// This is a pattern-of-last-resort used when we can't load a usable pattern out // of a resource. staticconst char16_t gDefaultPattern[] =
{
0x79, 0x4D, 0x4D, 0x64, 0x64, 0x20, 0x68, 0x68, 0x3A, 0x6D, 0x6D, 0x20, 0x61, 0
}; /* "yMMdd hh:mm a" */
// This prefix is designed to NEVER MATCH real text, in order to // suppress the parsing of negative numbers. Adjust as needed (if // this becomes valid Unicode). staticconst char16_t SUPPRESS_NEGATIVE_PREFIX[] = {0xAB00, 0};
/** * These are the tags we expect to see in normal resource bundle files associated * with a locale.
*/ staticconst char16_t QUOTE = 0x27; // Single quote
/* * The field range check bias for each UDateFormatField. * The bias is added to the minimum and maximum values * before they are compared to the parsed number. * For example, the calendar stores zero-based month numbers * but the parsed month numbers start at 1, so the bias is 1. * * A value of -1 means that the value is not checked.
*/ staticconst int32_t gFieldRangeBias[] = {
-1, // 'G' - UDAT_ERA_FIELD
-1, // 'y' - UDAT_YEAR_FIELD
1, // 'M' - UDAT_MONTH_FIELD
0, // 'd' - UDAT_DATE_FIELD
-1, // 'k' - UDAT_HOUR_OF_DAY1_FIELD
-1, // 'H' - UDAT_HOUR_OF_DAY0_FIELD
0, // 'm' - UDAT_MINUTE_FIELD
0, // 's' - UDAT_SECOND_FIELD
-1, // 'S' - UDAT_FRACTIONAL_SECOND_FIELD (0-999?)
-1, // 'E' - UDAT_DAY_OF_WEEK_FIELD (1-7?)
-1, // 'D' - UDAT_DAY_OF_YEAR_FIELD (1 - 366?)
-1, // 'F' - UDAT_DAY_OF_WEEK_IN_MONTH_FIELD (1-5?)
-1, // 'w' - UDAT_WEEK_OF_YEAR_FIELD (1-52?)
-1, // 'W' - UDAT_WEEK_OF_MONTH_FIELD (1-5?)
-1, // 'a' - UDAT_AM_PM_FIELD
-1, // 'h' - UDAT_HOUR1_FIELD
-1, // 'K' - UDAT_HOUR0_FIELD
-1, // 'z' - UDAT_TIMEZONE_FIELD
-1, // 'Y' - UDAT_YEAR_WOY_FIELD
-1, // 'e' - UDAT_DOW_LOCAL_FIELD
-1, // 'u' - UDAT_EXTENDED_YEAR_FIELD
-1, // 'g' - UDAT_JULIAN_DAY_FIELD
-1, // 'A' - UDAT_MILLISECONDS_IN_DAY_FIELD
-1, // 'Z' - UDAT_TIMEZONE_RFC_FIELD
-1, // 'v' - UDAT_TIMEZONE_GENERIC_FIELD
0, // 'c' - UDAT_STANDALONE_DAY_FIELD
1, // 'L' - UDAT_STANDALONE_MONTH_FIELD
-1, // 'Q' - UDAT_QUARTER_FIELD (1-4?)
-1, // 'q' - UDAT_STANDALONE_QUARTER_FIELD
-1, // 'V' - UDAT_TIMEZONE_SPECIAL_FIELD
-1, // 'U' - UDAT_YEAR_NAME_FIELD
-1, // 'O' - UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD
-1, // 'X' - UDAT_TIMEZONE_ISO_FIELD
-1, // 'x' - UDAT_TIMEZONE_ISO_LOCAL_FIELD
-1, // 'r' - UDAT_RELATED_YEAR_FIELD #if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
-1, // ':' - UDAT_TIME_SEPARATOR_FIELD #else
-1, // (no pattern character currently) - UDAT_TIME_SEPARATOR_FIELD #endif
};
// When calendar uses hebr numbering (i.e. he@calendar=hebrew), // offset the years within the current millennium down to 1-999 staticconst int32_t HEBREW_CAL_CUR_MILLENIUM_START_YEAR = 5000; staticconst int32_t HEBREW_CAL_CUR_MILLENIUM_END_YEAR = 6000;
/** * Maximum range for detecting daylight offset of a time zone when parsed time zone * string indicates it's daylight saving time, but the detected time zone does not * observe daylight saving time at the parsed date.
*/ staticconstdouble MAX_DAYLIGHT_DETECTION_RANGE = 30*365*24*60*60*1000.0;
// no matter what the locale's default number format looked like, we want // to modify it so that it doesn't use thousands separators, doesn't always // show the decimal point, and recognizes integers only when parsing staticvoid fixNumberFormatForDates(NumberFormat &nf) {
nf.setGroupingUsed(false);
DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(&nf); if (decfmt != nullptr) {
decfmt->setDecimalSeparatorAlwaysShown(false);
}
nf.setParseIntegerOnly(true);
nf.setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
}
/** * Not for public consumption; used by DateFormat. This constructor * never fails. If the resource data is not available, it uses the * the last resort symbols.
*/
SimpleDateFormat::SimpleDateFormat(const Locale& locale,
UErrorCode& status)
: fPattern(gDefaultPattern),
fLocale(locale)
{ if (U_FAILURE(status)) return;
initializeBooleanAttributes();
initializeCalendar(nullptr, fLocale, status);
fSymbols = DateFormatSymbols::createForLocale(fLocale, status); if (U_FAILURE(status))
{
status = U_ZERO_ERROR; delete fSymbols; // This constructor doesn't fail; it uses last resort data
fSymbols = new DateFormatSymbols(status); /* test for nullptr */ if (fSymbols == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return;
}
}
// fSimpleNumberFormatter references fNumberFormatter, delete it // before we call the = operator which may invalidate fNumberFormatter delete fSimpleNumberFormatter;
fSimpleNumberFormatter = nullptr;
// TimeZoneFormat can now be set independently via setter. // If it is nullptr, it will be lazily initialized from locale. delete fTimeZoneFormat;
fTimeZoneFormat = nullptr;
TimeZoneFormat *otherTZFormat;
{ // Synchronization is required here, when accessing other.fTimeZoneFormat, // because another thread may be concurrently executing other.tzFormat(), // a logically const function that lazily creates other.fTimeZoneFormat. // // Without synchronization, reordered memory writes could allow us // to see a non-null fTimeZoneFormat before the object itself was // fully initialized. In case of a race, it doesn't matter whether // we see a null or a fully initialized other.fTimeZoneFormat, // only that we avoid seeing a partially initialized object. // // Once initialized, no const function can modify fTimeZoneFormat, // meaning that once we have safely grabbed the other.fTimeZoneFormat // pointer, continued synchronization is not required to use it.
Mutex m(&LOCK);
otherTZFormat = other.fTimeZoneFormat;
} if (otherTZFormat) {
fTimeZoneFormat = new TimeZoneFormat(*otherTZFormat);
}
if (fSharedNumberFormatters != nullptr) {
freeSharedNumberFormatters(fSharedNumberFormatters);
fSharedNumberFormatters = nullptr;
} if (other.fSharedNumberFormatters != nullptr) {
fSharedNumberFormatters = allocSharedNumberFormatters(); if (fSharedNumberFormatters) { for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
SharedObject::copyPtr(
other.fSharedNumberFormatters[i],
fSharedNumberFormatters[i]);
}
}
}
UErrorCode localStatus = U_ZERO_ERROR; // SimpleNumberFormatter does not have a copy constructor. Furthermore, // it references data from an internal field, fNumberFormatter, // so we must rematerialize that reference after copying over the number formatter.
initSimpleNumberFormatter(localStatus); return *this;
}
void SimpleDateFormat::construct(EStyle timeStyle,
EStyle dateStyle, const Locale& locale,
UErrorCode& status)
{ // called by several constructors to load pattern data from the resources if (U_FAILURE(status)) return;
// We will need the calendar to know what type of symbols to load.
initializeCalendar(nullptr, locale, status); if (U_FAILURE(status)) return;
// Load date time patterns directly from resources. constchar* cType = fCalendar ? fCalendar->getType() : nullptr;
LocalUResourceBundlePointer bundle(ures_open(nullptr, locale.getBaseName(), &status)); if (U_FAILURE(status)) return;
// Check for "gregorian" fallback. if (cTypeIsGregorian || status == U_MISSING_RESOURCE_ERROR) {
status = U_ZERO_ERROR;
dateTimePatterns.adoptInstead(
ures_getByKeyWithFallback(bundle.getAlias(), "calendar/gregorian/DateTimePatterns",
(UResourceBundle*)nullptr, &status));
} if (U_FAILURE(status)) return;
LocalUResourceBundlePointer currentBundle;
if (ures_getSize(dateTimePatterns.getAlias()) <= kDateTime)
{
status = U_INVALID_FORMAT_ERROR; return;
}
// create a symbols object from the locale
fSymbols = DateFormatSymbols::createForLocale(locale, status); if (U_FAILURE(status)) return; /* test for nullptr */ if (fSymbols == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return;
}
UnicodeString timePattern; if (timeStyle >= kFull && timeStyle <= kShort) { bool hasRgOrHcSubtag = false; // also use DTPG if the locale has the "rg" or "hc" ("hours") subtag-- even if the overriding region // or hour cycle is the same as the one we get by default, we go through the DateTimePatternGenerator
UErrorCode dummyErr1 = U_ZERO_ERROR, dummyErr2 = U_ZERO_ERROR; if (locale.getKeywordValue("rg", nullptr, 0, dummyErr1) > 0 || locale.getKeywordValue("hours", nullptr, 0, dummyErr2) > 0) {
hasRgOrHcSubtag = true;
}
constchar* baseLocID = locale.getBaseName(); if (baseLocID != nullptr && uprv_strcmp(baseLocID,"und")!=0) {
UErrorCode useStatus = U_ZERO_ERROR;
Locale baseLoc(baseLocID);
Locale validLoc(getLocale(ULOC_VALID_LOCALE, useStatus)); if (hasRgOrHcSubtag || (U_SUCCESS(useStatus) && validLoc!=baseLoc)) { bool useDTPG = hasRgOrHcSubtag; constchar* baseReg = baseLoc.getCountry(); // empty string if no region if ((baseReg != nullptr && baseReg[0] != 0 &&
uprv_strncmp(baseReg,validLoc.getCountry(),ULOC_COUNTRY_CAPACITY)!=0)
|| uprv_strncmp(baseLoc.getLanguage(),validLoc.getLanguage(),ULOC_LANG_CAPACITY)!=0) { // use DTPG if // * baseLoc has a region and validLoc does not have the same one (or has none), OR // * validLoc has a different language code than baseLoc // * the original locale has the rg or hc subtag
useDTPG = true;
} if (useDTPG) { // The standard time formats may have the wrong time cycle, because: // the valid locale differs in important ways (region, language) from // the base locale. // We could *also* check whether they do actually have a mismatch with // the time cycle preferences for the region, but that is a lot more // work for little or no additional benefit, since just going ahead // and always synthesizing the time format as per the following should // create a locale-appropriate pattern with cycle that matches the // region preferences anyway.
LocalPointer<DateTimePatternGenerator> dtpg(DateTimePatternGenerator::createInstanceNoStdPat(locale, useStatus)); if (U_SUCCESS(useStatus)) {
UnicodeString timeSkeleton(true, timeSkeletons[timeStyle], -1);
timePattern = dtpg->getBestPattern(timeSkeleton, useStatus);
}
}
}
}
}
// if the pattern should include both date and time information, use the date/time // pattern string as a guide to tell use how to glue together the appropriate date // and time pattern strings. if ((timeStyle != kNone) && (dateStyle != kNone))
{
UnicodeString tempus1(timePattern); if (tempus1.length() == 0) {
currentBundle.adoptInstead(
ures_getByIndex(dateTimePatterns.getAlias(), static_cast<int32_t>(timeStyle), nullptr, &status)); if (U_FAILURE(status)) {
status = U_INVALID_FORMAT_ERROR; return;
} switch (ures_getType(currentBundle.getAlias())) { case URES_STRING: {
resStr = ures_getString(currentBundle.getAlias(), &resStrLen, &status); break;
} case URES_ARRAY: {
resStr = ures_getStringByIndex(currentBundle.getAlias(), 0, &resStrLen, &status);
ovrStr = ures_getStringByIndex(currentBundle.getAlias(), 1, &ovrStrLen, &status);
fTimeOverride.setTo(true, ovrStr, ovrStrLen); break;
} default: {
status = U_INVALID_FORMAT_ERROR; return;
}
}
void
SimpleDateFormat::initialize(const Locale& locale,
UErrorCode& status)
{ if (U_FAILURE(status)) return;
parsePattern(); // Need this before initNumberFormatters(), to set fHasHanYearChar
// Simple-minded hack to force Gannen year numbering for ja@calendar=japanese // if format is non-numeric (includes 年) and fDateOverride is not already specified. // Now this does get updated if applyPattern subsequently changes the pattern type. if (fDateOverride.isBogus() && fHasHanYearChar &&
fCalendar != nullptr && uprv_strcmp(fCalendar->getType(),"japanese") == 0 &&
uprv_strcmp(fLocale.getLanguage(),"ja") == 0) {
fDateOverride.setTo(u"y=jpanyear", -1);
}
// We don't need to check that the row count is >= 1, since all 2d arrays have at // least one row
fNumberFormat = NumberFormat::createInstance(locale, status); if (fNumberFormat != nullptr && U_SUCCESS(status))
{
fixNumberFormatForDates(*fNumberFormat); //fNumberFormat->setLenient(true); // Java uses a custom DateNumberFormat to format/parse
} elseif (U_SUCCESS(status))
{
status = U_MISSING_RESOURCE_ERROR;
}
}
/* Initialize the fields we use to disambiguate ambiguous years. Separate * so we can call it from readObject().
*/ void SimpleDateFormat::initializeDefaultCentury()
{ if(fCalendar) {
fHaveDefaultCentury = fCalendar->haveDefaultCentury(); if(fHaveDefaultCentury) {
fDefaultCenturyStart = fCalendar->defaultCenturyStart();
fDefaultCenturyStartYear = fCalendar->defaultCenturyStartYear();
} else {
fDefaultCenturyStart = DBL_MIN;
fDefaultCenturyStartYear = -1;
}
}
}
/* * Initialize the boolean attributes. Separate so we can call it from all constructors.
*/ void SimpleDateFormat::initializeBooleanAttributes()
{
UErrorCode status = U_ZERO_ERROR;
/* Define one-century window into which to disambiguate dates using * two-digit years. Make public in JDK 1.2.
*/ void SimpleDateFormat::parseAmbiguousDatesAsAfter(UDate startDate, UErrorCode& status)
{ if(U_FAILURE(status)) { return;
} if(!fCalendar) {
status = U_ILLEGAL_ARGUMENT_ERROR; return;
}
UnicodeString&
SimpleDateFormat::_format(Calendar& cal, UnicodeString& appendTo,
FieldPositionHandler& handler, UErrorCode& status) const
{ if ( U_FAILURE(status) ) { return appendTo;
}
Calendar* workCal = &cal;
Calendar* calClone = nullptr; if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) { // Different calendar type // We use the time and time zone from the input calendar, but // do not use the input calendar for field calculation.
calClone = fCalendar->clone(); if (calClone != nullptr) {
UDate t = cal.getTime(status);
calClone->setTime(t, status);
calClone->setTimeZone(cal.getTimeZone());
workCal = calClone;
} else {
status = U_MEMORY_ALLOCATION_ERROR; return appendTo;
}
}
// loop through the pattern string character by character
int32_t patternLength = fPattern.length(); for (int32_t i = 0; i < patternLength && U_SUCCESS(status); ++i) {
char16_t ch = fPattern[i];
// Use subFormat() to format a repeated pattern character // when a different pattern or non-pattern character is seen if (ch != prevCh && count > 0) {
subFormat(appendTo, prevCh, count, capitalizationContext, fieldNum++,
prevCh, handler, *workCal, status);
count = 0;
} if (ch == QUOTE) { // Consecutive single quotes are a single quote literal, // either outside of quotes or between quotes if ((i+1) < patternLength && fPattern[i+1] == QUOTE) {
appendTo += QUOTE;
++i;
} else {
inQuote = ! inQuote;
}
} elseif (!inQuote && isSyntaxChar(ch)) { // ch is a date-time pattern character to be interpreted // by subFormat(); count the number of times it is repeated
prevCh = ch;
++count;
} else { // Append quoted characters and unquoted non-pattern characters
appendTo += ch;
}
}
// Format the last item in the pattern, if any if (count > 0) {
subFormat(appendTo, prevCh, count, capitalizationContext, fieldNum++,
prevCh, handler, *workCal, status);
}
/* Map calendar field into calendar field level. * the larger the level, the smaller the field unit. * For example, UCAL_ERA level is 0, UCAL_YEAR level is 10, * UCAL_MONTH level is 20. * NOTE: if new fields adds in, the table needs to update.
*/ const int32_t
SimpleDateFormat::fgCalendarFieldToLevel[] =
{ /*GyM*/ 0, 10, 20, /*wW*/ 20, 30, /*dDEF*/ 30, 20, 30, 30, /*ahHm*/ 40, 50, 50, 60, /*sS*/ 70, 80, /*z?Y*/ 0, 0, 10, /*eug*/ 30, 10, 0, /*A?.*/ 40, 0, 0
};
int32_t SimpleDateFormat::getLevelFromChar(char16_t ch) { // Map date field LETTER into calendar field level. // the larger the level, the smaller the field unit. // NOTE: if new fields adds in, the table needs to update. staticconst int32_t mapCharToLevel[] = {
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, //
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, // ! " # $ % & ' ( ) * + , - . /
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, #if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR // 0 1 2 3 4 5 6 7 8 9 : ; < = > ?
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, -1, -1, -1, -1, -1, #else // 0 1 2 3 4 5 6 7 8 9 : ; < = > ?
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, #endif // @ A B C D E F G H I J K L M N O
-1, 40, -1, -1, 20, 30, 30, 0, 50, -1, -1, 50, 20, 20, -1, 0, // P Q R S T U V W X Y Z [ \ ] ^ _
-1, 20, -1, 80, -1, 10, 0, 30, 0, 10, 0, -1, -1, -1, -1, -1, // ` a b c d e f g h i j k l m n o
-1, 40, -1, 30, 30, 30, -1, 0, 50, -1, -1, 50, 0, 60, -1, -1, // p q r s t u v w x y z { | } ~
-1, 20, 10, 70, -1, 10, 0, 20, 0, 10, 0, -1, -1, -1, -1, -1
};
/** * Append symbols[value] to dst. Make sure the array index is not out * of bounds.
*/ staticinlinevoid
_appendSymbol(UnicodeString& dst,
int32_t value, const UnicodeString* symbols,
int32_t symbolsCount) {
U_ASSERT(0 <= value && value < symbolsCount); if (0 <= value && value < symbolsCount) {
dst += symbols[value];
}
}
staticinlinevoid
_appendSymbolWithMonthPattern(UnicodeString& dst, int32_t value, const UnicodeString* symbols, int32_t symbolsCount, const UnicodeString* monthPattern, UErrorCode& status) {
U_ASSERT(0 <= value && value < symbolsCount); if (0 <= value && value < symbolsCount) { if (monthPattern == nullptr) {
dst += symbols[value];
} else {
SimpleFormatter(*monthPattern, 1, 1, status).format(symbols[value], dst, status);
}
}
}
while (moreToProcess) {
int32_t delimiterPosition = str.indexOf(static_cast<char16_t>(ULOC_KEYWORD_ITEM_SEPARATOR_UNICODE), start); if (delimiterPosition == -1) {
moreToProcess = false;
len = str.length() - start;
} else {
len = delimiterPosition - start;
}
UnicodeString currentString(str,start,len);
int32_t equalSignPosition = currentString.indexOf(static_cast<char16_t>(ULOC_KEYWORD_ASSIGN_UNICODE), 0); if (equalSignPosition == -1) { // Simple override string such as "hebrew"
nsName.setTo(currentString);
ovrField.setToBogus();
} else { // Field specific override string such as "y=hebrew"
nsName.setTo(currentString,equalSignPosition+1);
ovrField.setTo(currentString,0,1); // We just need the first character.
}
int32_t nsNameHash = nsName.hashCode(); // See if the numbering system is in the override list, if not, then add it.
NSOverride *curr = overrideList; const SharedNumberFormat *snf = nullptr;
UBool found = false; while ( curr && !found ) { if ( curr->hash == nsNameHash ) {
snf = curr->snf;
found = true;
}
curr = curr->next;
}
if (!found) {
LocalPointer<NSOverride> cur(new NSOverride); if (!cur.isNull()) { char kw[ULOC_KEYWORD_AND_VALUES_CAPACITY];
uprv_strcpy(kw,"numbers=");
nsName.extract(0,len,kw+8,ULOC_KEYWORD_AND_VALUES_CAPACITY-8,US_INV);
// Now that we have an appropriate number formatter, fill in the appropriate spaces in the // number formatters table. if (ovrField.isBogus()) { switch (type) { case kOvrStrDate: case kOvrStrBoth: { for ( int8_t i=0 ; i<kDateFieldsCount; i++ ) {
SharedObject::copyPtr(snf, fSharedNumberFormatters[kDateFields[i]]);
} if (type==kOvrStrDate) { break;
}
U_FALLTHROUGH;
} case kOvrStrTime : { for ( int8_t i=0 ; i<kTimeFieldsCount; i++ ) {
SharedObject::copyPtr(snf, fSharedNumberFormatters[kTimeFields[i]]);
} break;
}
}
} else { // if the pattern character is unrecognized, signal an error and bail out
UDateFormatField patternCharIndex =
DateFormatSymbols::getPatternCharIndex(ovrField.charAt(0)); if (patternCharIndex == UDAT_FIELD_COUNT) {
status = U_INVALID_FORMAT_ERROR; if (overrideList) {
overrideList->free();
} return;
}
SharedObject::copyPtr(snf, fSharedNumberFormatters[patternCharIndex]);
}
// if the pattern character is unrecognized, signal an error and dump out if (patternCharIndex == UDAT_FIELD_COUNT)
{ if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
status = U_INVALID_FORMAT_ERROR;
} return;
}
UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex];
int32_t value = 0; // Don't get value unless it is useful if (field < UCAL_FIELD_COUNT) {
value = (patternCharIndex != UDAT_RELATED_YEAR_FIELD)? cal.get(field, status): cal.getRelatedYear(status); if (U_FAILURE(status)) { return;
}
}
const NumberFormat *currentNumberFormat = getNumberFormatByIndex(patternCharIndex); if (currentNumberFormat == nullptr) {
status = U_INTERNAL_PROGRAM_ERROR; return;
}
switch (patternCharIndex) {
// for any "G" symbol, write out the appropriate era string // "GGGG" is wide era name, "GGGGG" is narrow era name, anything else is abbreviated name case UDAT_ERA_FIELD:
{ constauto* calType = cal.getType(); if (uprv_strcmp(calType,"chinese") == 0 || uprv_strcmp(calType,"dangi") == 0) {
zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, 9); // as in ICU4J
} else { if (count == 5) {
_appendSymbol(appendTo, value, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageEraNarrow;
} elseif (count == 4) {
_appendSymbol(appendTo, value, fSymbols->fEraNames, fSymbols->fEraNamesCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageEraWide;
} else {
_appendSymbol(appendTo, value, fSymbols->fEras, fSymbols->fErasCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageEraAbbrev;
}
}
} break;
case UDAT_YEAR_NAME_FIELD: if (fSymbols->fShortYearNames != nullptr && value <= fSymbols->fShortYearNamesCount) { // the Calendar YEAR field runs 1 through 60 for cyclic years
_appendSymbol(appendTo, value - 1, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount); break;
} // else fall through to numeric year handling, do not break here
U_FALLTHROUGH;
// OLD: for "yyyy", write out the whole year; for "yy", write out the last 2 digits // NEW: UTS#35: //Year y yy yyy yyyy yyyyy //AD 1 1 01 001 0001 00001 //AD 12 12 12 012 0012 00012 //AD 123 123 23 123 0123 00123 //AD 1234 1234 34 1234 1234 01234 //AD 12345 12345 45 12345 12345 12345 case UDAT_YEAR_FIELD: case UDAT_YEAR_WOY_FIELD: if (fDateOverride.compare(hebr)==0 && value>HEBREW_CAL_CUR_MILLENIUM_START_YEAR && value<HEBREW_CAL_CUR_MILLENIUM_END_YEAR) {
value-=HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
} if(count == 2)
zeroPaddingNumber(currentNumberFormat, appendTo, value, 2, 2); else
zeroPaddingNumber(currentNumberFormat, appendTo, value, count, maxIntCount); break;
// for "MMMM"/"LLLL", write out the whole month name, for "MMM"/"LLL", write out the month // abbreviation, for "M"/"L" or "MM"/"LL", write out the month as a number with the // appropriate number of digits // for "MMMMM"/"LLLLL", use the narrow form case UDAT_MONTH_FIELD: case UDAT_STANDALONE_MONTH_FIELD: if (uprv_strcmp(cal.getType(),"hebrew") == 0) { if (HebrewCalendar::isLeapYear(cal.get(UCAL_YEAR,status)) && value == 6 && count >= 3 )
value = 13; // Show alternate form for Adar II in leap years in Hebrew calendar. if (!HebrewCalendar::isLeapYear(cal.get(UCAL_YEAR,status)) && value >= 6 && count < 3 )
value--; // Adjust the month number down 1 in Hebrew non-leap years, i.e. Adar is 6, not 7.
}
{
int32_t isLeapMonth = (fSymbols->fLeapMonthPatterns != nullptr && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount)?
cal.get(UCAL_IS_LEAP_MONTH, status): 0; // should consolidate the next section by using arrays of pointers & counts for the right symbols... if (count == 5) { if (patternCharIndex == UDAT_MONTH_FIELD) {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fNarrowMonths, fSymbols->fNarrowMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatNarrow]): nullptr, status);
} else {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneNarrowMonths, fSymbols->fStandaloneNarrowMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneNarrow]): nullptr, status);
}
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthNarrow;
} elseif (count == 4) { if (patternCharIndex == UDAT_MONTH_FIELD) {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fMonths, fSymbols->fMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide]): nullptr, status);
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
} else {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide]): nullptr, status);
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
}
} elseif (count == 3) { if (patternCharIndex == UDAT_MONTH_FIELD) {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fShortMonths, fSymbols->fShortMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev]): nullptr, status);
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
} else {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev]): nullptr, status);
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
}
} else {
UnicodeString monthNumber;
zeroPaddingNumber(currentNumberFormat,monthNumber, value + 1, count, maxIntCount);
_appendSymbolWithMonthPattern(appendTo, 0, &monthNumber, 1,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric]): nullptr, status);
}
} break;
// for "k" and "kk", write out the hour, adjusting midnight to appear as "24" case UDAT_HOUR_OF_DAY1_FIELD: if (value == 0)
zeroPaddingNumber(currentNumberFormat,appendTo, cal.getMaximum(UCAL_HOUR_OF_DAY) + 1, count, maxIntCount); else
zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount); break;
// for "ee" or "e", use local numeric day-of-the-week // for "EEEEEE" or "eeeeee", write out the short day-of-the-week name // for "EEEEE" or "eeeee", write out the narrow day-of-the-week name // for "EEEE" or "eeee", write out the wide day-of-the-week name // for "EEE" or "EE" or "E" or "eee", write out the abbreviated day-of-the-week name case UDAT_DOW_LOCAL_FIELD: if ( count < 3 ) {
zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount); break;
} // fall through to EEEEE-EEE handling, but for that we don't want local day-of-week, // we want standard day-of-week, so first fix value to work for EEEEE-EEE.
value = cal.get(UCAL_DAY_OF_WEEK, status); if (U_FAILURE(status)) { return;
} // fall through, do not break here
U_FALLTHROUGH; case UDAT_DAY_OF_WEEK_FIELD: if (count == 5) {
_appendSymbol(appendTo, value, fSymbols->fNarrowWeekdays,
fSymbols->fNarrowWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
} elseif (count == 4) {
_appendSymbol(appendTo, value, fSymbols->fWeekdays,
fSymbols->fWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
} elseif (count == 6) {
_appendSymbol(appendTo, value, fSymbols->fShorterWeekdays,
fSymbols->fShorterWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
} else {
_appendSymbol(appendTo, value, fSymbols->fShortWeekdays,
fSymbols->fShortWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
} break;
// for "ccc", write out the abbreviated day-of-the-week name // for "cccc", write out the wide day-of-the-week name // for "ccccc", use the narrow day-of-the-week name // for "ccccc", use the short day-of-the-week name case UDAT_STANDALONE_DAY_FIELD: if ( count < 3 ) {
zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, maxIntCount); break;
} // fall through to alpha DOW handling, but for that we don't want local day-of-week, // we want standard day-of-week, so first fix value.
value = cal.get(UCAL_DAY_OF_WEEK, status); if (U_FAILURE(status)) { return;
} if (count == 5) {
_appendSymbol(appendTo, value, fSymbols->fStandaloneNarrowWeekdays,
fSymbols->fStandaloneNarrowWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
} elseif (count == 4) {
_appendSymbol(appendTo, value, fSymbols->fStandaloneWeekdays,
fSymbols->fStandaloneWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
} elseif (count == 6) {
_appendSymbol(appendTo, value, fSymbols->fStandaloneShorterWeekdays,
fSymbols->fStandaloneShorterWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
} else { // count == 3
_appendSymbol(appendTo, value, fSymbols->fStandaloneShortWeekdays,
fSymbols->fStandaloneShortWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
} break;
// for "a" symbol, write out the whole AM/PM string case UDAT_AM_PM_FIELD: if (count < 5) {
_appendSymbol(appendTo, value, fSymbols->fAmPms,
fSymbols->fAmPmsCount);
} else {
_appendSymbol(appendTo, value, fSymbols->fNarrowAmPms,
fSymbols->fNarrowAmPmsCount);
} break;
// if we see pattern character for UDAT_TIME_SEPARATOR_FIELD (none currently defined), // write out the time separator string. Leave support in for future definition. case UDAT_TIME_SEPARATOR_FIELD:
{
UnicodeString separator;
appendTo += fSymbols->getTimeSeparatorString(separator);
} break;
// for "h" and "hh", write out the hour, adjusting noon and midnight to show up // as "12" case UDAT_HOUR1_FIELD: if (value == 0)
zeroPaddingNumber(currentNumberFormat,appendTo, cal.getLeastMaximum(UCAL_HOUR) + 1, count, maxIntCount); else
zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount); break;
// Note: "midnight" can be ambiguous as to whether it refers to beginning of day or end of day. // For ICU 57 output of "midnight" is temporarily suppressed.
// For "midnight" and "noon": // Time, as displayed, must be exactly noon or midnight. // This means minutes and seconds, if present, must be zero. if ((/*hour == 0 ||*/ hour == 12) &&
(!fHasMinute || cal.get(UCAL_MINUTE, status) == 0) &&
(!fHasSecond || cal.get(UCAL_SECOND, status) == 0)) { // Stealing am/pm value to use as our array index. // It works out: am/midnight are both 0, pm/noon are both 1, // 12 am is 12 midnight, and 12 pm is 12 noon.
int32_t val = cal.get(UCAL_AM_PM, status);
// toAppend is nullptr if time isn't exactly midnight or noon (as displayed). // toAppend is bogus if time is midnight or noon, but no localized string exists. // In either case, fall back to am/pm. if (toAppend == nullptr || toAppend->isBogus()) { // Reformat with identical arguments except ch, now changed to 'a'. // We are passing a different fieldToOutput because we want to add // 'b' to field position. This makes this fallback stable when // there is a data change on locales.
subFormat(appendTo, u'a', count, capitalizationContext, fieldNum, u'b', handler, cal, status); return;
} else {
appendTo += *toAppend;
}
break;
}
case UDAT_FLEXIBLE_DAY_PERIOD_FIELD:
{ // TODO: Maybe fetch the DayperiodRules during initialization (instead of at the first // loading of an instance) if a relevant pattern character (b or B) is used. const DayPeriodRules *ruleSet = DayPeriodRules::getInstance(this->getSmpFmtLocale(), status); if (U_FAILURE(status)) { // Data doesn't conform to spec, therefore loading failed. break;
} if (ruleSet == nullptr) { // Data doesn't exist for the locale we're looking for. // Falling back to am/pm. // We are passing a different fieldToOutput because we want to add // 'B' to field position. This makes this fallback stable when // there is a data change on locales.
subFormat(appendTo, u'a', count, capitalizationContext, fieldNum, u'B', handler, cal, status); return;
}
// Get current display time.
int32_t hour = cal.get(UCAL_HOUR_OF_DAY, status);
int32_t minute = 0; if (fHasMinute) {
minute = cal.get(UCAL_MINUTE, status);
}
int32_t second = 0; if (fHasSecond) {
second = cal.get(UCAL_SECOND, status);
}
// Rule set exists, therefore periodType can't be UNKNOWN. // Get localized string.
U_ASSERT(periodType != DayPeriodRules::DAYPERIOD_UNKNOWN);
UnicodeString *toAppend = nullptr;
int32_t index;
// Note: "midnight" can be ambiguous as to whether it refers to beginning of day or end of day. // For ICU 57 output of "midnight" is temporarily suppressed.
// General Periods -> AM/PM. if (periodType == DayPeriodRules::DAYPERIOD_AM ||
periodType == DayPeriodRules::DAYPERIOD_PM ||
toAppend->isBogus()) { // We are passing a different fieldToOutput because we want to add // 'B' to field position iterator. This makes this fallback stable when // there is a data change on locales.
subFormat(appendTo, u'a', count, capitalizationContext, fieldNum, u'B', handler, cal, status); return;
} else {
appendTo += *toAppend;
}
break;
}
// all of the other pattern symbols can be formatted as simple numbers with // appropriate zero padding default:
zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount); break;
} #if !UCONFIG_NO_BREAK_ITERATION // if first field, check to see whether we need to and are able to titlecase it if (fieldNum == 0 && fCapitalizationBrkIter != nullptr && appendTo.length() > beginOffset &&
u_islower(appendTo.char32At(beginOffset))) {
UBool titlecase = false; switch (capitalizationContext) { case UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE:
titlecase = true; break; case UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU:
titlecase = fSymbols->fCapitalization[capContextUsageType][0]; break; case UDISPCTX_CAPITALIZATION_FOR_STANDALONE:
titlecase = fSymbols->fCapitalization[capContextUsageType][1]; break; default: // titlecase = false; break;
} if (titlecase) {
BreakIterator* const mutableCapitalizationBrkIter = fCapitalizationBrkIter->clone();
UnicodeString firstField(appendTo, beginOffset);
firstField.toTitle(mutableCapitalizationBrkIter, fLocale, U_TITLECASE_NO_LOWERCASE | U_TITLECASE_NO_BREAK_ADJUSTMENT);
appendTo.replaceBetween(beginOffset, appendTo.length(), firstField); delete mutableCapitalizationBrkIter;
}
} #endif
void SimpleDateFormat::adoptNumberFormat(NumberFormat *formatToAdopt) { // Null out the fast formatter, it references fNumberFormat which we're // about to invalidate delete fSimpleNumberFormatter;
fSimpleNumberFormatter = nullptr;
// We successfully set the default number format. Now delete the overrides // (can't fail). if (fSharedNumberFormatters) {
freeSharedNumberFormatters(fSharedNumberFormatters);
fSharedNumberFormatters = nullptr;
}
// We must ensure fSharedNumberFormatters is allocated. if (fSharedNumberFormatters == nullptr) {
fSharedNumberFormatters = allocSharedNumberFormatters(); if (fSharedNumberFormatters == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return;
}
} const SharedNumberFormat *newFormat = createSharedNumberFormat(fmt.orphan()); if (newFormat == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; return;
} for (int i=0; i<fields.length(); i++) {
char16_t field = fields.charAt(i); // if the pattern character is unrecognized, signal an error and bail out
UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(field); if (patternCharIndex == UDAT_FIELD_COUNT) {
status = U_INVALID_FORMAT_ERROR;
newFormat->deleteIfZeroRefCount(); return;
}
// Set the number formatter in the table
SharedObject::copyPtr(
newFormat, fSharedNumberFormatters[patternCharIndex]);
}
newFormat->deleteIfZeroRefCount();
}
if (currentNumberFormat == fNumberFormat && fSimpleNumberFormatter) { // Can use fast path // We create UFormattedNumberData ourselves to avoid a heap allocation // and corresponding free. Set the pointer to null afterwards to prevent // the implementation from attempting to free it.
UErrorCode localStatus = U_ZERO_ERROR;
number::impl::UFormattedNumberData data;
data.quantity.setToLong(value);
number::SimpleNumber number(&data, localStatus);
number.setMinimumIntegerDigits(minDigits, localStatus);
number.setMaximumIntegerDigits(maxDigits, localStatus);
// Check for RBNF (no clone necessary) constauto* rbnf = dynamic_cast<const RuleBasedNumberFormat*>(currentNumberFormat); if (rbnf != nullptr) {
FieldPosition pos(FieldPosition::DONT_CARE);
rbnf->format(value, appendTo, pos); // 3rd arg is there to speed up processing return;
}
// Fall back to slow path (clone and mutate the NumberFormat) if (currentNumberFormat != nullptr) {
FieldPosition pos(FieldPosition::DONT_CARE);
LocalPointer<NumberFormat> nf(currentNumberFormat->clone());
nf->setMinimumIntegerDigits(minDigits);
nf->setMaximumIntegerDigits(maxDigits);
nf->format(value, appendTo, pos); // 3rd arg is there to speed up processing
}
}
/** * Return true if the given format character, occurring count * times, represents a numeric field.
*/
UBool SimpleDateFormat::isNumeric(char16_t formatChar, int32_t count) { return DateFormatSymbols::isNumericPatternChar(formatChar, count);
}
UBool
SimpleDateFormat::isAtNumericField(const UnicodeString &pattern, int32_t patternOffset) { if (patternOffset >= pattern.length()) { // not at any field returnfalse;
}
char16_t ch = pattern.charAt(patternOffset);
UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch); if (f == UDAT_FIELD_COUNT) { // not at any field returnfalse;
}
int32_t i = patternOffset; while (pattern.charAt(++i) == ch) {} return DateFormatSymbols::isNumericField(f, i - patternOffset);
}
UBool
SimpleDateFormat::isAfterNonNumericField(const UnicodeString &pattern, int32_t patternOffset) { if (patternOffset <= 0) { // not after any field returnfalse;
}
char16_t ch = pattern.charAt(--patternOffset);
UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch); if (f == UDAT_FIELD_COUNT) { // not after any field returnfalse;
}
int32_t i = patternOffset; while (pattern.charAt(--i) == ch) {} return !DateFormatSymbols::isNumericField(f, patternOffset - i);
}
// For parsing abutting numeric fields. 'abutPat' is the // offset into 'pattern' of the first of 2 or more abutting // numeric fields. 'abutStart' is the offset into 'text' // where parsing the fields begins. 'abutPass' starts off as 0 // and increments each time we try to parse the fields.
int32_t abutPat = -1; // If >=0, we are in a run of abutting numeric fields
int32_t abutStart = 0;
int32_t abutPass = 0;
UBool inQuote = false;
Calendar* calClone = nullptr;
Calendar *workCal = &cal; if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) { // Different calendar type // We use the time/zone from the input calendar, but // do not use the input calendar for field calculation.
calClone = fCalendar->clone(); if (calClone != nullptr) {
calClone->setTime(cal.getTime(status),status); if (U_FAILURE(status)) { goto ExitParse;
}
calClone->setTimeZone(cal.getTimeZone());
workCal = calClone;
} else {
status = U_MEMORY_ALLOCATION_ERROR; goto ExitParse;
}
}
if (fSymbols->fLeapMonthPatterns != nullptr && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
numericLeapMonthFormatter = new MessageFormat(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric], fLocale, status); if (numericLeapMonthFormatter == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; goto ExitParse;
} elseif (U_FAILURE(status)) { goto ExitParse; // this will delete numericLeapMonthFormatter
}
}
for (int32_t i=0; i<fPattern.length(); ++i) {
char16_t ch = fPattern.charAt(i);
// Handle alphabetic field characters. if (!inQuote && isSyntaxChar(ch)) {
int32_t fieldPat = i;
// Count the length of this field specifier
count = 1; while ((i+1)<fPattern.length() &&
fPattern.charAt(i+1) == ch) {
++count;
++i;
}
if (isNumeric(ch, count)) { if (abutPat < 0) { // Determine if there is an abutting numeric field. // Record the start of a set of abutting numeric fields. if (isAtNumericField(fPattern, i + 1)) {
abutPat = fieldPat;
abutStart = pos;
abutPass = 0;
}
}
} else {
abutPat = -1; // End of any abutting fields
}
// Handle fields within a run of abutting numeric fields. Take // the pattern "HHmmss" as an example. We will try to parse // 2/2/2 characters of the input text, then if that fails, // 1/2/2. We only adjust the width of the leftmost field; the // others remain fixed. This allows "123456" => 12:34:56, but // "12345" => 1:23:45. Likewise, for the pattern "yyyyMMdd" we // try 4/2/2, 3/2/2, 2/2/2, and finally 1/2/2. if (abutPat >= 0) { // If we are at the start of a run of abutting fields, then // shorten this field in each pass. If we can't shorten // this field any more, then the parse of this set of // abutting numeric fields has failed. if (fieldPat == abutPat) {
count -= abutPass++; if (count == 0) {
status = U_PARSE_ERROR; goto ExitParse;
}
}
// If the parse fails anywhere in the run, back up to the // start of the run and retry. if (pos < 0) {
i = abutPat - 1;
pos = abutStart; continue;
}
}
// Handle non-numeric fields and non-abutting numeric // fields. elseif (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
int32_t s = subParse(text, pos, ch, count, false, true, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter, &tzTimeType, &dayPeriodInt);
if (s == -pos-1) { // era not present, in special cases allow this to continue // from the position where the era was expected
s = pos;
if (i+1 < fPattern.length()) { // move to next pattern character
char16_t c = fPattern.charAt(i+1);
// check for whitespace if (PatternProps::isWhiteSpace(c)) {
i++; // Advance over run in pattern while ((i+1)<fPattern.length() &&
PatternProps::isWhiteSpace(fPattern.charAt(i+1))) {
++i;
}
}
}
} elseif (s <= 0) {
status = U_PARSE_ERROR; goto ExitParse;
}
pos = s;
}
}
// Handle literal pattern characters. These are any // quoted characters and non-alphabetic unquoted // characters. else {
abutPat = -1; // End of any abutting fields
if (! matchLiterals(fPattern, i, text, pos, getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status), getBooleanAttribute(UDAT_PARSE_PARTIAL_LITERAL_MATCH, status), isLenient())) {
status = U_PARSE_ERROR; goto ExitParse;
}
}
}
// Special hack for trailing "." after non-numeric field. if (text.charAt(pos) == 0x2e && getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) { // only do if the last field is not numeric if (isAfterNonNumericField(fPattern, fPattern.length())) {
pos++; // skip the extra "."
}
}
// If dayPeriod is set, use it in conjunction with hour-of-day to determine am/pm. if (dayPeriodInt >= 0) {
DayPeriodRules::DayPeriod dayPeriod = static_cast<DayPeriodRules::DayPeriod>(dayPeriodInt); const DayPeriodRules *ruleSet = DayPeriodRules::getInstance(this->getSmpFmtLocale(), status);
if (!cal.isSet(UCAL_HOUR) && !cal.isSet(UCAL_HOUR_OF_DAY)) { // If hour is not set, set time to the midpoint of current day period, overwriting // minutes if it's set. double midPoint = ruleSet->getMidPointForDayPeriod(dayPeriod, status);
// If we can't get midPoint we do nothing. if (U_SUCCESS(status)) { // Truncate midPoint toward zero to get the hour. // Any leftover means it was a half-hour.
int32_t midPointHour = static_cast<int32_t>(midPoint);
int32_t midPointMinute = (midPoint - midPointHour) > 0 ? 30 : 0;
// No need to set am/pm because hour-of-day is set last therefore takes precedence.
cal.set(UCAL_HOUR_OF_DAY, midPointHour);
cal.set(UCAL_MINUTE, midPointMinute);
}
} else { int hourOfDay;
if (cal.isSet(UCAL_HOUR_OF_DAY)) { // Hour is parsed in 24-hour format.
hourOfDay = cal.get(UCAL_HOUR_OF_DAY, status);
} else { // Hour is parsed in 12-hour format.
hourOfDay = cal.get(UCAL_HOUR, status); // cal.get() turns 12 to 0 for 12-hour time; change 0 to 12 // so 0 unambiguously means a 24-hour time from above. if (hourOfDay == 0) { hourOfDay = 12; }
}
U_ASSERT(0 <= hourOfDay && hourOfDay <= 23);
// If hour-of-day is 0 or 13 thru 23 then input time in unambiguously in 24-hour format. if (hourOfDay == 0 || (13 <= hourOfDay && hourOfDay <= 23)) { // Make hour-of-day take precedence over (hour + am/pm) by setting it again.
cal.set(UCAL_HOUR_OF_DAY, hourOfDay);
} else { // We have a 12-hour time and need to choose between am and pm. // Behave as if dayPeriod spanned 6 hours each way from its center point. // This will parse correctly for consistent time + period (e.g. 10 at night) as // well as provide a reasonable recovery for inconsistent time + period (e.g. // 9 in the afternoon).
// Assume current time is in the AM. // - Change 12 back to 0 for easier handling of 12am. // - Append minutes as fractional hours because e.g. 8:15 and 8:45 could be parsed // into different half-days if center of dayPeriod is at 14:30. // - cal.get(MINUTE) will return 0 if MINUTE is unset, which works. if (hourOfDay == 12) { hourOfDay = 0; } double currentHour = hourOfDay + (cal.get(UCAL_MINUTE, status)) / 60.0; double midPointHour = ruleSet->getMidPointForDayPeriod(dayPeriod, status);
if (U_SUCCESS(status)) { double hoursAheadMidPoint = currentHour - midPointHour;
// Assume current time is in the AM. if (-6 <= hoursAheadMidPoint && hoursAheadMidPoint < 6) { // Assumption holds; set time as such.
cal.set(UCAL_AM_PM, 0);
} else {
cal.set(UCAL_AM_PM, 1);
}
}
}
}
}
// At this point the fields of Calendar have been set. Calendar // will fill in default values for missing fields when the time // is computed.
parsePos.setIndex(pos);
// This part is a problem: When we call parsedDate.after, we compute the time. // Take the date April 3 2004 at 2:30 am. When this is first set up, the year // will be wrong if we're parsing a 2-digit year pattern. It will be 1904. // April 3 1904 is a Sunday (unlike 2004) so it is the DST onset day. 2:30 am // is therefore an "impossible" time, since the time goes from 1:59 to 3:00 am // on that day. It is therefore parsed out to fields as 3:30 am. Then we // add 100 years, and get April 3 2004 at 3:30 am. Note that April 3 2004 is // a Saturday, so it can have a 2:30 am -- and it should. [LIU] /* UDate parsedDate = calendar.getTime(); if( ambiguousYear[0] && !parsedDate.after(fDefaultCenturyStart) ) { calendar.add(Calendar.YEAR, 100); parsedDate = calendar.getTime(); }
*/ // Because of the above condition, save off the fields in case we need to readjust. // The procedure we use here is not particularly efficient, but there is no other // way to do this given the API restrictions present in Calendar. We minimize // inefficiency by only performing this computation when it might apply, that is, // when the two-digit year is equal to the start year, and thus might fall at the // front or the back of the default century. This only works because we adjust // the year correctly to start with in other cases -- see subParse(). if (ambiguousYear[0] || tzTimeType != UTZFMT_TIME_TYPE_UNKNOWN) // If this is true then the two-digit year == the default start year
{ // We need a copy of the fields, and we need to avoid triggering a call to // complete(), which will recalculate the fields. Since we can't access // the fields[] array in Calendar, we clone the entire object. This will // stop working if Calendar.clone() is ever rewritten to call complete().
Calendar *copy; if (ambiguousYear[0]) {
copy = cal.clone(); // Check for failed cloning. if (copy == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; goto ExitParse;
}
UDate parsedDate = copy->getTime(status); // {sfb} check internalGetDefaultCenturyStart if (fHaveDefaultCentury && (parsedDate < fDefaultCenturyStart)) { // We can't use add here because that does a complete() first.
cal.set(UCAL_YEAR, fDefaultCenturyStartYear + 100);
} delete copy;
}
if (tzTimeType != UTZFMT_TIME_TYPE_UNKNOWN) {
copy = cal.clone(); // Check for failed cloning. if (copy == nullptr) {
status = U_MEMORY_ALLOCATION_ERROR; goto ExitParse;
} const TimeZone & tz = cal.getTimeZone();
BasicTimeZone *btz = nullptr;
// Get local millis
copy->set(UCAL_ZONE_OFFSET, 0);
copy->set(UCAL_DST_OFFSET, 0);
UDate localMillis = copy->getTime(status);
// Make sure parsed time zone type (Standard or Daylight) // matches the rule used by the parsed time zone.
int32_t raw, dst; if (btz != nullptr) { if (tzTimeType == UTZFMT_TIME_TYPE_STANDARD) {
btz->getOffsetFromLocal(localMillis,
UCAL_TZ_LOCAL_STANDARD_FORMER, UCAL_TZ_LOCAL_STANDARD_LATTER, raw, dst, status);
} else {
btz->getOffsetFromLocal(localMillis,
UCAL_TZ_LOCAL_DAYLIGHT_FORMER, UCAL_TZ_LOCAL_DAYLIGHT_LATTER, raw, dst, status);
}
} else { // No good way to resolve ambiguous time at transition, // but following code work in most case.
tz.getOffset(localMillis, true, raw, dst, status);
}
// Now, compare the results with parsed type, either standard or daylight saving time
int32_t resolvedSavings = dst; if (tzTimeType == UTZFMT_TIME_TYPE_STANDARD) { if (dst != 0) { // Override DST_OFFSET = 0 in the result calendar
resolvedSavings = 0;
}
} else { // tztype == TZTYPE_DST if (dst == 0) { if (btz != nullptr) { // This implementation resolves daylight saving time offset // closest rule after the given time.
UDate baseTime = localMillis + raw;
UDate time = baseTime;
UDate limit = baseTime + MAX_DAYLIGHT_DETECTION_RANGE;
TimeZoneTransition trs;
UBool trsAvail;
// Search for DST rule after the given time while (time < limit) {
trsAvail = btz->getNextTransition(time, false, trs); if (!trsAvail) { break;
}
resolvedSavings = trs.getTo()->getDSTSavings(); if (resolvedSavings != 0) { break;
}
time = trs.getTime();
}
if (resolvedSavings == 0) { // If no DST rule after the given time was found, search for // DST rule before.
time = baseTime;
limit = baseTime - MAX_DAYLIGHT_DETECTION_RANGE; while (time > limit) {
trsAvail = btz->getPreviousTransition(time, true, trs); if (!trsAvail) { break;
}
resolvedSavings = trs.getFrom()->getDSTSavings(); if (resolvedSavings != 0) { break;
}
time = trs.getTime() - 1;
}
if (resolvedSavings == 0) {
resolvedSavings = btz->getDSTSavings();
}
}
} else {
resolvedSavings = tz.getDSTSavings();
} if (resolvedSavings == 0) { // final fallback
resolvedSavings = U_MILLIS_PER_HOUR;
}
}
}
cal.set(UCAL_ZONE_OFFSET, raw);
cal.set(UCAL_DST_OFFSET, resolvedSavings); delete copy;
}
}
ExitParse: // Set the parsed result if local calendar is used // instead of the input calendar if (U_SUCCESS(status) && workCal != &cal) {
cal.setTimeZone(workCal->getTimeZone());
cal.setTime(workCal->getTime(status), status);
}
// If any Calendar calls failed, we pretend that we // couldn't parse the string, when in reality this isn't quite accurate-- // we did parse it; the Calendar calls just failed. if (U_FAILURE(status)) {
parsePos.setErrorIndex(pos);
parsePos.setIndex(start);
}
}
// There may be multiple strings in the data[] array which begin with // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech). // We keep track of the longest match, and return that. Note that this // unfortunately requires us to test all array elements.
int32_t bestMatchLength = 0, bestMatch = -1;
UnicodeString bestMatchName;
for (; i < count; ++i) {
int32_t matchLength = 0; if ((matchLength = matchStringWithOptionalDot(text, start, data[i])) > bestMatchLength) {
bestMatchLength = matchLength;
bestMatch = i;
}
}
// scan pattern looking for contiguous literal characters for ( ; i < pattern.length(); i += 1) {
char16_t ch = pattern.charAt(i);
if (!inQuote && isSyntaxChar(ch)) { break;
}
if (ch == QUOTE) { // Match a quote literal ('') inside OR outside of quotes if ((i + 1) < pattern.length() && pattern.charAt(i + 1) == QUOTE) {
i += 1;
} else {
inQuote = !inQuote; continue;
}
}
literal += ch;
}
// at this point, literal contains the literal text // and i is the index of the next non-literal pattern character.
int32_t p;
int32_t t = textOffset;
if (whitespaceLenient) { // trim leading, trailing whitespace from // the literal text
literal.trim();
// ignore any leading whitespace in the text while (t < text.length() && u_isWhitespace(text.charAt(t))) {
t += 1;
}
}
for (p = 0; p < literal.length() && t < text.length();) {
UBool needWhitespace = false;
while (p < literal.length() && PatternProps::isWhiteSpace(literal.charAt(p))) {
needWhitespace = true;
p += 1;
}
if (needWhitespace) {
int32_t tStart = t;
while (t < text.length()) {
char16_t tch = text.charAt(t);
if (!u_isUWhiteSpace(tch) && !PatternProps::isWhiteSpace(tch)) { break;
}
t += 1;
}
// TODO: should we require internal spaces // in lenient mode? (There won't be any // leading or trailing spaces) if (!whitespaceLenient && t == tStart) { // didn't find matching whitespace: // an error in strict mode returnfalse;
}
// In strict mode, this run of whitespace // may have been at the end. if (p >= literal.length()) { break;
}
} if (t >= text.length() || literal.charAt(p) != text.charAt(t)) { // Ran out of text, or found a non-matching character: // OK in lenient mode, an error in strict mode. if (whitespaceLenient) { if (t == textOffset && text.charAt(t) == 0x2e &&
isAfterNonNumericField(pattern, patternOffset)) { // Lenient mode and the literal input text begins with a "." and // we are after a non-numeric field: We skip the "."
++t; continue; // Do not update p.
} // if it is actual whitespace and we're whitespace lenient it's OK
char16_t wsc = text.charAt(t); if(PatternProps::isWhiteSpace(wsc)) { // Lenient mode and it's just whitespace we skip it
++t; continue; // Do not update p.
}
} // hack around oldleniency being a bit of a catch-all bucket and we're just adding support specifically for partial matches if(partialMatchLenient && oldLeniency) { break;
}
returnfalse;
}
++p;
++t;
}
// At this point if we're in strict mode we have a complete match. // If we're in lenient mode we may have a partial match, or no // match at all. if (p <= 0) { // no match. Pretend it matched a run of whitespace // and ignorables in the text. const UnicodeSet *ignorables = nullptr;
UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(pattern.charAt(i)); if (patternCharIndex != UDAT_FIELD_COUNT) {
ignorables = SimpleDateFormatStaticSets::getIgnorables(patternCharIndex);
}
for (t = textOffset; t < text.length(); t += 1) {
char16_t ch = text.charAt(t);
// if we get here, we've got a complete match.
patternOffset = i - 1;
textOffset = t;
returntrue;
}
//---------------------------------------------------------------------- // check both wide and abbrev months. // Does not currently handle monthPattern. // UCalendarDateFields field = UCAL_MONTH
// There may be multiple strings in the data[] array which begin with // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech). // We keep track of the longest match, and return that. Note that this // unfortunately requires us to test all array elements. // But this does not really work for cases such as Chuvash in which // May is "ҫу" and August is "ҫурла"/"ҫур.", hence matchAlphaMonthStrings.
int32_t bestMatchLength = 0, bestMatch = -1;
UnicodeString bestMatchName;
int32_t isLeapMonth = 0;
for (; i < count; ++i) {
int32_t matchLen = 0; if ((matchLen = matchStringWithOptionalDot(text, start, data[i])) > bestMatchLength) {
bestMatch = i;
bestMatchLength = matchLen;
}
if (bestMatch >= 0) { if (field < UCAL_FIELD_COUNT) { // Adjustment for Hebrew Calendar month Adar II if (!strcmp(cal.getType(),"hebrew") && field==UCAL_MONTH && bestMatch==13) {
cal.set(field,6);
} else { if (field == UCAL_YEAR) {
bestMatch++; // only get here for cyclic year names, which match 1-based years 1-60
}
cal.set(field, bestMatch);
} if (monthPattern != nullptr) {
cal.set(UCAL_IS_LEAP_MONTH, isLeapMonth);
}
}
// If there are any spaces here, skip over them. If we hit the end // of the string, then fail. for (;;) { if (start >= text.length()) { return -start;
}
UChar32 c = text.char32At(start); if (!u_isUWhiteSpace(c) /*||*/ && !PatternProps::isWhiteSpace(c)) { break;
}
start += U16_LENGTH(c);
}
pos.setIndex(start);
// We handle a few special cases here where we need to parse // a number value. We handle further, more generic cases below. We need // to handle some of them here because some fields require extra processing on // the parsed value. if (patternCharIndex == UDAT_HOUR_OF_DAY1_FIELD || // k
patternCharIndex == UDAT_HOUR_OF_DAY0_FIELD || // H
patternCharIndex == UDAT_HOUR1_FIELD || // h
patternCharIndex == UDAT_HOUR0_FIELD || // K
(patternCharIndex == UDAT_DOW_LOCAL_FIELD && count <= 2) || // e
(patternCharIndex == UDAT_STANDALONE_DAY_FIELD && count <= 2) || // c
(patternCharIndex == UDAT_MONTH_FIELD && count <= 2) || // M
(patternCharIndex == UDAT_STANDALONE_MONTH_FIELD && count <= 2) || // L
(patternCharIndex == UDAT_QUARTER_FIELD && count <= 2) || // Q
(patternCharIndex == UDAT_STANDALONE_QUARTER_FIELD && count <= 2) || // q
patternCharIndex == UDAT_YEAR_FIELD || // y
patternCharIndex == UDAT_YEAR_WOY_FIELD || // Y
patternCharIndex == UDAT_YEAR_NAME_FIELD || // U (falls back to numeric)
(patternCharIndex == UDAT_ERA_FIELD && isChineseCalendar) || // G
patternCharIndex == UDAT_FRACTIONAL_SECOND_FIELD) // S
{
int32_t parseStart = pos.getIndex(); // It would be good to unify this with the obeyCount logic below, // but that's going to be difficult. const UnicodeString* src;
if (!getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) { // Check the range of the value
int32_t bias = gFieldRangeBias[patternCharIndex]; if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) { return -start;
}
}
pos.setIndex(txtLoc);
}
}
// Make sure that we got a number if // we want one, and didn't get one // if we don't want one. switch (patternCharIndex) { case UDAT_HOUR_OF_DAY1_FIELD: case UDAT_HOUR_OF_DAY0_FIELD: case UDAT_HOUR1_FIELD: case UDAT_HOUR0_FIELD: // special range check for hours: if (value < 0 || value > 24) { return -start;
}
// fall through to gotNumber check
U_FALLTHROUGH; case UDAT_YEAR_FIELD: case UDAT_YEAR_WOY_FIELD: case UDAT_FRACTIONAL_SECOND_FIELD: // these must be a number if (! gotNumber) { return -start;
}
break;
default: // we check the rest of the fields below. break;
}
// check return position, if it equals -start, then matchString error // special case the return code so we don't necessarily fail out until we // verify no year information also if (ps == -start)
ps--;
return ps;
case UDAT_YEAR_FIELD: // If there are 3 or more YEAR pattern characters, this indicates // that the year value is to be treated literally, without any // two-digit year adjustments (e.g., from "01" to 2001). Otherwise // we made adjustments to place the 2-digit year in the proper // century, for parsed strings from "00" to "99". Any other string // is treated literally: "2250", "-1", "1", "002". if (fDateOverride.compare(hebr)==0 && value < 1000) {
value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
} elseif (text.moveIndex32(start, 2) == pos.getIndex() && !isChineseCalendar
&& u_isdigit(text.char32At(start))
&& u_isdigit(text.char32At(text.moveIndex32(start, 1))))
{ // only adjust year for patterns less than 3. if(count < 3) { // Assume for example that the defaultCenturyStart is 6/18/1903. // This means that two-digit years will be forced into the range // 6/18/1903 to 6/17/2003. As a result, years 00, 01, and 02 // correspond to 2000, 2001, and 2002. Years 04, 05, etc. correspond // to 1904, 1905, etc. If the year is 03, then it is 2003 if the // other fields specify a date before 6/18, or 1903 if they specify a // date afterwards. As a result, 03 is an ambiguous year. All other // two-digit years are unambiguous. if(fHaveDefaultCentury) { // check if this formatter even has a pivot year
int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
ambiguousYear[0] = (value == ambiguousTwoDigitYear);
value += (fDefaultCenturyStartYear/100)*100 +
(value < ambiguousTwoDigitYear ? 100 : 0);
}
}
}
cal.set(UCAL_YEAR, value);
// Delayed checking for adjustment of Hebrew month numbers in non-leap years. if (saveHebrewMonth >= 0) {
HebrewCalendar *hc = (HebrewCalendar*)&cal; if (!hc->isLeapYear(value) && saveHebrewMonth >= 6) {
cal.set(UCAL_MONTH,saveHebrewMonth);
} else {
cal.set(UCAL_MONTH,saveHebrewMonth-1);
}
saveHebrewMonth = -1;
} return pos.getIndex();
case UDAT_YEAR_WOY_FIELD: // Comment is the same as for UDAT_Year_FIELDs - look above if (fDateOverride.compare(hebr)==0 && value < 1000) {
value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
} elseif (text.moveIndex32(start, 2) == pos.getIndex()
&& u_isdigit(text.char32At(start))
&& u_isdigit(text.char32At(text.moveIndex32(start, 1)))
&& fHaveDefaultCentury )
{
int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
ambiguousYear[0] = (value == ambiguousTwoDigitYear);
value += (fDefaultCenturyStartYear/100)*100 +
(value < ambiguousTwoDigitYear ? 100 : 0);
}
cal.set(UCAL_YEAR_WOY, value); return pos.getIndex();
case UDAT_YEAR_NAME_FIELD: if (fSymbols->fShortYearNames != nullptr) {
int32_t newStart = matchString(text, start, UCAL_YEAR, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount, nullptr, cal); if (newStart > 0) { return newStart;
}
} if (gotNumber && (getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC,status) || value > fSymbols->fShortYearNamesCount)) {
cal.set(UCAL_YEAR, value); return pos.getIndex();
} return -start;
case UDAT_MONTH_FIELD: case UDAT_STANDALONE_MONTH_FIELD: if (gotNumber) // i.e., M or MM.
{ // When parsing month numbers from the Hebrew Calendar, we might need to adjust the month depending on whether // or not it was a leap year. We may or may not yet know what year it is, so might have to delay checking until // the year is parsed. if (!strcmp(cal.getType(),"hebrew")) {
HebrewCalendar *hc = (HebrewCalendar*)&cal; if (cal.isSet(UCAL_YEAR)) {
UErrorCode monthStatus = U_ZERO_ERROR; if (!hc->isLeapYear(hc->get(UCAL_YEAR, monthStatus)) && value >= 6) {
cal.set(UCAL_MONTH, value);
} else {
cal.set(UCAL_MONTH, value - 1);
}
} else {
saveHebrewMonth = value;
}
} else { // Don't want to parse the month if it is a string // while pattern uses numeric style: M/MM, L/LL // [We computed 'value' above.]
cal.set(UCAL_MONTH, value - 1);
} return pos.getIndex();
} else { // count >= 3 // i.e., MMM/MMMM, LLL/LLLL // Want to be able to parse both short and long forms. // Try count == 4 first:
UnicodeString * wideMonthPat = nullptr;
UnicodeString * shortMonthPat = nullptr; if (fSymbols->fLeapMonthPatterns != nullptr && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) { if (patternCharIndex==UDAT_MONTH_FIELD) {
wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide];
shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev];
} else {
wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide];
shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev];
}
}
int32_t newStart = 0; if (patternCharIndex==UDAT_MONTH_FIELD) { if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) && count>=3 && count <=4 &&
fSymbols->fLeapMonthPatterns==nullptr && fSymbols->fMonthsCount==fSymbols->fShortMonthsCount) { // single function to check both wide and short, an experiment
newStart = matchAlphaMonthStrings(text, start, fSymbols->fMonths, fSymbols->fShortMonths, fSymbols->fMonthsCount, cal); // try MMMM,MMM if (newStart > 0) { return newStart;
}
} if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
newStart = matchString(text, start, UCAL_MONTH, fSymbols->fMonths, fSymbols->fMonthsCount, wideMonthPat, cal); // try MMMM if (newStart > 0) { return newStart;
}
} if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
newStart = matchString(text, start, UCAL_MONTH, fSymbols->fShortMonths, fSymbols->fShortMonthsCount, shortMonthPat, cal); // try MMM
}
} else { if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) && count>=3 && count <=4 &&
fSymbols->fLeapMonthPatterns==nullptr && fSymbols->fStandaloneMonthsCount==fSymbols->fStandaloneShortMonthsCount) { // single function to check both wide and short, an experiment
newStart = matchAlphaMonthStrings(text, start, fSymbols->fStandaloneMonths, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneMonthsCount, cal); // try MMMM,MMM if (newStart > 0) { return newStart;
}
} if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount, wideMonthPat, cal); // try LLLL if (newStart > 0) { return newStart;
}
} if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount, shortMonthPat, cal); // try LLL
}
} if (newStart > 0 || !getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status)) // currently we do not try to parse MMMMM/LLLLL: #8860 return newStart; // else we allowing parsing as number, below
} break;
case UDAT_HOUR_OF_DAY1_FIELD: // [We computed 'value' above.] if (value == cal.getMaximum(UCAL_HOUR_OF_DAY) + 1)
value = 0;
// fall through to set field
U_FALLTHROUGH; case UDAT_HOUR_OF_DAY0_FIELD:
cal.set(UCAL_HOUR_OF_DAY, value); return pos.getIndex();
case UDAT_FRACTIONAL_SECOND_FIELD: // Fractional seconds left-justify
i = countDigits(text, start, pos.getIndex()); if (i < 3) { while (i < 3) {
value *= 10;
i++;
}
} else {
int32_t a = 1; while (i > 3) {
a *= 10;
i--;
}
value /= a;
}
cal.set(UCAL_MILLISECOND, value); return pos.getIndex();
case UDAT_DOW_LOCAL_FIELD: if (gotNumber) // i.e., e or ee
{ // [We computed 'value' above.]
cal.set(UCAL_DOW_LOCAL, value); return pos.getIndex();
} // else for eee-eeeee fall through to handling of EEE-EEEEE // fall through, do not break here
U_FALLTHROUGH; case UDAT_DAY_OF_WEEK_FIELD:
{ // Want to be able to parse both short and long forms. // Try count == 4 (EEEE) wide first:
int32_t newStart = 0; if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) { if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
fSymbols->fWeekdays, fSymbols->fWeekdaysCount, nullptr, cal)) > 0) return newStart;
} // EEEE wide failed, now try EEE abbreviated if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) { if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
fSymbols->fShortWeekdays, fSymbols->fShortWeekdaysCount, nullptr, cal)) > 0) return newStart;
} // EEE abbreviated failed, now try EEEEEE short if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) { if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
fSymbols->fShorterWeekdays, fSymbols->fShorterWeekdaysCount, nullptr, cal)) > 0) return newStart;
} // EEEEEE short failed, now try EEEEE narrow if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 5) { if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
fSymbols->fNarrowWeekdays, fSymbols->fNarrowWeekdaysCount, nullptr, cal)) > 0) return newStart;
} if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status) || patternCharIndex == UDAT_DAY_OF_WEEK_FIELD) return newStart; // else we allowing parsing as number, below
} break;
case UDAT_STANDALONE_DAY_FIELD:
{ if (gotNumber) // c or cc
{ // [We computed 'value' above.]
cal.set(UCAL_DOW_LOCAL, value); return pos.getIndex();
} // Want to be able to parse both short and long forms. // Try count == 4 (cccc) first:
int32_t newStart = 0; if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) { if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
fSymbols->fStandaloneWeekdays, fSymbols->fStandaloneWeekdaysCount, nullptr, cal)) > 0) return newStart;
} if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) { if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
fSymbols->fStandaloneShortWeekdays, fSymbols->fStandaloneShortWeekdaysCount, nullptr, cal)) > 0) return newStart;
} if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) { if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
fSymbols->fStandaloneShorterWeekdays, fSymbols->fStandaloneShorterWeekdaysCount, nullptr, cal)) > 0) return newStart;
} if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status)) return newStart; // else we allowing parsing as number, below
} break;
case UDAT_AM_PM_FIELD:
{ // optionally try both wide/abbrev and narrow forms
int32_t newStart = 0; // try wide/abbrev if( getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count < 5 ) { if ((newStart = matchString(text, start, UCAL_AM_PM, fSymbols->fAmPms, fSymbols->fAmPmsCount, nullptr, cal)) > 0) { return newStart;
}
} // try narrow if( getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count >= 5 ) { if ((newStart = matchString(text, start, UCAL_AM_PM, fSymbols->fNarrowAmPms, fSymbols->fNarrowAmPmsCount, nullptr, cal)) > 0) { return newStart;
}
} // no matches for given options return -start;
}
case UDAT_HOUR1_FIELD: // [We computed 'value' above.] if (value == cal.getLeastMaximum(UCAL_HOUR)+1)
value = 0;
// fall through to set field
U_FALLTHROUGH; case UDAT_HOUR0_FIELD:
cal.set(UCAL_HOUR, value); return pos.getIndex();
case UDAT_QUARTER_FIELD: if (gotNumber) // i.e., Q or QQ.
{ // Don't want to parse the month if it is a string // while pattern uses numeric style: Q or QQ. // [We computed 'value' above.]
cal.set(UCAL_MONTH, (value - 1) * 3); return pos.getIndex();
} else { // count >= 3 // i.e., QQQ or QQQQ // Want to be able to parse short, long, and narrow forms. // Try count == 4 first:
int32_t newStart = 0;
case UDAT_STANDALONE_QUARTER_FIELD: if (gotNumber) // i.e., q or qq.
{ // Don't want to parse the month if it is a string // while pattern uses numeric style: q or q. // [We computed 'value' above.]
cal.set(UCAL_MONTH, (value - 1) * 3); return pos.getIndex();
} else { // count >= 3 // i.e., qqq or qqqq // Want to be able to parse both short and long forms. // Try count == 4 first:
int32_t newStart = 0;
case UDAT_TIMEZONE_FIELD: // 'z'
{
UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_SPECIFIC_SHORT : UTZFMT_STYLE_SPECIFIC_LONG; const TimeZoneFormat *tzfmt = tzFormat(status); if (U_SUCCESS(status)) {
TimeZone *tz = tzfmt->parse(style, text, pos, tzTimeType); if (tz != nullptr) {
cal.adoptTimeZone(tz); return pos.getIndex();
}
} return -start;
} break; case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
{
UTimeZoneFormatStyle style = (count < 4) ?
UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL : ((count == 5) ? UTZFMT_STYLE_ISO_EXTENDED_FULL: UTZFMT_STYLE_LOCALIZED_GMT); const TimeZoneFormat *tzfmt = tzFormat(status); if (U_SUCCESS(status)) {
TimeZone *tz = tzfmt->parse(style, text, pos, tzTimeType); if (tz != nullptr) {
cal.adoptTimeZone(tz); return pos.getIndex();
}
} return -start;
} case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
{
UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_GENERIC_SHORT : UTZFMT_STYLE_GENERIC_LONG; const TimeZoneFormat *tzfmt = tzFormat(status); if (U_SUCCESS(status)) {
TimeZone *tz = tzfmt->parse(style, text, pos, tzTimeType); if (tz != nullptr) {
cal.adoptTimeZone(tz); return pos.getIndex();
}
} return -start;
} case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
{
UTimeZoneFormatStyle style; switch (count) { case 1:
style = UTZFMT_STYLE_ZONE_ID_SHORT; break; case 2:
style = UTZFMT_STYLE_ZONE_ID; break; case 3:
style = UTZFMT_STYLE_EXEMPLAR_LOCATION; break; default:
style = UTZFMT_STYLE_GENERIC_LOCATION; break;
} const TimeZoneFormat *tzfmt = tzFormat(status); if (U_SUCCESS(status)) {
TimeZone *tz = tzfmt->parse(style, text, pos, tzTimeType); if (tz != nullptr) {
cal.adoptTimeZone(tz); return pos.getIndex();
}
} return -start;
} case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
{
UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_LOCALIZED_GMT_SHORT : UTZFMT_STYLE_LOCALIZED_GMT; const TimeZoneFormat *tzfmt = tzFormat(status); if (U_SUCCESS(status)) {
TimeZone *tz = tzfmt->parse(style, text, pos, tzTimeType); if (tz != nullptr) {
cal.adoptTimeZone(tz); return pos.getIndex();
}
} return -start;
} case UDAT_TIMEZONE_ISO_FIELD: // 'X'
{
UTimeZoneFormatStyle style; switch (count) { case 1:
style = UTZFMT_STYLE_ISO_BASIC_SHORT; break; case 2:
style = UTZFMT_STYLE_ISO_BASIC_FIXED; break; case 3:
style = UTZFMT_STYLE_ISO_EXTENDED_FIXED; break; case 4:
style = UTZFMT_STYLE_ISO_BASIC_FULL; break; default:
style = UTZFMT_STYLE_ISO_EXTENDED_FULL; break;
} const TimeZoneFormat *tzfmt = tzFormat(status); if (U_SUCCESS(status)) {
TimeZone *tz = tzfmt->parse(style, text, pos, tzTimeType); if (tz != nullptr) {
cal.adoptTimeZone(tz); return pos.getIndex();
}
} return -start;
} case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
{
UTimeZoneFormatStyle style; switch (count) { case 1:
style = UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT; break; case 2:
style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED; break; case 3:
style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED; break; case 4:
style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL; break; default:
style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL; break;
} const TimeZoneFormat *tzfmt = tzFormat(status); if (U_SUCCESS(status)) {
TimeZone *tz = tzfmt->parse(style, text, pos, tzTimeType); if (tz != nullptr) {
cal.adoptTimeZone(tz); return pos.getIndex();
}
} return -start;
} // currently no pattern character is defined for UDAT_TIME_SEPARATOR_FIELD // so we should not get here. Leave support in for future definition. case UDAT_TIME_SEPARATOR_FIELD:
{ staticconst char16_t def_sep = DateFormatSymbols::DEFAULT_TIME_SEPARATOR; staticconst char16_t alt_sep = DateFormatSymbols::ALTERNATE_TIME_SEPARATOR;
// Try matching a time separator.
int32_t count_sep = 1;
UnicodeString data[3];
fSymbols->getTimeSeparatorString(data[0]);
// Add the default, if different from the locale. if (data[0].compare(&def_sep, 1) != 0) {
data[count_sep++].setTo(def_sep);
}
// If lenient, add also the alternate, if different from the locale. if (isLenient() && data[0].compare(&alt_sep, 1) != 0) {
data[count_sep++].setTo(alt_sep);
}
return matchString(text, start, UCAL_FIELD_COUNT /* => nothing to set */, data, count_sep, nullptr, cal);
}
default: // Handle "generic" fields // this is now handled below, outside the switch block break;
} // Handle "generic" fields: // switch default case now handled here (outside switch block) to allow // parsing of some string fields as digits for lenient case
// Don't need suffix processing here (as in number processing at the beginning of the function); // the new fields being handled as numeric values (month, weekdays, quarters) should not have suffixes.
if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status)) { // Check the range of the value
int32_t bias = gFieldRangeBias[patternCharIndex]; if (bias >= 0 && (val > cal.getMaximum(field) + bias || val < cal.getMinimum(field) + bias)) { return -start;
}
}
// For the following, need to repeat some of the "if (gotNumber)" code above: // UDAT_[STANDALONE_]MONTH_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_STANDALONE_DAY_FIELD, // UDAT_[STANDALONE_]QUARTER_FIELD switch (patternCharIndex) { case UDAT_MONTH_FIELD: // See notes under UDAT_MONTH_FIELD case above if (!strcmp(cal.getType(),"hebrew")) {
HebrewCalendar *hc = (HebrewCalendar*)&cal; if (cal.isSet(UCAL_YEAR)) {
UErrorCode monthStatus = U_ZERO_ERROR; if (!hc->isLeapYear(hc->get(UCAL_YEAR, monthStatus)) && val >= 6) {
cal.set(UCAL_MONTH, val);
} else {
cal.set(UCAL_MONTH, val - 1);
}
} else {
saveHebrewMonth = val;
}
} else {
cal.set(UCAL_MONTH, val - 1);
} break; case UDAT_STANDALONE_MONTH_FIELD:
cal.set(UCAL_MONTH, val - 1); break; case UDAT_DOW_LOCAL_FIELD: case UDAT_STANDALONE_DAY_FIELD:
cal.set(UCAL_DOW_LOCAL, val); break; case UDAT_QUARTER_FIELD: case UDAT_STANDALONE_QUARTER_FIELD:
cal.set(UCAL_MONTH, (val - 1) * 3); break; case UDAT_RELATED_YEAR_FIELD:
cal.setRelatedYear(val); break; default:
cal.set(field, val); break;
} return pos.getIndex();
} return -start;
}
/** * Parse an integer using fNumberFormat. This method is semantically * const, but actually may modify fNumberFormat.
*/ void SimpleDateFormat::parseInt(const UnicodeString& text,
Formattable& number,
ParsePosition& pos,
UBool allowNegative, const NumberFormat *fmt) const {
parseInt(text, number, -1, pos, allowNegative,fmt);
}
if (maxDigits > 0) { // adjust the result to fit into // the maxDigits and move the position back
int32_t nDigits = pos.getIndex() - oldPos; if (nDigits > maxDigits) {
int32_t val = number.getLong();
nDigits -= maxDigits; while (nDigits > 0) {
val /= 10;
nDigits--;
}
pos.setIndex(oldPos + maxDigits);
number.setLong(val);
}
}
}
void SimpleDateFormat::translatePattern(const UnicodeString& originalPattern,
UnicodeString& translatedPattern, const UnicodeString& from, const UnicodeString& to,
UErrorCode& status)
{ // run through the pattern and convert any pattern symbols from the version // in "from" to the corresponding character in "to". This code takes // quoted strings into account (it doesn't try to translate them), and it signals // an error if a particular "pattern character" doesn't appear in "from". // Depending on the values of "from" and "to" this can convert from generic // to localized patterns or localized to generic. if (U_FAILURE(status)) { return;
}
translatedPattern.remove();
UBool inQuote = false; for (int32_t i = 0; i < originalPattern.length(); ++i) {
char16_t c = originalPattern[i]; if (inQuote) { if (c == QUOTE) {
inQuote = false;
}
} else { if (c == QUOTE) {
inQuote = true;
} elseif (isSyntaxChar(c)) {
int32_t ci = from.indexOf(c); if (ci == -1) {
status = U_INVALID_FORMAT_ERROR; return;
}
c = to[ci];
}
}
translatedPattern += c;
} if (inQuote) {
status = U_INVALID_FORMAT_ERROR; return;
}
}
// Hack to update use of Gannen year numbering for ja@calendar=japanese - // use only if format is non-numeric (includes 年) and no other fDateOverride. if (fCalendar != nullptr && uprv_strcmp(fCalendar->getType(),"japanese") == 0 &&
uprv_strcmp(fLocale.getLanguage(),"ja") == 0) { if (fDateOverride==UnicodeString(u"y=jpanyear") && !fHasHanYearChar) { // Gannen numbering is set but new pattern should not use it, unset; // use procedure from adoptNumberFormat to clear overrides if (fSharedNumberFormatters) {
freeSharedNumberFormatters(fSharedNumberFormatters);
fSharedNumberFormatters = nullptr;
}
fDateOverride.setToBogus(); // record status
} elseif (fDateOverride.isBogus() && fHasHanYearChar) { // No current override (=> no Gannen numbering) but new pattern needs it; // use procedures from initNUmberFormatters / adoptNumberFormat
umtx_lock(&LOCK); if (fSharedNumberFormatters == nullptr) {
fSharedNumberFormatters = allocSharedNumberFormatters();
}
umtx_unlock(&LOCK); if (fSharedNumberFormatters != nullptr) {
Locale ovrLoc(fLocale.getLanguage(),fLocale.getCountry(),fLocale.getVariant(),"numbers=jpanyear");
UErrorCode status = U_ZERO_ERROR; const SharedNumberFormat *snf = createSharedNumberFormat(ovrLoc, status); if (U_SUCCESS(status)) { // Now that we have an appropriate number formatter, fill in the // appropriate slot in the number formatters table.
UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(u'y');
SharedObject::copyPtr(snf, fSharedNumberFormatters[patternCharIndex]);
snf->deleteIfZeroRefCount();
fDateOverride.setTo(u"y=jpanyear", -1); // record status
}
}
}
}
}
//---------------------------------------------------------------------- const TimeZoneFormat*
SimpleDateFormat::getTimeZoneFormat() const { // TimeZoneFormat initialization might fail when out of memory. // If we always initialize TimeZoneFormat instance, we can return // such status there. For now, this implementation lazily instantiates // a TimeZoneFormat for performance optimization reasons, but cannot // propagate such error (probably just out of memory case) to the caller.
UErrorCode status = U_ZERO_ERROR; return (const TimeZoneFormat*)tzFormat(status);
}
// check that we are still in range if ( (start > text.length()) ||
(start < 0) ||
(patLoc < 0) ||
(patLoc > fPattern.length())) { // out of range, don't advance location in text return start;
}
// get the suffix
DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(fNumberFormat); if (decfmt != nullptr) { if (isNegative) {
suf = decfmt->getNegativeSuffix(suf);
} else {
suf = decfmt->getPositiveSuffix(suf);
}
}
// check for suffix if (suf.length() <= 0) { return start;
}
// check suffix will be encountered in the pattern
patternMatch = compareSimpleAffix(suf,fPattern,patLoc);
// check if a suffix will be encountered in the text
textPreMatch = compareSimpleAffix(suf,text,start);
// check if a suffix was encountered in the text
textPostMatch = compareSimpleAffix(suf,text,start-suf.length());
int32_t
SimpleDateFormat::compareSimpleAffix(const UnicodeString& affix, const UnicodeString& input,
int32_t pos) const {
int32_t start = pos; for (int32_t i=0; i<affix.length(); ) {
UChar32 c = affix.char32At(i);
int32_t len = U16_LENGTH(c); if (PatternProps::isWhiteSpace(c)) { // We may have a pattern like: \u200F \u0020 // and input text like: \u200F \u0020 // Note that U+200F and U+0020 are Pattern_White_Space but only // U+0020 is UWhiteSpace. So we have to first do a direct // match of the run of Pattern_White_Space in the pattern, // then match any extra characters.
UBool literalMatch = false; while (pos < input.length() &&
input.char32At(pos) == c) {
literalMatch = true;
i += len;
pos += len; if (i == affix.length()) { break;
}
c = affix.char32At(i);
len = U16_LENGTH(c); if (!PatternProps::isWhiteSpace(c)) { break;
}
}
// Advance over run in pattern
i = skipPatternWhiteSpace(affix, i);
// Advance over run in input text // Must see at least one white space char in input, // unless we've already matched some characters literally.
int32_t s = pos;
pos = skipUWhiteSpace(input, pos); if (pos == s && !literalMatch) { return -1;
}
// If we skip UWhiteSpace in the input text, we need to skip it in the pattern. // Otherwise, the previous lines may have skipped over text (such as U+00A0) that // is also in the affix.
i = skipUWhiteSpace(affix, i);
} else { if (pos < input.length() &&
input.char32At(pos) == c) {
i += len;
pos += len;
} else { return -1;
}
}
} return pos - start;
}
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.
