Google Git
Sign in
skia / external / github.com / unicode-org / icu / aa9f5bc2799cc562db159da0d4468440362156da / . / icu4c / source / i18n / smpdtfmt.cpp
blob: a3ec7cb026591c72f45a804a975496c4f2f2f046 [file] [log] [blame]
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
* Copyright (C) 1997-2016, International Business Machines Corporation and *
* others. All Rights Reserved. *
*******************************************************************************
*
* File SMPDTFMT.CPP
*
* Modification History:
*
* Date Name Description
* 02/19/97 aliu Converted from java.
* 03/31/97 aliu Modified extensively to work with 50 locales.
* 04/01/97 aliu Added support for centuries.
* 07/09/97 helena Made ParsePosition into a class.
* 07/21/98 stephen Added initializeDefaultCentury.
* Removed getZoneIndex (added in DateFormatSymbols)
* Removed subParseLong
* Removed chk
* 02/22/99 stephen Removed character literals for EBCDIC safety
* 10/14/99 aliu Updated 2-digit year parsing so that only "00" thru
* "99" are recognized. {j28 4182066}
* 11/15/99 weiv Added support for week of year/day of week format
********************************************************************************
*/
#define ZID_KEY_MAX 128
#include "unicode/utypes.h"
#if !UCONFIG_NO_FORMATTING
#include "unicode/smpdtfmt.h"
#include "unicode/dtfmtsym.h"
#include "unicode/ures.h"
#include "unicode/msgfmt.h"
#include "unicode/calendar.h"
#include "unicode/gregocal.h"
#include "unicode/timezone.h"
#include "unicode/decimfmt.h"
#include "unicode/dcfmtsym.h"
#include "unicode/uchar.h"
#include "unicode/uniset.h"
#include "unicode/ustring.h"
#include "unicode/basictz.h"
#include "unicode/simpleformatter.h"
#include "unicode/simpletz.h"
#include "unicode/rbtz.h"
#include "unicode/tzfmt.h"
#include "unicode/ucasemap.h"
#include "unicode/utf16.h"
#include "unicode/vtzone.h"
#include "unicode/udisplaycontext.h"
#include "unicode/brkiter.h"
#include "unicode/rbnf.h"
#include "unicode/dtptngen.h"
#include "uresimp.h"
#include "olsontz.h"
#include "patternprops.h"
#include "fphdlimp.h"
#include "hebrwcal.h"
#include "cstring.h"
#include "uassert.h"
#include "cmemory.h"
#include "umutex.h"
#include <float.h>
#include "smpdtfst.h"
#include "sharednumberformat.h"
#include "ucasemap_imp.h"
#include "ustr_imp.h"
#include "charstr.h"
#include "uvector.h"
#include "cstr.h"
#include "dayperiodrules.h"
#include "tznames_impl.h" // ZONE_NAME_U16_MAX
#include "number_utypes.h"
#if defined( U_DEBUG_CALSVC ) || defined (U_DEBUG_CAL)
#include <stdio.h>
#endif
// *****************************************************************************
// class SimpleDateFormat
// *****************************************************************************
U_NAMESPACE_BEGIN
/**
* Last-resort string to use for "GMT" when constructing time zone strings.
*/
// For time zones that have no names, use strings GMT+minutes and
// GMT-minutes. For instance, in France the time zone is GMT+60.
// Also accepted are GMT+H:MM or GMT-H:MM.
// Currently not being used
//static const UChar gGmt[] = {0x0047, 0x004D, 0x0054, 0x0000}; // "GMT"
//static const UChar gGmtPlus[] = {0x0047, 0x004D, 0x0054, 0x002B, 0x0000}; // "GMT+"
//static const UChar gGmtMinus[] = {0x0047, 0x004D, 0x0054, 0x002D, 0x0000}; // "GMT-"
//static const UChar gDefGmtPat[] = {0x0047, 0x004D, 0x0054, 0x007B, 0x0030, 0x007D, 0x0000}; /* GMT{0} */
//static const UChar gDefGmtNegHmsPat[] = {0x002D, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x003A, 0x0073, 0x0073, 0x0000}; /* -HH:mm:ss */
//static const UChar gDefGmtNegHmPat[] = {0x002D, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x0000}; /* -HH:mm */
//static const UChar gDefGmtPosHmsPat[] = {0x002B, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x003A, 0x0073, 0x0073, 0x0000}; /* +HH:mm:ss */
//static const UChar gDefGmtPosHmPat[] = {0x002B, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x0000}; /* +HH:mm */
//static const UChar gUt[] = {0x0055, 0x0054, 0x0000}; // "UT"
//static const UChar gUtc[] = {0x0055, 0x0054, 0x0043, 0x0000}; // "UT"
typedef enum GmtPatSize {
kGmtLen = 3,
kGmtPatLen = 6,
kNegHmsLen = 9,
kNegHmLen = 6,
kPosHmsLen = 9,
kPosHmLen = 6,
kUtLen = 2,
kUtcLen = 3
} GmtPatSize;
// Stuff needed for numbering system overrides
typedef enum OvrStrType {
kOvrStrDate = 0,
kOvrStrTime = 1,
kOvrStrBoth = 2
} OvrStrType;
static const UDateFormatField kDateFields[] = {
UDAT_YEAR_FIELD,
UDAT_MONTH_FIELD,
UDAT_DATE_FIELD,
UDAT_DAY_OF_YEAR_FIELD,
UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
UDAT_WEEK_OF_YEAR_FIELD,
UDAT_WEEK_OF_MONTH_FIELD,
UDAT_YEAR_WOY_FIELD,
UDAT_EXTENDED_YEAR_FIELD,
UDAT_JULIAN_DAY_FIELD,
UDAT_STANDALONE_DAY_FIELD,
UDAT_STANDALONE_MONTH_FIELD,
UDAT_QUARTER_FIELD,
UDAT_STANDALONE_QUARTER_FIELD,
UDAT_YEAR_NAME_FIELD,
UDAT_RELATED_YEAR_FIELD };
static const int8_t kDateFieldsCount = 16;
static const UDateFormatField kTimeFields[] = {
UDAT_HOUR_OF_DAY1_FIELD,
UDAT_HOUR_OF_DAY0_FIELD,
UDAT_MINUTE_FIELD,
UDAT_SECOND_FIELD,
UDAT_FRACTIONAL_SECOND_FIELD,
UDAT_HOUR1_FIELD,
UDAT_HOUR0_FIELD,
UDAT_MILLISECONDS_IN_DAY_FIELD,
UDAT_TIMEZONE_RFC_FIELD,
UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD };
static const int8_t kTimeFieldsCount = 10;
// This is a pattern-of-last-resort used when we can't load a usable pattern out
// of a resource.
static const UChar gDefaultPattern[] =
{
0x79, 0x79, 0x79, 0x79, 0x4D, 0x4D, 0x64, 0x64, 0x20, 0x68, 0x68, 0x3A, 0x6D, 0x6D, 0x20, 0x61, 0
}; /* "yyyyMMdd 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).
static const UChar SUPPRESS_NEGATIVE_PREFIX[] = {0xAB00, 0};
/**
* These are the tags we expect to see in normal resource bundle files associated
* with a locale.
*/
static const UChar 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.
*/
static const 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 millenium down to 1-999
static const int32_t HEBREW_CAL_CUR_MILLENIUM_START_YEAR = 5000;
static const 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.
*/
static const double MAX_DAYLIGHT_DETECTION_RANGE = 30*365*24*60*60*1000.0;
static UMutex LOCK;
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)
SimpleDateFormat::NSOverride::~NSOverride() {
if (snf != NULL) {
snf->removeRef();
}
}
void SimpleDateFormat::NSOverride::free() {
NSOverride *cur = this;
while (cur) {
NSOverride *next_temp = cur->next;
delete cur;
cur = next_temp;
}
}
// 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
static void fixNumberFormatForDates(NumberFormat &nf) {
nf.setGroupingUsed(FALSE);
DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(&nf);
if (decfmt != NULL) {
decfmt->setDecimalSeparatorAlwaysShown(FALSE);
}
nf.setParseIntegerOnly(TRUE);
nf.setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
}
static const SharedNumberFormat *createSharedNumberFormat(
NumberFormat *nfToAdopt) {
fixNumberFormatForDates(*nfToAdopt);
const SharedNumberFormat *result = new SharedNumberFormat(nfToAdopt);
if (result == NULL) {
delete nfToAdopt;
}
return result;
}
static const SharedNumberFormat *createSharedNumberFormat(
const Locale &loc, UErrorCode &status) {
NumberFormat *nf = NumberFormat::createInstance(loc, status);
if (U_FAILURE(status)) {
return NULL;
}
const SharedNumberFormat *result = createSharedNumberFormat(nf);
if (result == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
}
return result;
}
static const SharedNumberFormat **allocSharedNumberFormatters() {
const SharedNumberFormat **result = (const SharedNumberFormat**)
uprv_malloc(UDAT_FIELD_COUNT * sizeof(const SharedNumberFormat*));
if (result == NULL) {
return NULL;
}
for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
result[i] = NULL;
}
return result;
}
static void freeSharedNumberFormatters(const SharedNumberFormat ** list) {
for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
SharedObject::clearPtr(list[i]);
}
uprv_free(list);
}
const NumberFormat *SimpleDateFormat::getNumberFormatByIndex(
UDateFormatField index) const {
if (fSharedNumberFormatters == NULL ||
fSharedNumberFormatters[index] == NULL) {
return fNumberFormat;
}
return &(**fSharedNumberFormatters[index]);
}
//----------------------------------------------------------------------
SimpleDateFormat::~SimpleDateFormat()
{
delete fSymbols;
if (fSharedNumberFormatters) {
freeSharedNumberFormatters(fSharedNumberFormatters);
}
if (fTimeZoneFormat) {
delete fTimeZoneFormat;
}
freeFastNumberFormatters();
#if !UCONFIG_NO_BREAK_ITERATION
delete fCapitalizationBrkIter;
#endif
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(UErrorCode& status)
: fLocale(Locale::getDefault()),
fSymbols(NULL),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
initializeBooleanAttributes();
construct(kShort, (EStyle) (kShort + kDateOffset), fLocale, status);
initializeDefaultCentury();
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
UErrorCode &status)
: fPattern(pattern),
fLocale(Locale::getDefault()),
fSymbols(NULL),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
fDateOverride.setToBogus();
fTimeOverride.setToBogus();
initializeBooleanAttributes();
initializeCalendar(NULL,fLocale,status);
fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
initialize(fLocale, status);
initializeDefaultCentury();
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
const UnicodeString& override,
UErrorCode &status)
: fPattern(pattern),
fLocale(Locale::getDefault()),
fSymbols(NULL),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
fDateOverride.setTo(override);
fTimeOverride.setToBogus();
initializeBooleanAttributes();
initializeCalendar(NULL,fLocale,status);
fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
initialize(fLocale, status);
initializeDefaultCentury();
processOverrideString(fLocale,override,kOvrStrBoth,status);
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
const Locale& locale,
UErrorCode& status)
: fPattern(pattern),
fLocale(locale),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
fDateOverride.setToBogus();
fTimeOverride.setToBogus();
initializeBooleanAttributes();
initializeCalendar(NULL,fLocale,status);
fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
initialize(fLocale, status);
initializeDefaultCentury();
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
const UnicodeString& override,
const Locale& locale,
UErrorCode& status)
: fPattern(pattern),
fLocale(locale),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
fDateOverride.setTo(override);
fTimeOverride.setToBogus();
initializeBooleanAttributes();
initializeCalendar(NULL,fLocale,status);
fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
initialize(fLocale, status);
initializeDefaultCentury();
processOverrideString(locale,override,kOvrStrBoth,status);
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
DateFormatSymbols* symbolsToAdopt,
UErrorCode& status)
: fPattern(pattern),
fLocale(Locale::getDefault()),
fSymbols(symbolsToAdopt),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
fDateOverride.setToBogus();
fTimeOverride.setToBogus();
initializeBooleanAttributes();
initializeCalendar(NULL,fLocale,status);
initialize(fLocale, status);
initializeDefaultCentury();
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
const DateFormatSymbols& symbols,
UErrorCode& status)
: fPattern(pattern),
fLocale(Locale::getDefault()),
fSymbols(new DateFormatSymbols(symbols)),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
fDateOverride.setToBogus();
fTimeOverride.setToBogus();
initializeBooleanAttributes();
initializeCalendar(NULL, fLocale, status);
initialize(fLocale, status);
initializeDefaultCentury();
}
//----------------------------------------------------------------------
// Not for public consumption; used by DateFormat
SimpleDateFormat::SimpleDateFormat(EStyle timeStyle,
EStyle dateStyle,
const Locale& locale,
UErrorCode& status)
: fLocale(locale),
fSymbols(NULL),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
initializeBooleanAttributes();
construct(timeStyle, dateStyle, fLocale, status);
if(U_SUCCESS(status)) {
initializeDefaultCentury();
}
}
//----------------------------------------------------------------------
/**
* 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),
fSymbols(NULL),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
if (U_FAILURE(status)) return;
initializeBooleanAttributes();
initializeCalendar(NULL, 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 NULL */
if (fSymbols == 0) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
}
fDateOverride.setToBogus();
fTimeOverride.setToBogus();
initialize(fLocale, status);
if(U_SUCCESS(status)) {
initializeDefaultCentury();
}
}
//----------------------------------------------------------------------
SimpleDateFormat::SimpleDateFormat(const SimpleDateFormat& other)
: DateFormat(other),
fLocale(other.fLocale),
fSymbols(NULL),
fTimeZoneFormat(NULL),
fSharedNumberFormatters(NULL),
fCapitalizationBrkIter(NULL)
{
initializeBooleanAttributes();
*this = other;
}
//----------------------------------------------------------------------
SimpleDateFormat& SimpleDateFormat::operator=(const SimpleDateFormat& other)
{
if (this == &other) {
return *this;
}
DateFormat::operator=(other);
fDateOverride = other.fDateOverride;
fTimeOverride = other.fTimeOverride;
delete fSymbols;
fSymbols = NULL;
if (other.fSymbols)
fSymbols = new DateFormatSymbols(*other.fSymbols);
fDefaultCenturyStart = other.fDefaultCenturyStart;
fDefaultCenturyStartYear = other.fDefaultCenturyStartYear;
fHaveDefaultCentury = other.fHaveDefaultCentury;
fPattern = other.fPattern;
fHasMinute = other.fHasMinute;
fHasSecond = other.fHasSecond;
fLocale = other.fLocale;
// TimeZoneFormat can now be set independently via setter.
// If it is NULL, it will be lazily initialized from locale
delete fTimeZoneFormat;
fTimeZoneFormat = NULL;
if (other.fTimeZoneFormat) {
fTimeZoneFormat = new TimeZoneFormat(*other.fTimeZoneFormat);
}
#if !UCONFIG_NO_BREAK_ITERATION
if (other.fCapitalizationBrkIter != NULL) {
fCapitalizationBrkIter = (other.fCapitalizationBrkIter)->clone();
}
#endif
if (fSharedNumberFormatters != NULL) {
freeSharedNumberFormatters(fSharedNumberFormatters);
fSharedNumberFormatters = NULL;
}
if (other.fSharedNumberFormatters != NULL) {
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;
freeFastNumberFormatters();
initFastNumberFormatters(localStatus);
return *this;
}
//----------------------------------------------------------------------
SimpleDateFormat*
SimpleDateFormat::clone() const
{
return new SimpleDateFormat(*this);
}
//----------------------------------------------------------------------
UBool
SimpleDateFormat::operator==(const Format& other) const
{
if (DateFormat::operator==(other)) {
// The DateFormat::operator== check for fCapitalizationContext equality above
// is sufficient to check equality of all derived context-related data.
// DateFormat::operator== guarantees following cast is safe
SimpleDateFormat* that = (SimpleDateFormat*)&other;
return (fPattern == that->fPattern &&
fSymbols != NULL && // Check for pathological object
that->fSymbols != NULL && // Check for pathological object
*fSymbols == *that->fSymbols &&
fHaveDefaultCentury == that->fHaveDefaultCentury &&
fDefaultCenturyStart == that->fDefaultCenturyStart);
}
return FALSE;
}
//----------------------------------------------------------------------
static const UChar* timeSkeletons[4] = {
u"jmmsszzzz", // kFull
u"jmmssz", // kLong
u"jmmss", // kMedium
u"jmm", // kShort
};
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(NULL, locale, status);
if (U_FAILURE(status)) return;
// Load date time patterns directly from resources.
const char* cType = fCalendar ? fCalendar->getType() : NULL;
LocalUResourceBundlePointer bundle(ures_open(NULL, locale.getBaseName(), &status));
if (U_FAILURE(status)) return;
UBool cTypeIsGregorian = TRUE;
LocalUResourceBundlePointer dateTimePatterns;
if (cType != NULL && uprv_strcmp(cType, "gregorian") != 0) {
CharString resourcePath("calendar/", status);
resourcePath.append(cType, status).append("/DateTimePatterns", status);
dateTimePatterns.adoptInstead(
ures_getByKeyWithFallback(bundle.getAlias(), resourcePath.data(),
(UResourceBundle*)NULL, &status));
cTypeIsGregorian = FALSE;
}
// 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*)NULL, &status));
}
if (U_FAILURE(status)) return;
LocalUResourceBundlePointer currentBundle;
if (ures_getSize(dateTimePatterns.getAlias()) <= kDateTime)
{
status = U_INVALID_FORMAT_ERROR;
return;
}
setLocaleIDs(ures_getLocaleByType(dateTimePatterns.getAlias(), ULOC_VALID_LOCALE, &status),
ures_getLocaleByType(dateTimePatterns.getAlias(), ULOC_ACTUAL_LOCALE, &status));
// create a symbols object from the locale
fSymbols = DateFormatSymbols::createForLocale(locale, status);
if (U_FAILURE(status)) return;
/* test for NULL */
if (fSymbols == 0) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
const UChar *resStr,*ovrStr;
int32_t resStrLen,ovrStrLen = 0;
fDateOverride.setToBogus();
fTimeOverride.setToBogus();
UnicodeString timePattern;
if (timeStyle >= kFull && timeStyle <= kShort) {
const char* baseLocID = locale.getBaseName();
if (baseLocID[0]!=0 && uprv_strcmp(baseLocID,"und")!=0) {
UErrorCode useStatus = U_ZERO_ERROR;
Locale baseLoc(baseLocID);
Locale validLoc(getLocale(ULOC_VALID_LOCALE, useStatus));
if (U_SUCCESS(useStatus) && validLoc!=baseLoc) {
bool useDTPG = false;
const char* baseReg = baseLoc.getCountry(); // empty string if no region
if ((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
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(), (int32_t)timeStyle, NULL, &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;
}
}
tempus1.setTo(TRUE, resStr, resStrLen);
}
currentBundle.adoptInstead(
ures_getByIndex(dateTimePatterns.getAlias(), (int32_t)dateStyle, NULL, &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);
fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
break;
}
default: {
status = U_INVALID_FORMAT_ERROR;
return;
}
}
UnicodeString tempus2(TRUE, resStr, resStrLen);
int32_t glueIndex = kDateTime;
int32_t patternsSize = ures_getSize(dateTimePatterns.getAlias());
if (patternsSize >= (kDateTimeOffset + kShort + 1)) {
// Get proper date time format
glueIndex = (int32_t)(kDateTimeOffset + (dateStyle - kDateOffset));
}
resStr = ures_getStringByIndex(dateTimePatterns.getAlias(), glueIndex, &resStrLen, &status);
SimpleFormatter(UnicodeString(TRUE, resStr, resStrLen), 2, 2, status).
format(tempus1, tempus2, fPattern, status);
}
// if the pattern includes just time data or just date date, load the appropriate
// pattern string from the resources
// setTo() - see DateFormatSymbols::assignArray comments
else if (timeStyle != kNone) {
fPattern.setTo(timePattern);
if (fPattern.length() == 0) {
currentBundle.adoptInstead(
ures_getByIndex(dateTimePatterns.getAlias(), (int32_t)timeStyle, NULL, &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);
fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
break;
}
default: {
status = U_INVALID_FORMAT_ERROR;
return;
}
}
fPattern.setTo(TRUE, resStr, resStrLen);
}
}
else if (dateStyle != kNone) {
currentBundle.adoptInstead(
ures_getByIndex(dateTimePatterns.getAlias(), (int32_t)dateStyle, NULL, &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);
fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
break;
}
default: {
status = U_INVALID_FORMAT_ERROR;
return;
}
}
fPattern.setTo(TRUE, resStr, resStrLen);
}
// and if it includes _neither_, that's an error
else
status = U_INVALID_FORMAT_ERROR;
// finally, finish initializing by creating a Calendar and a NumberFormat
initialize(locale, status);
}
//----------------------------------------------------------------------
Calendar*
SimpleDateFormat::initializeCalendar(TimeZone* adoptZone, const Locale& locale, UErrorCode& status)
{
if(!U_FAILURE(status)) {
fCalendar = Calendar::createInstance(
adoptZone ? adoptZone : TimeZone::forLocaleOrDefault(locale), locale, status);
}
return fCalendar;
}
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 != NULL && U_SUCCESS(status))
{
fixNumberFormatForDates(*fNumberFormat);
//fNumberFormat->setLenient(TRUE); // Java uses a custom DateNumberFormat to format/parse
initNumberFormatters(locale, status);
initFastNumberFormatters(status);
}
else if (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;
setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status);
setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
setBooleanAttribute(UDAT_PARSE_PARTIAL_LITERAL_MATCH, true, status);
setBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, true, status);
}
/* 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;
}
fCalendar->setTime(startDate, status);
if(U_SUCCESS(status)) {
fHaveDefaultCentury = TRUE;
fDefaultCenturyStart = startDate;
fDefaultCenturyStartYear = fCalendar->get(UCAL_YEAR, status);
}
}
//----------------------------------------------------------------------
UnicodeString&
SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo, FieldPosition& pos) const
{
UErrorCode status = U_ZERO_ERROR;
FieldPositionOnlyHandler handler(pos);
return _format(cal, appendTo, handler, status);
}
//----------------------------------------------------------------------
UnicodeString&
SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo,
FieldPositionIterator* posIter, UErrorCode& status) const
{
FieldPositionIteratorHandler handler(posIter, status);
return _format(cal, appendTo, handler, status);
}
//----------------------------------------------------------------------
UnicodeString&
SimpleDateFormat::_format(Calendar& cal, UnicodeString& appendTo,
FieldPositionHandler& handler, UErrorCode& status) const
{
if ( U_FAILURE(status) ) {
return appendTo;
}
Calendar* workCal = &cal;
Calendar* calClone = NULL;
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 != NULL) {
UDate t = cal.getTime(status);
calClone->setTime(t, status);
calClone->setTimeZone(cal.getTimeZone());
workCal = calClone;
} else {
status = U_MEMORY_ALLOCATION_ERROR;
return appendTo;
}
}
UBool inQuote = FALSE;
UChar prevCh = 0;
int32_t count = 0;
int32_t fieldNum = 0;
UDisplayContext capitalizationContext = getContext(UDISPCTX_TYPE_CAPITALIZATION, status);
// loop through the pattern string character by character
for (int32_t i = 0; i < fPattern.length() && U_SUCCESS(status); ++i) {
UChar 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) < fPattern.length() && fPattern[i+1] == QUOTE) {
appendTo += (UChar)QUOTE;
++i;
} else {
inQuote = ! inQuote;
}
}
else if (!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);
}
if (calClone != NULL) {
delete calClone;
}
return appendTo;
}
//----------------------------------------------------------------------
/* 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(UChar 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.
static const 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
};
return ch < UPRV_LENGTHOF(mapCharToLevel) ? mapCharToLevel[ch] : -1;
}
UBool SimpleDateFormat::isSyntaxChar(UChar ch) {
static const UBool mapCharToIsSyntax[] = {
//
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
//
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
//
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
//
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
// ! " # $ % & '
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
// ( ) * + , - . /
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
// 0 1 2 3 4 5 6 7
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
#if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
// 8 9 : ; < = > ?
FALSE, FALSE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE,
#else
// 8 9 : ; < = > ?
FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
#endif
// @ A B C D E F G
FALSE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
// H I J K L M N O
TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
// P Q R S T U V W
TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
// X Y Z [ \ ] ^ _
TRUE, TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE,
// ` a b c d e f g
FALSE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
// h i j k l m n o
TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
// p q r s t u v w
TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
// x y z { | } ~
TRUE, TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE
};
return ch < UPRV_LENGTHOF(mapCharToIsSyntax) ? mapCharToIsSyntax[ch] : FALSE;
}
// Map index into pattern character string to Calendar field number.
const UCalendarDateFields
SimpleDateFormat::fgPatternIndexToCalendarField[] =
{
/*GyM*/ UCAL_ERA, UCAL_YEAR, UCAL_MONTH,
/*dkH*/ UCAL_DATE, UCAL_HOUR_OF_DAY, UCAL_HOUR_OF_DAY,
/*msS*/ UCAL_MINUTE, UCAL_SECOND, UCAL_MILLISECOND,
/*EDF*/ UCAL_DAY_OF_WEEK, UCAL_DAY_OF_YEAR, UCAL_DAY_OF_WEEK_IN_MONTH,
/*wWa*/ UCAL_WEEK_OF_YEAR, UCAL_WEEK_OF_MONTH, UCAL_AM_PM,
/*hKz*/ UCAL_HOUR, UCAL_HOUR, UCAL_ZONE_OFFSET,
/*Yeu*/ UCAL_YEAR_WOY, UCAL_DOW_LOCAL, UCAL_EXTENDED_YEAR,
/*gAZ*/ UCAL_JULIAN_DAY, UCAL_MILLISECONDS_IN_DAY, UCAL_ZONE_OFFSET,
/*v*/ UCAL_ZONE_OFFSET,
/*c*/ UCAL_DOW_LOCAL,
/*L*/ UCAL_MONTH,
/*Q*/ UCAL_MONTH,
/*q*/ UCAL_MONTH,
/*V*/ UCAL_ZONE_OFFSET,
/*U*/ UCAL_YEAR,
/*O*/ UCAL_ZONE_OFFSET,
/*Xx*/ UCAL_ZONE_OFFSET, UCAL_ZONE_OFFSET,
/*r*/ UCAL_EXTENDED_YEAR,
/*bB*/ UCAL_FIELD_COUNT, UCAL_FIELD_COUNT, // no mappings to calendar fields
#if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
/*:*/ UCAL_FIELD_COUNT, /* => no useful mapping to any calendar field */
#else
/*no pattern char for UDAT_TIME_SEPARATOR_FIELD*/ UCAL_FIELD_COUNT, /* => no useful mapping to any calendar field */
#endif
};
// Map index into pattern character string to DateFormat field number
const UDateFormatField
SimpleDateFormat::fgPatternIndexToDateFormatField[] = {
/*GyM*/ UDAT_ERA_FIELD, UDAT_YEAR_FIELD, UDAT_MONTH_FIELD,
/*dkH*/ UDAT_DATE_FIELD, UDAT_HOUR_OF_DAY1_FIELD, UDAT_HOUR_OF_DAY0_FIELD,
/*msS*/ UDAT_MINUTE_FIELD, UDAT_SECOND_FIELD, UDAT_FRACTIONAL_SECOND_FIELD,
/*EDF*/ UDAT_DAY_OF_WEEK_FIELD, UDAT_DAY_OF_YEAR_FIELD, UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
/*wWa*/ UDAT_WEEK_OF_YEAR_FIELD, UDAT_WEEK_OF_MONTH_FIELD, UDAT_AM_PM_FIELD,
/*hKz*/ UDAT_HOUR1_FIELD, UDAT_HOUR0_FIELD, UDAT_TIMEZONE_FIELD,
/*Yeu*/ UDAT_YEAR_WOY_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_EXTENDED_YEAR_FIELD,
/*gAZ*/ UDAT_JULIAN_DAY_FIELD, UDAT_MILLISECONDS_IN_DAY_FIELD, UDAT_TIMEZONE_RFC_FIELD,
/*v*/ UDAT_TIMEZONE_GENERIC_FIELD,
/*c*/ UDAT_STANDALONE_DAY_FIELD,
/*L*/ UDAT_STANDALONE_MONTH_FIELD,
/*Q*/ UDAT_QUARTER_FIELD,
/*q*/ UDAT_STANDALONE_QUARTER_FIELD,
/*V*/ UDAT_TIMEZONE_SPECIAL_FIELD,
/*U*/ UDAT_YEAR_NAME_FIELD,
/*O*/ UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD,
/*Xx*/ UDAT_TIMEZONE_ISO_FIELD, UDAT_TIMEZONE_ISO_LOCAL_FIELD,
/*r*/ UDAT_RELATED_YEAR_FIELD,
/*bB*/ UDAT_AM_PM_MIDNIGHT_NOON_FIELD, UDAT_FLEXIBLE_DAY_PERIOD_FIELD,
#if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
/*:*/ UDAT_TIME_SEPARATOR_FIELD,
#else
/*no pattern char for UDAT_TIME_SEPARATOR_FIELD*/ UDAT_TIME_SEPARATOR_FIELD,
#endif
};
//----------------------------------------------------------------------
/**
* Append symbols[value] to dst. Make sure the array index is not out
* of bounds.
*/
static inline void
_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];
}
}
static inline void
_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 == NULL) {
dst += symbols[value];
} else {
SimpleFormatter(*monthPattern, 1, 1, status).format(symbols[value], dst, status);
}
}
}
//----------------------------------------------------------------------
static number::LocalizedNumberFormatter*
createFastFormatter(const DecimalFormat* df, int32_t minInt, int32_t maxInt, UErrorCode& status) {
const number::LocalizedNumberFormatter* lnfBase = df->toNumberFormatter(status);
if (U_FAILURE(status)) {
return nullptr;
}
return lnfBase->integerWidth(
number::IntegerWidth::zeroFillTo(minInt).truncateAt(maxInt)
).clone().orphan();
}
void SimpleDateFormat::initFastNumberFormatters(UErrorCode& status) {
if (U_FAILURE(status)) {
return;
}
auto* df = dynamic_cast<const DecimalFormat*>(fNumberFormat);
if (df == nullptr) {
return;
}
fFastNumberFormatters[SMPDTFMT_NF_1x10] = createFastFormatter(df, 1, 10, status);
fFastNumberFormatters[SMPDTFMT_NF_2x10] = createFastFormatter(df, 2, 10, status);
fFastNumberFormatters[SMPDTFMT_NF_3x10] = createFastFormatter(df, 3, 10, status);
fFastNumberFormatters[SMPDTFMT_NF_4x10] = createFastFormatter(df, 4, 10, status);
fFastNumberFormatters[SMPDTFMT_NF_2x2] = createFastFormatter(df, 2, 2, status);
}
void SimpleDateFormat::freeFastNumberFormatters() {
delete fFastNumberFormatters[SMPDTFMT_NF_1x10];
delete fFastNumberFormatters[SMPDTFMT_NF_2x10];
delete fFastNumberFormatters[SMPDTFMT_NF_3x10];
delete fFastNumberFormatters[SMPDTFMT_NF_4x10];
delete fFastNumberFormatters[SMPDTFMT_NF_2x2];
fFastNumberFormatters[SMPDTFMT_NF_1x10] = nullptr;
fFastNumberFormatters[SMPDTFMT_NF_2x10] = nullptr;
fFastNumberFormatters[SMPDTFMT_NF_3x10] = nullptr;
fFastNumberFormatters[SMPDTFMT_NF_4x10] = nullptr;
fFastNumberFormatters[SMPDTFMT_NF_2x2] = nullptr;
}
void
SimpleDateFormat::initNumberFormatters(const Locale &locale,UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
if ( fDateOverride.isBogus() && fTimeOverride.isBogus() ) {
return;
}
umtx_lock(&LOCK);
if (fSharedNumberFormatters == NULL) {
fSharedNumberFormatters = allocSharedNumberFormatters();
if (fSharedNumberFormatters == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
}
}
umtx_unlock(&LOCK);
if (U_FAILURE(status)) {
return;
}
processOverrideString(locale,fDateOverride,kOvrStrDate,status);
processOverrideString(locale,fTimeOverride,kOvrStrTime,status);
}
void
SimpleDateFormat::processOverrideString(const Locale &locale, const UnicodeString &str, int8_t type, UErrorCode &status) {
if (str.isBogus() || U_FAILURE(status)) {
return;
}
int32_t start = 0;
int32_t len;
UnicodeString nsName;
UnicodeString ovrField;
UBool moreToProcess = TRUE;
NSOverride *overrideList = NULL;
while (moreToProcess) {
int32_t delimiterPosition = str.indexOf((UChar)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((UChar)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 = NULL;
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);
Locale ovrLoc(locale.getLanguage(),locale.getCountry(),locale.getVariant(),kw);
cur->hash = nsNameHash;
cur->next = overrideList;
SharedObject::copyPtr(
createSharedNumberFormat(ovrLoc, status), cur->snf);
if (U_FAILURE(status)) {
if (overrideList) {
overrideList->free();
}
return;
}
snf = cur->snf;
overrideList = cur.orphan();
} else {
status = U_MEMORY_ALLOCATION_ERROR;
if (overrideList) {
overrideList->free();
}
return;
}
}
// 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]);
}
start = delimiterPosition + 1;
}
if (overrideList) {
overrideList->free();
}
}
//---------------------------------------------------------------------
void
SimpleDateFormat::subFormat(UnicodeString &appendTo,
char16_t ch,
int32_t count,
UDisplayContext capitalizationContext,
int32_t fieldNum,
char16_t fieldToOutput,
FieldPositionHandler& handler,
Calendar& cal,
UErrorCode& status) const
{
if (U_FAILURE(status)) {
return;
}
// this function gets called by format() to produce the appropriate substitution
// text for an individual pattern symbol (e.g., "HH" or "yyyy")
UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
const int32_t maxIntCount = 10;
int32_t beginOffset = appendTo.length();
const NumberFormat *currentNumberFormat;
DateFormatSymbols::ECapitalizationContextUsageType capContextUsageType = DateFormatSymbols::kCapContextUsageOther;
UBool isHebrewCalendar = (uprv_strcmp(cal.getType(),"hebrew") == 0);
UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0 || uprv_strcmp(cal.getType(),"dangi") == 0);
// 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;
}
currentNumberFormat = getNumberFormatByIndex(patternCharIndex);
if (currentNumberFormat == NULL) {
status = U_INTERNAL_PROGRAM_ERROR;
return;
}
UnicodeString hebr("hebr", 4, US_INV);
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:
if (isChineseCalendar) {
zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, 9); // as in ICU4J
} else {
if (count == 5) {
_appendSymbol(appendTo, value, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageEraNarrow;
} else if (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 != NULL && 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 ( isHebrewCalendar ) {
HebrewCalendar *hc = (HebrewCalendar*)&cal;
if (hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value == 6 && count >= 3 )
value = 13; // Show alternate form for Adar II in leap years in Hebrew calendar.
if (!hc->isLeapYear(hc->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 != NULL && 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]): NULL, status);
} else {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneNarrowMonths, fSymbols->fStandaloneNarrowMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneNarrow]): NULL, status);
}
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthNarrow;
} else if (count == 4) {
if (patternCharIndex == UDAT_MONTH_FIELD) {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fMonths, fSymbols->fMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide]): NULL, status);
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
} else {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide]): NULL, status);
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
}
} else if (count == 3) {
if (patternCharIndex == UDAT_MONTH_FIELD) {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fShortMonths, fSymbols->fShortMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev]): NULL, status);
capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
} else {
_appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount,
(isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev]): NULL, 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]): NULL, 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;
case UDAT_FRACTIONAL_SECOND_FIELD:
// Fractional seconds left-justify
{
int32_t minDigits = (count > 3) ? 3 : count;
if (count == 1) {
value /= 100;
} else if (count == 2) {
value /= 10;
}
zeroPaddingNumber(currentNumberFormat, appendTo, value, minDigits, maxIntCount);
if (count > 3) {
zeroPaddingNumber(currentNumberFormat, appendTo, 0, count - 3, 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;
} else if (count == 4) {
_appendSymbol(appendTo, value, fSymbols->fWeekdays,
fSymbols->fWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
} else if (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;
} else if (count == 4) {
_appendSymbol(appendTo, value, fSymbols->fStandaloneWeekdays,
fSymbols->fStandaloneWeekdaysCount);
capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
} else if (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;
case UDAT_TIMEZONE_FIELD: // 'z'
case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
case UDAT_TIMEZONE_ISO_FIELD: // 'X'
case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
{
UChar zsbuf[ZONE_NAME_U16_MAX];
UnicodeString zoneString(zsbuf, 0, UPRV_LENGTHOF(zsbuf));
const TimeZone& tz = cal.getTimeZone();
UDate date = cal.getTime(status);
const TimeZoneFormat *tzfmt = tzFormat(status);
if (U_SUCCESS(status)) {
if (patternCharIndex == UDAT_TIMEZONE_FIELD) {
if (count < 4) {
// "z", "zz", "zzz"
tzfmt->format(UTZFMT_STYLE_SPECIFIC_SHORT, tz, date, zoneString);
capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
} else {
// "zzzz" or longer
tzfmt->format(UTZFMT_STYLE_SPECIFIC_LONG, tz, date, zoneString);
capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
}
}
else if (patternCharIndex == UDAT_TIMEZONE_RFC_FIELD) {
if (count < 4) {
// "Z"
tzfmt->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
} else if (count == 5) {
// "ZZZZZ"
tzfmt->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
} else {
// "ZZ", "ZZZ", "ZZZZ"
tzfmt->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
}
}
else if (patternCharIndex == UDAT_TIMEZONE_GENERIC_FIELD) {
if (count == 1) {
// "v"
tzfmt->format(UTZFMT_STYLE_GENERIC_SHORT, tz, date, zoneString);
capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
} else if (count == 4) {
// "vvvv"
tzfmt->format(UTZFMT_STYLE_GENERIC_LONG, tz, date, zoneString);
capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
}
}
else if (patternCharIndex == UDAT_TIMEZONE_SPECIAL_FIELD) {
if (count == 1) {
// "V"
tzfmt->format(UTZFMT_STYLE_ZONE_ID_SHORT, tz, date, zoneString);
} else if (count == 2) {
// "VV"
tzfmt->format(UTZFMT_STYLE_ZONE_ID, tz, date, zoneString);
} else if (count == 3) {
// "VVV"
tzfmt->format(UTZFMT_STYLE_EXEMPLAR_LOCATION, tz, date, zoneString);
} else if (count == 4) {
// "VVVV"
tzfmt->format(UTZFMT_STYLE_GENERIC_LOCATION, tz, date, zoneString);
capContextUsageType = DateFormatSymbols::kCapContextUsageZoneLong;
}
}
else if (patternCharIndex == UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD) {
if (count == 1) {
// "O"
tzfmt->format(UTZFMT_STYLE_LOCALIZED_GMT_SHORT, tz, date, zoneString);
} else if (count == 4) {
// "OOOO"
tzfmt->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
}
}
else if (patternCharIndex == UDAT_TIMEZONE_ISO_FIELD) {
if (count == 1) {
// "X"
tzfmt->format(UTZFMT_STYLE_ISO_BASIC_SHORT, tz, date, zoneString);
} else if (count == 2) {
// "XX"
tzfmt->format(UTZFMT_STYLE_ISO_BASIC_FIXED, tz, date, zoneString);
} else if (count == 3) {
// "XXX"
tzfmt->format(UTZFMT_STYLE_ISO_EXTENDED_FIXED, tz, date, zoneString);
} else if (count == 4) {
// "XXXX"
tzfmt->format(UTZFMT_STYLE_ISO_BASIC_FULL, tz, date, zoneString);
} else if (count == 5) {
// "XXXXX"
tzfmt->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
}
}
else if (patternCharIndex == UDAT_TIMEZONE_ISO_LOCAL_FIELD) {
if (count == 1) {
// "x"
tzfmt->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT, tz, date, zoneString);
} else if (count == 2) {
// "xx"
tzfmt->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED, tz, date, zoneString);
} else if (count == 3) {
// "xxx"
tzfmt->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED, tz, date, zoneString);
} else if (count == 4) {
// "xxxx"
tzfmt->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
} else if (count == 5) {
// "xxxxx"
tzfmt->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL, tz, date, zoneString);
}
}
else {
UPRV_UNREACHABLE;
}
}
appendTo += zoneString;
}
break;
case UDAT_QUARTER_FIELD:
if (count >= 4)
_appendSymbol(appendTo, value/3, fSymbols->fQuarters,
fSymbols->fQuartersCount);
else if (count == 3)
_appendSymbol(appendTo, value/3, fSymbols->fShortQuarters,
fSymbols->fShortQuartersCount);
else
zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
break;
case UDAT_STANDALONE_QUARTER_FIELD:
if (count >= 4)
_appendSymbol(appendTo, value/3, fSymbols->fStandaloneQuarters,
fSymbols->fStandaloneQuartersCount);
else if (count == 3)
_appendSymbol(appendTo, value/3, fSymbols->fStandaloneShortQuarters,
fSymbols->fStandaloneShortQuartersCount);
else
zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
break;
case UDAT_AM_PM_MIDNIGHT_NOON_FIELD:
{
const UnicodeString *toAppend = NULL;
int32_t hour = cal.get(UCAL_HOUR_OF_DAY, status);
// 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);
if (count <= 3) {
toAppend = &fSymbols->fAbbreviatedDayPeriods[val];
} else if (count == 4 || count > 5) {
toAppend = &fSymbols->fWideDayPeriods[val];
} else { // count == 5
toAppend = &fSymbols->fNarrowDayPeriods[val];
}
}
// toAppend is NULL 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 == NULL || 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 == NULL) {
// 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);
}
// Determine day period.
DayPeriodRules::DayPeriod periodType;
if (hour == 0 && minute == 0 && second == 0 && ruleSet->hasMidnight()) {
periodType = DayPeriodRules::DAYPERIOD_MIDNIGHT;
} else if (hour == 12 && minute == 0 && second == 0 && ruleSet->hasNoon()) {
periodType = DayPeriodRules::DAYPERIOD_NOON;
} else {
periodType = ruleSet->getDayPeriodForHour(hour);
}
// Rule set exists, therefore periodType can't be UNKNOWN.
// Get localized string.
U_ASSERT(periodType != DayPeriodRules::DAYPERIOD_UNKNOWN);
UnicodeString *toAppend = NULL;
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.
if (periodType != DayPeriodRules::DAYPERIOD_AM &&
periodType != DayPeriodRules::DAYPERIOD_PM &&
periodType != DayPeriodRules::DAYPERIOD_MIDNIGHT) {
index = (int32_t)periodType;
if (count <= 3) {
toAppend = &fSymbols->fAbbreviatedDayPeriods[index]; // i.e. short
} else if (count == 4 || count > 5) {
toAppend = &fSymbols->fWideDayPeriods[index];
} else { // count == 5
toAppend = &fSymbols->fNarrowDayPeriods[index];
}
}
// Fallback schedule:
// Midnight/Noon -> General Periods -> AM/PM.
// Midnight/Noon -> General Periods.
if ((toAppend == NULL || toAppend->isBogus()) &&
(periodType == DayPeriodRules::DAYPERIOD_MIDNIGHT ||
periodType == DayPeriodRules::DAYPERIOD_NOON)) {
periodType = ruleSet->getDayPeriodForHour(hour);
index = (int32_t)periodType;
if (count <= 3) {
toAppend = &fSymbols->fAbbreviatedDayPeriods[index]; // i.e. short
} else if (count == 4 || count > 5) {
toAppend = &fSymbols->fWideDayPeriods[index];
} else { // count == 5
toAppend = &fSymbols->fNarrowDayPeriods[index];
}
}
// 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 != NULL && 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
handler.addAttribute(DateFormatSymbols::getPatternCharIndex(fieldToOutput), beginOffset, appendTo.length());
}
//----------------------------------------------------------------------
void SimpleDateFormat::adoptNumberFormat(NumberFormat *formatToAdopt) {
fixNumberFormatForDates(*formatToAdopt);
delete fNumberFormat;
fNumberFormat = formatToAdopt;
// We successfully set the default number format. Now delete the overrides
// (can't fail).
if (fSharedNumberFormatters) {
freeSharedNumberFormatters(fSharedNumberFormatters);
fSharedNumberFormatters = NULL;
}
// Also re-compute the fast formatters.
UErrorCode localStatus = U_ZERO_ERROR;
freeFastNumberFormatters();
initFastNumberFormatters(localStatus);
}
void SimpleDateFormat::adoptNumberFormat(const UnicodeString& fields, NumberFormat *formatToAdopt, UErrorCode &status){
fixNumberFormatForDates(*formatToAdopt);
LocalPointer<NumberFormat> fmt(formatToAdopt);
if (U_FAILURE(status)) {
return;
}
// We must ensure fSharedNumberFormatters is allocated.
if (fSharedNumberFormatters == NULL) {
fSharedNumberFormatters = allocSharedNumberFormatters();
if (fSharedNumberFormatters == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
}
const SharedNumberFormat *newFormat = createSharedNumberFormat(fmt.orphan());
if (newFormat == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
for (int i=0; i<fields.length(); i++) {
UChar 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();
}
const NumberFormat *
SimpleDateFormat::getNumberFormatForField(UChar field) const {
UDateFormatField index = DateFormatSymbols::getPatternCharIndex(field);
if (index == UDAT_FIELD_COUNT) {
return NULL;
}
return getNumberFormatByIndex(index);
}
//----------------------------------------------------------------------
void
SimpleDateFormat::zeroPaddingNumber(
const NumberFormat *currentNumberFormat,
UnicodeString &appendTo,
int32_t value, int32_t minDigits, int32_t maxDigits) const
{
const number::LocalizedNumberFormatter* fastFormatter = nullptr;
// NOTE: This uses the heuristic that these five min/max int settings account for the vast majority
// of SimpleDateFormat number formatting cases at the time of writing (ICU 62).
if (currentNumberFormat == fNumberFormat) {
if (maxDigits == 10) {
if (minDigits == 1) {
fastFormatter = fFastNumberFormatters[SMPDTFMT_NF_1x10];
} else if (minDigits == 2) {
fastFormatter = fFastNumberFormatters[SMPDTFMT_NF_2x10];
} else if (minDigits == 3) {
fastFormatter = fFastNumberFormatters[SMPDTFMT_NF_3x10];
} else if (minDigits == 4) {
fastFormatter = fFastNumberFormatters[SMPDTFMT_NF_4x10];
}
} else if (maxDigits == 2) {
if (minDigits == 2) {
fastFormatter = fFastNumberFormatters[SMPDTFMT_NF_2x2];
}
}
}
if (fastFormatter != nullptr) {
// Can use fast path
number::impl::UFormattedNumberData result;
result.quantity.setToInt(value);
UErrorCode localStatus = U_ZERO_ERROR;
fastFormatter->formatImpl(&result, localStatus);
if (U_FAILURE(localStatus)) {
return;
}
appendTo.append(result.getStringRef().toTempUnicodeString());
return;
}
// Check for RBNF (no clone necessary)
auto* 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, occuring count
* times, represents a numeric field.
*/
UBool SimpleDateFormat::isNumeric(UChar 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
return FALSE;
}
UChar ch = pattern.charAt(patternOffset);
UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
if (f == UDAT_FIELD_COUNT) {
// not at any field
return FALSE;
}
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
return FALSE;
}
UChar ch = pattern.charAt(--patternOffset);
UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
if (f == UDAT_FIELD_COUNT) {
// not after any field
return FALSE;
}
int32_t i = patternOffset;
while (pattern.charAt(--i) == ch) {}
return !DateFormatSymbols::isNumericField(f, patternOffset - i);
}
void
SimpleDateFormat::parse(const UnicodeString& text, Calendar& cal, ParsePosition& parsePos) const
{
UErrorCode status = U_ZERO_ERROR;
int32_t pos = parsePos.getIndex();
if(parsePos.getIndex() < 0) {
parsePos.setErrorIndex(0);
return;
}
int32_t start = pos;
// Hold the day period until everything else is parsed, because we need
// the hour to interpret time correctly.
int32_t dayPeriodInt = -1;
UBool ambiguousYear[] = { FALSE };
int32_t saveHebrewMonth = -1;
int32_t count = 0;
UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
// 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;
MessageFormat * numericLeapMonthFormatter = NULL;
Calendar* calClone = NULL;
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 != NULL) {
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 != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
numericLeapMonthFormatter = new MessageFormat(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric], fLocale, status);
if (numericLeapMonthFormatter == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto ExitParse;
} else if (U_FAILURE(status)) {
goto ExitParse; // this will delete numericLeapMonthFormatter
}
}
for (int32_t i=0; i<fPattern.length(); ++i) {
UChar 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;
}
}
pos = subParse(text, pos, ch, count,
TRUE, FALSE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter, &tzTimeType);
// 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.
else if (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
UChar 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;
}
}
}
}
else if (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 = (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 = (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 == NULL) {
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 == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto ExitParse;
}
const TimeZone & tz = cal.getTimeZone();
BasicTimeZone *btz = NULL;
if (dynamic_cast<const OlsonTimeZone *>(&tz) != NULL
|| dynamic_cast<const SimpleTimeZone *>(&tz) != NULL
|| dynamic_cast<const RuleBasedTimeZone *>(&tz) != NULL
|| dynamic_cast<const VTimeZone *>(&tz) != NULL) {
btz = (BasicTimeZone*)&tz;
}
// 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 != NULL) {
if (tzTimeType == UTZFMT_TIME_TYPE_STANDARD) {
btz->getOffsetFromLocal(localMillis,
BasicTimeZone::kStandard, BasicTimeZone::kStandard, raw, dst, status);
} else {
btz->getOffsetFromLocal(localMillis,
BasicTimeZone::kDaylight, BasicTimeZone::kDaylight, 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 != NULL) {
// 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 (numericLeapMonthFormatter != NULL) {
delete numericLeapMonthFormatter;
}
if (calClone != NULL) {
delete calClone;
}
// 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);
}
}
//----------------------------------------------------------------------
static int32_t
matchStringWithOptionalDot(const UnicodeString &text,
int32_t index,
const UnicodeString &data);
int32_t SimpleDateFormat::matchQuarterString(const UnicodeString& text,
int32_t start,
UCalendarDateFields field,
const UnicodeString* data,
int32_t dataCount,
Calendar& cal) const
{
int32_t i = 0;
int32_t count = dataCount;
// 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;
}
}
if (bestMatch >= 0) {
cal.set(field, bestMatch * 3);
return start + bestMatchLength;
}
return -start;
}
int32_t SimpleDateFormat::matchDayPeriodStrings(const UnicodeString& text, int32_t start,
const UnicodeString* data, int32_t dataCount,
int32_t &dayPeriod) const
{
int32_t bestMatchLength = 0, bestMatch = -1;
for (int32_t i = 0; i < dataCount; ++i) {
int32_t matchLength = 0;
if ((matchLength = matchStringWithOptionalDot(text, start, data[i])) > bestMatchLength) {
bestMatchLength = matchLength;
bestMatch = i;
}
}
if (bestMatch >= 0) {
dayPeriod = bestMatch;
return start + bestMatchLength;
}
return -start;
}
//----------------------------------------------------------------------
UBool SimpleDateFormat::matchLiterals(const UnicodeString &pattern,
int32_t &patternOffset,
const UnicodeString &text,
int32_t &textOffset,
UBool whitespaceLenient,
UBool partialMatchLenient,
UBool oldLeniency)
{
UBool inQuote = FALSE;
UnicodeString literal;
int32_t i = patternOffset;
// scan pattern looking for contiguous literal characters
for ( ; i < pattern.length(); i += 1) {
UChar 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()) {
UChar 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
return FALSE;
}
// 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
UChar 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 paritial matches
if(partialMatchLenient && oldLeniency) {