum(const UnicodeString &setNumStr, UErrorCode &errorCode) {
CharString cs;
cs.appendInvariantChars(setNumStr, errorCode);
return parseSetNum(cs.data(), errorCode);
}
NumberFormat*
NumberFormat::makeInstance(const Locale& desiredLocale,
UNumberFormatStyle style,
UBool mustBeDecimalFormat,
UErrorCode& status) {
if (U_FAILURE(status)) return NULL;
if (style < 0 || style >= UNUM_FORMAT_STYLE_COUNT) {
status = U_ILLEGAL_ARGUMENT_ERROR;
return NULL;
}
// Some styles are not supported. This is a result of merging
// the @draft ICU 4.2 NumberFormat::EStyles into the long-existing UNumberFormatStyle.
// Ticket #8503 is for reviewing/fixing/merging the two relevant implementations:
// this one and unum_open().
// The UNUM_PATTERN_ styles are not supported here
// because this method does not take a pattern string.
if (!isStyleSupported(style)) {
status = U_UNSUPPORTED_ERROR;
return NULL;
}
#if U_PLATFORM_USES_ONLY_WIN32_API
if (!mustBeDecimalFormat) {
char buffer[8];
int32_t count = desiredLocale.getKeywordValue("compat", buffer, sizeof(buffer), status);
// if the locale has "@compat=host", create a host-specific NumberFormat
if (U_SUCCESS(status) && count > 0 && uprv_strcmp(buffer, "host") == 0) {
Win32NumberFormat *f = NULL;
UBool curr = TRUE;
switch (style) {
case UNUM_DECIMAL:
curr = FALSE;
// fall-through
case UNUM_CURRENCY:
case UNUM_CURRENCY_ISO: // do not support plural formatting here
case UNUM_CURRENCY_PLURAL:
f = new Win32NumberFormat(desiredLocale, curr, status);
if (U_SUCCESS(status)) {
return f;
}
delete f;
break;
default:
break;
}
}
}
#endif
// Use numbering system cache hashtable
umtx_initOnce(gNSCacheInitOnce, &nscacheInit);
// Get cached numbering system
LocalPointer<NumberingSystem> ownedNs;
NumberingSystem *ns = NULL;
if (NumberingSystem_cache != NULL) {
// TODO: Bad hash key usage, see ticket #8504.
int32_t hashKey = desiredLocale.hashCode();
Mutex lock(&nscacheMutex);
ns = (NumberingSystem *)uhash_iget(NumberingSystem_cache, hashKey);
if (ns == NULL) {
ns = NumberingSystem::createInstance(desiredLocale,status);
uhash_iput(NumberingSystem_cache, hashKey, (void*)ns, &status);
}
} else {
ownedNs.adoptInstead(NumberingSystem::createInstance(desiredLocale,status));
ns = ownedNs.getAlias();
}
// check results of getting a numbering system
if (U_FAILURE(status)) {
return NULL;
}
if (mustBeDecimalFormat && ns->isAlgorithmic()) {
status = U_UNSUPPORTED_ERROR;
return NULL;
}
LocalPointer<DecimalFormatSymbols> symbolsToAdopt;
UnicodeString pattern;
LocalUResourceBundlePointer ownedResource(ures_open(NULL, desiredLocale.getName(), &status));
if (U_FAILURE(status)) {
// We don't appear to have resource data available -- use the last-resort data
status = U_USING_FALLBACK_WARNING;
// When the data is unavailable, and locale isn't passed in, last resort data is used.
symbolsToAdopt.adoptInstead(new DecimalFormatSymbols(status));
if (symbolsToAdopt.isNull()) {
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
// Creates a DecimalFormat instance with the last resort number patterns.
pattern.setTo(TRUE, gLastResortNumberPatterns[style], -1);
}
else {
// Loads the decimal symbols of the desired locale.
symbolsToAdopt.adoptInstead(new DecimalFormatSymbols(desiredLocale, status));
if (symbolsToAdopt.isNull()) {
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
UResourceBundle *resource = ownedResource.orphan();
UResourceBundle *numElements = ures_getByKeyWithFallback(resource, gNumberElements, NULL, &status);
resource = ures_getByKeyWithFallback(numElements, ns->getName(), resource, &status);
resource = ures_getByKeyWithFallback(resource, gPatterns, resource, &status);
ownedResource.adoptInstead(resource);
int32_t patLen = 0;
const UChar *patResStr = ures_getStringByKeyWithFallback(resource, gFormatKeys[style], &patLen, &status);
// Didn't find a pattern specific to the numbering system, so fall back to "latn"
if ( status == U_MISSING_RESOURCE_ERROR && uprv_strcmp(gLatn,ns->getName())) {
status = U_ZERO_ERROR;
resource = ures_getByKeyWithFallback(numElements, gLatn, resource, &status);
resource = ures_getByKeyWithFallback(resource, gPatterns, resource, &status);
patResStr = ures_getStringByKeyWithFallback(resource, gFormatKeys[style], &patLen, &status);
}
ures_close(numElements);
// Creates the specified decimal format style of the desired locale.
pattern.setTo(TRUE, patResStr, patLen);
}
if (U_FAILURE(status)) {
return NULL;
}
if(style==UNUM_CURRENCY || style == UNUM_CURRENCY_ISO){
const UChar* currPattern = symbolsToAdopt->getCurrencyPattern();
if(currPattern!=NULL){
pattern.setTo(currPattern, u_strlen(currPattern));
}
}
NumberFormat *f;
if (ns->isAlgorithmic()) {
UnicodeString nsDesc;
UnicodeString nsRuleSetGroup;
UnicodeString nsRuleSetName;
Locale nsLoc;
URBNFRuleSetTag desiredRulesType = URBNF_NUMBERING_SYSTEM;
nsDesc.setTo(ns->getDescription());
int32_t firstSlash = nsDesc.indexOf(gSlash);
int32_t lastSlash = nsDesc.lastIndexOf(gSlash);
if ( lastSlash > firstSlash ) {
CharString nsLocID;
nsLocID.appendInvariantChars(nsDesc.tempSubString(0, firstSlash), status);
nsRuleSetGroup.setTo(nsDesc,firstSlash+1,lastSlash-firstSlash-1);
nsRuleSetName.setTo(nsDesc,lastSlash+1);
nsLoc = Locale::createFromName(nsLocID.data());
UnicodeString SpelloutRules = UNICODE_STRING_SIMPLE("SpelloutRules");
if ( nsRuleSetGroup.compare(SpelloutRules) == 0 ) {
desiredRulesType = URBNF_SPELLOUT;
}
} else {
nsLoc = desiredLocale;
nsRuleSetName.setTo(nsDesc);
}
RuleBasedNumberFormat *r = new RuleBasedNumberFormat(desiredRulesType,nsLoc,status);
if (r == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
r->setDefaultRuleSet(nsRuleSetName,status);
f = r;
} else {
// replace single currency sign in the pattern with double currency sign
// if the style is UNUM_CURRENCY_ISO
if (style == UNUM_CURRENCY_ISO) {
pattern.findAndReplace(UnicodeString(TRUE, gSingleCurrencySign, 1),
UnicodeString(TRUE, gDoubleCurrencySign, 2));
}
// "new DecimalFormat()" does not adopt the symbols if its memory allocation fails.
DecimalFormatSymbols *syms = symbolsToAdopt.orphan();
f = new DecimalFormat(pattern, syms, style, status);
if (f == NULL) {
delete syms;
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
}
f->setLocaleIDs(ures_getLocaleByType(ownedResource.getAlias(), ULOC_VALID_LOCALE, &status),
ures_getLocaleByType(ownedResource.getAlias(), ULOC_ACTUAL_LOCALE, &status));
if (U_FAILURE(status)) {
delete f;
return NULL;
}
return f;
}
UnicodeString &QuantityFormatter::format(
const Formattable& quantity,
const NumberFormat &fmt,
const PluralRules &rules,
UnicodeString &appendTo,
FieldPosition &pos,
UErrorCode &status) const {
if (U_FAILURE(status)) {
return appendTo;
}
UnicodeString count;
const DecimalFormat *decFmt = dynamic_cast<const DecimalFormat *>(&fmt);
if (decFmt != NULL) {
FixedDecimal fd = decFmt->getFixedDecimal(quantity, status);
if (U_FAILURE(status)) {
return appendTo;
}
count = rules.select(fd);
} else {
if (quantity.getType() == Formattable::kDouble) {
count = rules.select(quantity.getDouble());
} else if (quantity.getType() == Formattable::kLong) {
count = rules.select(quantity.getLong());
} else if (quantity.getType() == Formattable::kInt64) {
count = rules.select((double) quantity.getInt64());
} else {
status = U_ILLEGAL_ARGUMENT_ERROR;
return appendTo;
}
}
CharString buffer;
buffer.appendInvariantChars(count, status);
if (U_FAILURE(status)) {
return appendTo;
}
int32_t pluralIndex = getPluralIndex(buffer.data());
if (pluralIndex == -1) {
pluralIndex = 0;
}
const SimplePatternFormatter *pattern = formatters[pluralIndex];
if (pattern == NULL) {
pattern = formatters[0];
}
if (pattern == NULL) {
status = U_INVALID_STATE_ERROR;
return appendTo;
}
UnicodeString formattedNumber;
FieldPosition fpos(pos.getField());
fmt.format(quantity, formattedNumber, fpos, status);
const UnicodeString *params[1] = {&formattedNumber};
int32_t offsets[1];
pattern->format(params, LENGTHOF(params), appendTo, offsets, LENGTHOF(offsets), status);
if (offsets[0] != -1) {
if (fpos.getBeginIndex() != 0 || fpos.getEndIndex() != 0) {
pos.setBeginIndex(fpos.getBeginIndex() + offsets[0]);
pos.setEndIndex(fpos.getEndIndex() + offsets[0]);
}
}
return appendTo;
}
const Hashtable *
LocaleUtility::getAvailableLocaleNames(const UnicodeString & bundleID)
{
// LocaleUtility_cache is a hash-of-hashes. The top-level keys
// are path strings ('bundleID') passed to
// ures_openAvailableLocales. The top-level values are
// second-level hashes. The second-level keys are result strings
// from ures_openAvailableLocales. The second-level values are
// garbage ((void*)1 or other random pointer).
UErrorCode status = U_ZERO_ERROR;
Hashtable * cache;
umtx_lock(NULL);
cache = LocaleUtility_cache;
umtx_unlock(NULL);
if (cache == NULL)
{
cache = new Hashtable(status);
if (cache == NULL || U_FAILURE(status))
{
return NULL; // catastrophic failure; e.g. out of memory
}
cache->setValueDeleter(uhash_deleteHashtable);
Hashtable * h; // set this to final LocaleUtility_cache value
umtx_lock(NULL);
h = LocaleUtility_cache;
if (h == NULL)
{
LocaleUtility_cache = h = cache;
cache = NULL;
ucln_common_registerCleanup(UCLN_COMMON_SERVICE, service_cleanup);
}
umtx_unlock(NULL);
if (cache != NULL)
{
delete cache;
}
cache = h;
}
U_ASSERT(cache != NULL);
Hashtable * htp;
umtx_lock(NULL);
htp = (Hashtable *) cache->get(bundleID);
umtx_unlock(NULL);
if (htp == NULL)
{
htp = new Hashtable(status);
if (htp && U_SUCCESS(status))
{
CharString cbundleID;
cbundleID.appendInvariantChars(bundleID, status);
const char * path = cbundleID.isEmpty() ? NULL : cbundleID.data();
UEnumeration * uenum = ures_openAvailableLocales(path, &status);
for (;;)
{
const UChar * id = uenum_unext(uenum, NULL, &status);
if (id == NULL)
{
break;
}
htp->put(UnicodeString(id), (void *)htp, status);
}
uenum_close(uenum);
if (U_FAILURE(status))
{
delete htp;
return NULL;
}
umtx_lock(NULL);
cache->put(bundleID, (void *)htp, status);
umtx_unlock(NULL);
}
}
return htp;
}