X-Git-Url: https://git.saurik.com/apple/icu.git/blobdiff_plain/51004dcb01e06fef634b61be77ed73dd61cb6db9..HEAD:/icuSources/i18n/dtptngen.cpp diff --git a/icuSources/i18n/dtptngen.cpp b/icuSources/i18n/dtptngen.cpp index ddb55327..3c643369 100644 --- a/icuSources/i18n/dtptngen.cpp +++ b/icuSources/i18n/dtptngen.cpp @@ -1,6 +1,8 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html /* ******************************************************************************* -* Copyright (C) 2007-2013, International Business Machines Corporation and +* Copyright (C) 2007-2016, International Business Machines Corporation and * others. All Rights Reserved. ******************************************************************************* * @@ -16,7 +18,9 @@ #include "unicode/decimfmt.h" #include "unicode/dtfmtsym.h" #include "unicode/dtptngen.h" -#include "unicode/msgfmt.h" +#include "unicode/localpointer.h" +#include "unicode/schriter.h" +#include "unicode/simpleformatter.h" #include "unicode/smpdtfmt.h" #include "unicode/udat.h" #include "unicode/udatpg.h" @@ -25,18 +29,20 @@ #include "unicode/ures.h" #include "unicode/ustring.h" #include "unicode/rep.h" +#include "unicode/region.h" #include "cpputils.h" -#include "ucln_in.h" #include "mutex.h" +#include "umutex.h" #include "cmemory.h" #include "cstring.h" #include "locbased.h" -#include "gregoimp.h" #include "hash.h" +#include "uhash.h" #include "uresimp.h" #include "dtptngen_impl.h" - -#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0])) +#include "ucln_in.h" +#include "charstr.h" +#include "uassert.h" #if U_CHARSET_FAMILY==U_EBCDIC_FAMILY /** @@ -85,17 +91,17 @@ static void ures_a_open(UResourceBundleAIterator *aiter, UResourceBundle *bund, aiter->num = ures_getSize(aiter->bund); aiter->cursor = 0; #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR) - aiter->entries = NULL; + aiter->entries = nullptr; #else aiter->entries = (UResAEntry*)uprv_malloc(sizeof(UResAEntry)*aiter->num); for(int i=0;inum;i++) { - aiter->entries[i].item = ures_getByIndex(aiter->bund, i, NULL, status); + aiter->entries[i].item = ures_getByIndex(aiter->bund, i, nullptr, status); const char *akey = ures_getKey(aiter->entries[i].item); int32_t len = uprv_strlen(akey)+1; aiter->entries[i].key = (UChar*)uprv_malloc(len*sizeof(UChar)); u_charsToUChars(akey, aiter->entries[i].key, len); } - uprv_sortArray(aiter->entries, aiter->num, sizeof(UResAEntry), ures_a_codepointSort, NULL, TRUE, status); + uprv_sortArray(aiter->entries, aiter->num, sizeof(UResAEntry), ures_a_codepointSort, nullptr, TRUE, status); #endif } @@ -112,7 +118,7 @@ static const UChar *ures_a_getNextString(UResourceBundleAIterator *aiter, int32_ #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR) return ures_getNextString(aiter->bund, len, key, err); #else - if(U_FAILURE(*err)) return NULL; + if(U_FAILURE(*err)) return nullptr; UResourceBundle *item = aiter->entries[aiter->cursor].item; const UChar* ret = ures_getString(item, len, err); *key = ures_getKey(item); @@ -127,32 +133,39 @@ static const UChar *ures_a_getNextString(UResourceBundleAIterator *aiter, int32_ U_NAMESPACE_BEGIN - // ***************************************************************************** // class DateTimePatternGenerator // ***************************************************************************** static const UChar Canonical_Items[] = { - // GyQMwWEdDFHmsSv - CAP_G, LOW_Y, CAP_Q, CAP_M, LOW_W, CAP_W, CAP_E, LOW_D, CAP_D, CAP_F, + // GyQMwWEDFdaHmsSv + CAP_G, LOW_Y, CAP_Q, CAP_M, LOW_W, CAP_W, CAP_E, + CAP_D, CAP_F, LOW_D, LOW_A, // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J CAP_H, LOW_M, LOW_S, CAP_S, LOW_V, 0 }; static const dtTypeElem dtTypes[] = { // patternChar, field, type, minLen, weight {CAP_G, UDATPG_ERA_FIELD, DT_SHORT, 1, 3,}, - {CAP_G, UDATPG_ERA_FIELD, DT_LONG, 4, 0}, + {CAP_G, UDATPG_ERA_FIELD, DT_LONG, 4, 0}, + {CAP_G, UDATPG_ERA_FIELD, DT_NARROW, 5, 0}, + {LOW_Y, UDATPG_YEAR_FIELD, DT_NUMERIC, 1, 20}, {CAP_Y, UDATPG_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 20}, {LOW_U, UDATPG_YEAR_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 20}, + {LOW_R, UDATPG_YEAR_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20}, {CAP_U, UDATPG_YEAR_FIELD, DT_SHORT, 1, 3}, {CAP_U, UDATPG_YEAR_FIELD, DT_LONG, 4, 0}, {CAP_U, UDATPG_YEAR_FIELD, DT_NARROW, 5, 0}, + {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2}, {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0}, {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0}, + {CAP_Q, UDATPG_QUARTER_FIELD, DT_NARROW, 5, 0}, {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, - {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT + DT_DELTA, 3, 0}, - {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG + DT_DELTA, 4, 0}, + {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT - DT_DELTA, 3, 0}, + {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG - DT_DELTA, 4, 0}, + {LOW_Q, UDATPG_QUARTER_FIELD, DT_NARROW - DT_DELTA, 5, 0}, + {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2}, {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0}, {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0}, @@ -162,56 +175,109 @@ static const dtTypeElem dtTypes[] = { {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0}, {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0}, {LOW_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 1}, + {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2}, - {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 0}, + + {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC, 1, 0}, + {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3}, {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0}, {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0}, + {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER, 6, 0}, {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2}, {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0}, {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0}, {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0}, + {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORTER - 2*DT_DELTA, 6, 0}, {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0}, {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0}, {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0}, + {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER - DT_DELTA, 6, 0}, + {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2}, - {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 3}, - {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 0}, - {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20}, // really internal use, so we don't care - {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 0}, + {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 20}, // really internal use, so we don't care + + {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC, 1, 3}, + + {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC, 1, 0}, + + {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 3}, + {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_LONG, 4, 0}, + {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_NARROW, 5, 0}, + {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - DT_DELTA, 1, 3}, + {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - DT_DELTA, 4, 0}, + {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - DT_DELTA, 5, 0}, + // b needs to be closer to a than to B, so we make this 3*DT_DELTA + {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - 3*DT_DELTA, 1, 3}, + {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - 3*DT_DELTA, 4, 0}, + {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - 3*DT_DELTA, 5, 0}, + {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour - {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2}, + {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2}, // 24 hour {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour - {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, + {CAP_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // 12 hour + // The C code has had versions of the following 3, keep & update. Should not need these, but... + // Without these, certain tests using e.g. staticGetSkeleton fail because j/J in patterns + // get skipped instead of mapped to the right hour chars, for example in + // DateFormatTest::TestPatternFromSkeleton + // IntlTestDateTimePatternGeneratorAPI:: testStaticGetSkeleton + // DateIntervalFormatTest::testTicket11985 + // Need to investigate better handling of jJC replacement e.g. in staticGetSkeleton. + {CAP_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 5*DT_DELTA, 1, 2}, // 12/24 hour no AM/PM + {LOW_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 6*DT_DELTA, 1, 6}, // 12/24 hour + {CAP_C, UDATPG_HOUR_FIELD, DT_NUMERIC + 7*DT_DELTA, 1, 6}, // 12/24 hour with preferred dayPeriods for 12 + {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2}, + {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2}, - {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000}, - {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 1000}, + {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000}, + + {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC, 1, 1000}, + {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0}, {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0}, {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3}, {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0}, - {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3}, + {CAP_Z, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 3}, {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, - {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3}, - {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, + {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 5, 0}, + {CAP_O, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0}, + {CAP_O, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, + {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0}, + {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 2, 0}, + {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-1 - DT_DELTA, 3, 0}, + {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-2 - DT_DELTA, 4, 0}, + {CAP_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0}, + {CAP_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0}, + {CAP_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, + {LOW_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0}, + {LOW_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0}, + {LOW_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, + {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[] }; static const char* const CLDR_FIELD_APPEND[] = { - "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week", "Day", "*", "*", "*", + "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week", + "*", "*", "Day", "*", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J "Hour", "Minute", "Second", "*", "Timezone" }; -static const char* const CLDR_FIELD_NAME[] = { - "era", "year", "quarter", "month", "week", "*", "weekday", "*", "*", "day", "dayperiod", +static const char* const CLDR_FIELD_NAME[UDATPG_FIELD_COUNT] = { + "era", "year", "quarter", "month", "week", "weekOfMonth", "weekday", + "dayOfYear", "weekdayOfMonth", "day", "dayperiod", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J "hour", "minute", "second", "*", "zone" }; -static const char* const Resource_Fields[] = { - "day", "dayperiod", "era", "hour", "minute", "month", "second", "week", - "weekday", "year", "zone", "quarter" }; +static const char* const CLDR_FIELD_WIDTH[] = { // [UDATPG_WIDTH_COUNT] + "", "-short", "-narrow" +}; + +// TODO(ticket:13619): remove when definition uncommented in dtptngen.h. +static const int32_t UDATPG_WIDTH_COUNT = UDATPG_NARROW + 1; +static constexpr UDateTimePGDisplayWidth UDATPG_WIDTH_APPENDITEM = UDATPG_WIDE; +static constexpr int32_t UDATPG_FIELD_KEY_MAX = 24; // max length of CLDR field tag (type + width) // For appendItems static const UChar UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A, @@ -237,74 +303,80 @@ DateTimePatternGenerator::createInstance(UErrorCode& status) { } DateTimePatternGenerator* U_EXPORT2 -DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) { - DateTimePatternGenerator *result = new DateTimePatternGenerator(locale, status); - if (result == NULL) { - status = U_MEMORY_ALLOCATION_ERROR; - } +DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status, UBool skipICUData) { if (U_FAILURE(status)) { - delete result; - result = NULL; + return nullptr; } - return result; + LocalPointer result( + new DateTimePatternGenerator(locale, status, skipICUData), status); + return U_SUCCESS(status) ? result.orphan() : nullptr; } DateTimePatternGenerator* U_EXPORT2 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) { - DateTimePatternGenerator *result = new DateTimePatternGenerator(status); - if (result == NULL) { - status = U_MEMORY_ALLOCATION_ERROR; - } if (U_FAILURE(status)) { - delete result; - result = NULL; + return nullptr; } - return result; + LocalPointer result( + new DateTimePatternGenerator(status), status); + return U_SUCCESS(status) ? result.orphan() : nullptr; } DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) : - skipMatcher(NULL), - fAvailableFormatKeyHash(NULL) + skipMatcher(nullptr), + fAvailableFormatKeyHash(nullptr), + internalErrorCode(U_ZERO_ERROR) { fp = new FormatParser(); dtMatcher = new DateTimeMatcher(); distanceInfo = new DistanceInfo(); patternMap = new PatternMap(); - if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) { - status = U_MEMORY_ALLOCATION_ERROR; + if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) { + internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR; } } -DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status) : - skipMatcher(NULL), - fAvailableFormatKeyHash(NULL) +DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status, UBool skipICUData) : + skipMatcher(nullptr), + fAvailableFormatKeyHash(nullptr), + internalErrorCode(U_ZERO_ERROR), + pLocale(locale) { fp = new FormatParser(); dtMatcher = new DateTimeMatcher(); distanceInfo = new DistanceInfo(); patternMap = new PatternMap(); - if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) { - status = U_MEMORY_ALLOCATION_ERROR; + if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) { + internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR; } else { - initData(locale, status); + initData(locale, status, skipICUData); } } DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) : UObject(), - skipMatcher(NULL), - fAvailableFormatKeyHash(NULL) + skipMatcher(nullptr), + fAvailableFormatKeyHash(nullptr), + internalErrorCode(U_ZERO_ERROR) { fp = new FormatParser(); dtMatcher = new DateTimeMatcher(); distanceInfo = new DistanceInfo(); patternMap = new PatternMap(); + if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) { + internalErrorCode = U_MEMORY_ALLOCATION_ERROR; + } *this=other; } DateTimePatternGenerator& DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) { + // reflexive case + if (&other == this) { + return *this; + } + internalErrorCode = other.internalErrorCode; pLocale = other.pLocale; fDefaultHourFormatChar = other.fDefaultHourFormatChar; *fp = *(other.fp); @@ -316,22 +388,27 @@ DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) { dateTimeFormat.getTerminatedBuffer(); decimal.getTerminatedBuffer(); delete skipMatcher; - if ( other.skipMatcher == NULL ) { - skipMatcher = NULL; + if ( other.skipMatcher == nullptr ) { + skipMatcher = nullptr; } else { skipMatcher = new DateTimeMatcher(*other.skipMatcher); + if (skipMatcher == nullptr) + { + internalErrorCode = U_MEMORY_ALLOCATION_ERROR; + return *this; + } } for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) { appendItemFormats[i] = other.appendItemFormats[i]; - appendItemNames[i] = other.appendItemNames[i]; - // NUL-terminate for the C API. - appendItemFormats[i].getTerminatedBuffer(); - appendItemNames[i].getTerminatedBuffer(); + appendItemFormats[i].getTerminatedBuffer(); // NUL-terminate for the C API. + for (int32_t j=0; j< UDATPG_WIDTH_COUNT; ++j ) { + fieldDisplayNames[i][j] = other.fieldDisplayNames[i][j]; + fieldDisplayNames[i][j].getTerminatedBuffer(); // NUL-terminate for the C API. + } } - UErrorCode status = U_ZERO_ERROR; - patternMap->copyFrom(*other.patternMap, status); - copyHashtable(other.fAvailableFormatKeyHash, status); + patternMap->copyFrom(*other.patternMap, internalErrorCode); + copyHashtable(other.fAvailableFormatKeyHash, internalErrorCode); return *this; } @@ -344,10 +421,14 @@ DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) cons if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) && (dateTimeFormat==other.dateTimeFormat) && (decimal==other.decimal)) { for ( int32_t i=0 ; i list; + int32_t length = 0; + int32_t preferredFormat = ALLOWED_HOUR_FORMAT_UNKNOWN; + for (int32_t j = 0; formatList.getKeyAndValue(j, key, value); ++j) { + if (uprv_strcmp(key, "allowed") == 0) { + if (value.getType() == URES_STRING) { + length = 2; // 1 preferred to add later, 1 allowed to add now + if (list.allocateInsteadAndReset(length + 1) == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + list[1] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode)); + } + else { + ResourceArray allowedFormats = value.getArray(errorCode); + length = allowedFormats.getSize() + 1; // 1 preferred, getSize allowed + if (list.allocateInsteadAndReset(length + 1) == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + for (int32_t k = 1; k < length; ++k) { + allowedFormats.getValue(k-1, value); + list[k] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode)); + } + } + } else if (uprv_strcmp(key, "preferred") == 0) { + preferredFormat = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode)); + } + } + if (length > 1) { + list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: list[1]; + } else { + // fallback handling for missing data + length = 2; // 1 preferred, 1 allowed + if (list.allocateInsteadAndReset(length + 1) == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: ALLOWED_HOUR_FORMAT_H; + list[1] = list[0]; + } + list[length] = ALLOWED_HOUR_FORMAT_UNKNOWN; + // At this point list[] will have at least two non-ALLOWED_HOUR_FORMAT_UNKNOWN entries, + // followed by ALLOWED_HOUR_FORMAT_UNKNOWN. + uhash_put(localeToAllowedHourFormatsMap, const_cast(regionOrLocale), list.orphan(), &errorCode); + if (U_FAILURE(errorCode)) { return; } + } + } + + AllowedHourFormat getHourFormatFromUnicodeString(const UnicodeString &s) { + if (s.length() == 1) { + if (s[0] == LOW_H) { return ALLOWED_HOUR_FORMAT_h; } + if (s[0] == CAP_H) { return ALLOWED_HOUR_FORMAT_H; } + if (s[0] == CAP_K) { return ALLOWED_HOUR_FORMAT_K; } + if (s[0] == LOW_K) { return ALLOWED_HOUR_FORMAT_k; } + } else if (s.length() == 2) { + if (s[0] == LOW_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_hb; } + if (s[0] == LOW_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_hB; } + if (s[0] == CAP_K && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Kb; } + if (s[0] == CAP_K && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_KB; } + if (s[0] == CAP_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Hb; } + if (s[0] == CAP_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_HB; } + } + + return ALLOWED_HOUR_FORMAT_UNKNOWN; + } +}; + +} // namespace + +AllowedHourFormatsSink::~AllowedHourFormatsSink() {} + +U_CFUNC void U_CALLCONV DateTimePatternGenerator::loadAllowedHourFormatsData(UErrorCode &status) { + if (U_FAILURE(status)) { return; } + localeToAllowedHourFormatsMap = uhash_open( + uhash_hashChars, uhash_compareChars, nullptr, &status); + if (U_FAILURE(status)) { return; } + + uhash_setValueDeleter(localeToAllowedHourFormatsMap, deleteAllowedHourFormats); + ucln_i18n_registerCleanup(UCLN_I18N_ALLOWED_HOUR_FORMATS, allowedHourFormatsCleanup); + + LocalUResourceBundlePointer rb(ures_openDirect(nullptr, "supplementalData", &status)); + if (U_FAILURE(status)) { return; } + + AllowedHourFormatsSink sink; + // TODO: Currently in the enumeration each table allocates a new array. + // Try to reduce the number of memory allocations. Consider storing a + // UVector32 with the concatenation of all of the sub-arrays, put the start index + // into the hashmap, store 6 single-value sub-arrays right at the beginning of the + // vector (at index enum*2) for easy data sharing, copy sub-arrays into runtime + // object. Remember to clean up the vector, too. + ures_getAllItemsWithFallback(rb.getAlias(), "timeData", sink, status); +} + +static int32_t* getAllowedHourFormatsLangCountry(const char* language, const char* country, UErrorCode& status) { + CharString langCountry; + langCountry.append(language, status); + langCountry.append('_', status); + langCountry.append(country, status); + + int32_t* allowedFormats; + allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, langCountry.data()); + if (allowedFormats == nullptr) { + allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, const_cast(country)); + } + + return allowedFormats; +} + +void DateTimePatternGenerator::getAllowedHourFormats(const Locale &locale, UErrorCode &status) { + if (U_FAILURE(status)) { return; } + + const char *language = locale.getLanguage(); + const char *country = locale.getCountry(); + const char *locName = locale.getName(); // Apple addition + if (*locName==0 || uprv_strcmp(locName,"root")==0 || uprv_strcmp(locName,"und")==0) { // Apple addition + language = "und"; + country = "001"; + } + Locale maxLocale; // must be here for correct lifetime + if (*language == '\0' || *country == '\0') { + maxLocale = locale; + UErrorCode localStatus = U_ZERO_ERROR; + maxLocale.addLikelySubtags(localStatus); + if (U_SUCCESS(localStatus)) { + language = maxLocale.getLanguage(); + country = maxLocale.getCountry(); + } + } + if (*language == '\0') { + // Unexpected, but fail gracefully + language = "und"; + } + if (*country == '\0') { + country = "001"; + } + + int32_t* allowedFormats = getAllowedHourFormatsLangCountry(language, country, status); + + // Check if the region has an alias + if (allowedFormats == nullptr) { + UErrorCode localStatus = U_ZERO_ERROR; + const Region* region = Region::getInstance(country, localStatus); + if (U_SUCCESS(localStatus)) { + country = region->getRegionCode(); // the real region code + allowedFormats = getAllowedHourFormatsLangCountry(language, country, status); + } + } + + if (allowedFormats != nullptr) { // Lookup is successful + // Here allowedFormats points to a list consisting of key for preferredFormat, + // followed by one or more keys for allowedFormats, then followed by ALLOWED_HOUR_FORMAT_UNKNOWN. + switch (allowedFormats[0]) { + case ALLOWED_HOUR_FORMAT_h: fDefaultHourFormatChar = LOW_H; break; + case ALLOWED_HOUR_FORMAT_H: fDefaultHourFormatChar = CAP_H; break; + case ALLOWED_HOUR_FORMAT_K: fDefaultHourFormatChar = CAP_K; break; + case ALLOWED_HOUR_FORMAT_k: fDefaultHourFormatChar = LOW_K; break; + default: fDefaultHourFormatChar = CAP_H; break; + } + for (int32_t i = 0; i < UPRV_LENGTHOF(fAllowedHourFormats); ++i) { + fAllowedHourFormats[i] = allowedFormats[i + 1]; + if (fAllowedHourFormats[i] == ALLOWED_HOUR_FORMAT_UNKNOWN) { + break; + } + } + } else { // Lookup failed, twice + fDefaultHourFormatChar = CAP_H; + fAllowedHourFormats[0] = ALLOWED_HOUR_FORMAT_H; + fAllowedHourFormats[1] = ALLOWED_HOUR_FORMAT_UNKNOWN; + } +} + UnicodeString DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) { - dtMatcher->set(pattern, fp); - return dtMatcher->getSkeletonPtr()->getSkeleton(); + FormatParser fp2; + DateTimeMatcher matcher; + PtnSkeleton localSkeleton; + matcher.set(pattern, &fp2, localSkeleton); + return localSkeleton.getSkeleton(); +} + +UnicodeString +DateTimePatternGenerator::staticGetSkeleton( + const UnicodeString& pattern, UErrorCode& /*status*/) { + FormatParser fp; + DateTimeMatcher matcher; + PtnSkeleton localSkeleton; + matcher.set(pattern, &fp, localSkeleton); + return localSkeleton.getSkeleton(); } UnicodeString DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) { - dtMatcher->set(pattern, fp); - return dtMatcher->getSkeletonPtr()->getBaseSkeleton(); + FormatParser fp2; + DateTimeMatcher matcher; + PtnSkeleton localSkeleton; + matcher.set(pattern, &fp2, localSkeleton); + return localSkeleton.getBaseSkeleton(); +} + +UnicodeString +DateTimePatternGenerator::staticGetBaseSkeleton( + const UnicodeString& pattern, UErrorCode& /*status*/) { + FormatParser fp; + DateTimeMatcher matcher; + PtnSkeleton localSkeleton; + matcher.set(pattern, &fp, localSkeleton); + return localSkeleton.getBaseSkeleton(); } void DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) { + if (U_FAILURE(status)) { return; } UnicodeString dfPattern; UnicodeString conflictingString; DateFormat* df; - if (U_FAILURE(status)) { - return; - } - // Load with ICU patterns for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) { DateFormat::EStyle style = (DateFormat::EStyle)i; df = DateFormat::createDateInstance(style, locale); SimpleDateFormat* sdf; - if (df != NULL && (sdf = dynamic_cast(df)) != NULL) { - addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status); + if (df != nullptr && (sdf = dynamic_cast(df)) != nullptr) { + sdf->toPattern(dfPattern); + addPattern(dfPattern, FALSE, conflictingString, status); } // TODO Maybe we should return an error when the date format isn't simple. delete df; - if (U_FAILURE(status)) { - return; - } + if (U_FAILURE(status)) { return; } df = DateFormat::createTimeInstance(style, locale); - if (df != NULL && (sdf = dynamic_cast(df)) != NULL) { - addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status); - // HACK for hh:ss - if ( i==DateFormat::kMedium ) { - hackPattern = dfPattern; + if (df != nullptr && (sdf = dynamic_cast(df)) != nullptr) { + sdf->toPattern(dfPattern); + addPattern(dfPattern, FALSE, conflictingString, status); + + // TODO: C++ and Java are inconsistent (see #12568). + // C++ uses MEDIUM, but Java uses SHORT. + if ( i==DateFormat::kShort && !dfPattern.isEmpty() ) { + consumeShortTimePattern(dfPattern, status); } } // TODO Maybe we should return an error when the date format isn't simple. delete df; - if (U_FAILURE(status)) { - return; - } + if (U_FAILURE(status)) { return; } } } @@ -490,238 +820,238 @@ DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY) -static const UChar hourFormatChars[] = { CAP_H, LOW_H, CAP_K, LOW_K, 0 }; // HhKk, the hour format characters - void -DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& err) { - UResourceBundle *rb, *calTypeBundle, *calBundle; - UResourceBundle *patBundle, *fieldBundle, *fBundle; - UnicodeString rbPattern, value, field; - UnicodeString conflictingPattern; - const char *key=NULL; - int32_t i; - - UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, LENGTHOF(UDATPG_ItemFormat)-1); // Read-only alias. - - err = U_ZERO_ERROR; - - fDefaultHourFormatChar = 0; - for (i=0; i0) { - hackTimes(hackPattern, err); } +}; + +// Virtual destructors must be defined out of line. +DateTimePatternGenerator::AppendItemFormatsSink::~AppendItemFormatsSink() {} +DateTimePatternGenerator::AppendItemNamesSink::~AppendItemNamesSink() {} +DateTimePatternGenerator::AvailableFormatsSink::~AvailableFormatsSink() {} + +void +DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& errorCode) { + if (U_FAILURE(errorCode)) { return; } + UnicodeString rbPattern, value, field; + CharString path; + + UBool hasCountryFallbackResource = FALSE; + LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &errorCode)); + LocalUResourceBundlePointer countryRB(ures_openWithCountryFallback(nullptr, locale.getName(), &hasCountryFallbackResource, &errorCode)); + if (U_FAILURE(errorCode)) { return; } + + CharString calendarTypeToUse; // to be filled in with the type to use, if all goes well + getCalendarTypeToUse(locale, calendarTypeToUse, errorCode); + if (U_FAILURE(errorCode)) { return; } + + // Local err to ignore resource not found exceptions + UErrorCode err = U_ZERO_ERROR; + + // Load append item formats. + AppendItemFormatsSink appendItemFormatsSink(*this); + path.clear() + .append(DT_DateTimeCalendarTag, errorCode) + .append('/', errorCode) + .append(calendarTypeToUse, errorCode) + .append('/', errorCode) + .append(DT_DateTimeAppendItemsTag, errorCode); // i.e., calendar/xxx/appendItems + if (U_FAILURE(errorCode)) { return; } + ures_getAllItemsWithFallback(rb.getAlias(), path.data(), appendItemFormatsSink, err); + appendItemFormatsSink.fillInMissing(); + + // Load CLDR item names. + err = U_ZERO_ERROR; + AppendItemNamesSink appendItemNamesSink(*this); + ures_getAllItemsWithFallback(rb.getAlias(), DT_DateTimeFieldsTag, appendItemNamesSink, err); + appendItemNamesSink.fillInMissing(); + + // Load the available formats from CLDR. + err = U_ZERO_ERROR; + initHashtable(errorCode); + if (U_FAILURE(errorCode)) { return; } + path.clear() + .append(DT_DateTimeCalendarTag, errorCode) + .append('/', errorCode) + .append(calendarTypeToUse, errorCode) + .append('/', errorCode) + .append(DT_DateTimeAvailableFormatsTag, errorCode); // i.e., calendar/xxx/availableFormats + if (U_FAILURE(errorCode)) { return; } + if (hasCountryFallbackResource) { + AvailableFormatsSink countryAvailableFormatsSink(*this, TRUE); + ures_getAllItemsWithFallback(countryRB.getAlias(), path.data(), countryAvailableFormatsSink, err); + } + AvailableFormatsSink availableFormatsSink(*this, FALSE); + ures_getAllItemsWithFallback(rb.getAlias(), path.data(), availableFormatsSink, err); } void DateTimePatternGenerator::initHashtable(UErrorCode& err) { - if (fAvailableFormatKeyHash!=NULL) { + if (U_FAILURE(err)) { return; } + if (fAvailableFormatKeyHash!=nullptr) { return; } - if ((fAvailableFormatKeyHash = new Hashtable(FALSE, err))==NULL) { - err=U_MEMORY_ALLOCATION_ERROR; - return; + LocalPointer hash(new Hashtable(FALSE, err), err); + if (U_SUCCESS(err)) { + fAvailableFormatKeyHash = hash.orphan(); } } - void DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) { appendItemFormats[field] = value; @@ -736,20 +1066,35 @@ DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const void DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) { - appendItemNames[field] = value; - // NUL-terminate for the C API. - appendItemNames[field].getTerminatedBuffer(); + setFieldDisplayName(field, UDATPG_WIDTH_APPENDITEM, value); } const UnicodeString& -DateTimePatternGenerator:: getAppendItemName(UDateTimePatternField field) const { - return appendItemNames[field]; +DateTimePatternGenerator::getAppendItemName(UDateTimePatternField field) const { + return fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM]; +} + +void +DateTimePatternGenerator::setFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width, const UnicodeString& value) { + fieldDisplayNames[field][width] = value; + // NUL-terminate for the C API. + fieldDisplayNames[field][width].getTerminatedBuffer(); +} + +UnicodeString +DateTimePatternGenerator::getFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) const { + return fieldDisplayNames[field][width]; +} + +UnicodeString& +DateTimePatternGenerator::getMutableFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) { + return fieldDisplayNames[field][width]; } void DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) { value = SINGLE_QUOTE; - value += appendItemNames[field]; + value += fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM]; value += SINGLE_QUOTE; } @@ -760,28 +1105,54 @@ DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErro UnicodeString DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) { - const UnicodeString *bestPattern=NULL; + if (U_FAILURE(status)) { + return UnicodeString(); + } + if (U_FAILURE(internalErrorCode)) { + status = internalErrorCode; + return UnicodeString(); + } + const UnicodeString *bestPattern = nullptr; UnicodeString dtFormat; UnicodeString resultPattern; + int32_t flags = kDTPGNoFlags; int32_t dateMask=(1<set(patternFormCopy, fp); - const PtnSkeleton* specifiedSkeleton=NULL; - bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, &specifiedSkeleton); + dtMatcher->set(patternFormMapped, fp); + const PtnSkeleton* specifiedSkeleton = nullptr; + bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, status, &specifiedSkeleton); + + // getBestRaw() might return a pattern with an era field in it, even if the skeleton didn't specifically ask for it. + // Check for that and take it out of distanceInfo->missingFieldMask so that we don't end up adding a SECOND era + // field by mistake + if (bestPattern->indexOf(u'G') != -1) { + distanceInfo->missingFieldMask &= ~(1 << UDATPG_ERA_FIELD); + } + + if (U_FAILURE(status)) { + return UnicodeString(); + } + if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) { - resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, FALSE, options); + resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, flags, options); return resultPattern; } int32_t neededFields = dtMatcher->getFieldMask(); - UnicodeString datePattern=getBestAppending(neededFields & dateMask, options); - UnicodeString timePattern=getBestAppending(neededFields & timeMask, options); + UnicodeString datePattern=getBestAppending(neededFields & dateMask, flags, status, options); + UnicodeString timePattern=getBestAppending(neededFields & timeMask, flags, status, options); + if (U_FAILURE(status)) { + return UnicodeString(); + } if (datePattern.length()==0) { if (timePattern.length()==0) { resultPattern.remove(); @@ -796,11 +1167,96 @@ DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDate resultPattern.remove(); status = U_ZERO_ERROR; dtFormat=getDateTimeFormat(); - Formattable dateTimeObject[] = { timePattern, datePattern }; - resultPattern = MessageFormat::format(dtFormat, dateTimeObject, 2, resultPattern, status ); + SimpleFormatter(dtFormat, 2, 2, status).format(timePattern, datePattern, resultPattern, status); return resultPattern; } +/* + * Map a skeleton that may have metacharacters jJC to one without, by replacing + * the metacharacters with locale-appropriate fields of h/H/k/K and of a/b/B + * (depends on fDefaultHourFormatChar and fAllowedHourFormats being set, which in + * turn depends on initData having been run). This method also updates the flags + * as necessary. Returns the updated skeleton. + */ +UnicodeString +DateTimePatternGenerator::mapSkeletonMetacharacters(const UnicodeString& patternForm, int32_t* flags, UDateTimePatternMatchOptions options, UErrorCode& status) { + UnicodeString patternFormMapped; + patternFormMapped.remove(); + UChar hourFormatSkeletonCharForLowJ = fDefaultHourFormatChar; + switch (options & UADATPG_FORCE_HOUR_CYCLE_MASK) { + case UADATPG_FORCE_12_HOUR_CYCLE: hourFormatSkeletonCharForLowJ = LOW_H; break; + case UADATPG_FORCE_24_HOUR_CYCLE: hourFormatSkeletonCharForLowJ = CAP_H; break; + default: break; + } + UBool inQuoted = FALSE; + int32_t patPos, patLen = patternForm.length(); + for (patPos = 0; patPos < patLen; patPos++) { + UChar patChr = patternForm.charAt(patPos); + if (patChr == SINGLE_QUOTE) { + inQuoted = !inQuoted; + } else if (!inQuoted) { + // Handle special mappings for 'j' and 'C' in which fields lengths + // 1,3,5 => hour field length 1 + // 2,4,6 => hour field length 2 + // 1,2 => abbreviated dayPeriod (field length 1..3) + // 3,4 => long dayPeriod (field length 4) + // 5,6 => narrow dayPeriod (field length 5) + if (patChr == LOW_J || patChr == CAP_C) { + int32_t extraLen = 0; // 1 less than total field length + while (patPos+1 < patLen && patternForm.charAt(patPos+1)==patChr) { + extraLen++; + patPos++; + } + int32_t hourLen = 1 + (extraLen & 1); + int32_t dayPeriodLen = (extraLen < 2)? 1: 3 + (extraLen >> 1); + UChar hourChar = LOW_H; + UChar dayPeriodChar = LOW_A; + if (patChr == LOW_J) { + hourChar = hourFormatSkeletonCharForLowJ; + } else { + AllowedHourFormat bestAllowed; + if (fAllowedHourFormats[0] != ALLOWED_HOUR_FORMAT_UNKNOWN) { + bestAllowed = (AllowedHourFormat)fAllowedHourFormats[0]; + } else { + status = U_INVALID_FORMAT_ERROR; + return UnicodeString(); + } + if (bestAllowed == ALLOWED_HOUR_FORMAT_H || bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_Hb) { + hourChar = CAP_H; + } else if (bestAllowed == ALLOWED_HOUR_FORMAT_K || bestAllowed == ALLOWED_HOUR_FORMAT_KB || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) { + hourChar = CAP_K; + } else if (bestAllowed == ALLOWED_HOUR_FORMAT_k) { + hourChar = LOW_K; + } + // in #13183 just add b/B to skeleton, no longer need to set special flags + if (bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_hB || bestAllowed == ALLOWED_HOUR_FORMAT_KB) { + dayPeriodChar = CAP_B; + } else if (bestAllowed == ALLOWED_HOUR_FORMAT_Hb || bestAllowed == ALLOWED_HOUR_FORMAT_hb || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) { + dayPeriodChar = LOW_B; + } + } + if (hourChar==CAP_H || hourChar==LOW_K) { + dayPeriodLen = 0; + } + while (dayPeriodLen-- > 0) { + patternFormMapped.append(dayPeriodChar); + } + while (hourLen-- > 0) { + patternFormMapped.append(hourChar); + } + } else if (patChr == CAP_J) { + // Get pattern for skeleton with H, then replace H or k + // with fDefaultHourFormatChar (if different) + patternFormMapped.append(CAP_H); + *flags |= kDTPGSkeletonUsesCapJ; + } else { + patternFormMapped.append(patChr); + } + } + } + return patternFormMapped; +} + UnicodeString DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern, const UnicodeString& skeleton, @@ -812,9 +1268,16 @@ UnicodeString DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern, const UnicodeString& skeleton, UDateTimePatternMatchOptions options, - UErrorCode& /*status*/) { + UErrorCode& status) { + if (U_FAILURE(status)) { + return UnicodeString(); + } + if (U_FAILURE(internalErrorCode)) { + status = internalErrorCode; + return UnicodeString(); + } dtMatcher->set(skeleton, fp); - UnicodeString result = adjustFieldTypes(pattern, NULL, FALSE, options); + UnicodeString result = adjustFieldTypes(pattern, nullptr, kDTPGNoFlags, options); return result; } @@ -831,12 +1294,15 @@ DateTimePatternGenerator::getDecimal() const { } void -DateTimePatternGenerator::addCanonicalItems() { +DateTimePatternGenerator::addCanonicalItems(UErrorCode& status) { + if (U_FAILURE(status)) { return; } UnicodeString conflictingPattern; - UErrorCode status = U_ZERO_ERROR; for (int32_t i=0; i 0) { + addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status); + } + if (U_FAILURE(status)) { return; } } } @@ -854,23 +1320,44 @@ DateTimePatternGenerator::getDateTimeFormat() const { void DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) { + if (U_FAILURE(status)) { return; } + const UChar *resStr; int32_t resStrLen = 0; - Calendar* fCalendar = Calendar::createInstance(locale, status); - CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status); - UResourceBundle *dateTimePatterns = calData.getByKey(DT_DateTimePatternsTag, status); - if (U_FAILURE(status)) return; + LocalPointer fCalendar(Calendar::createInstance(locale, status), status); + if (U_FAILURE(status)) { return; } + + LocalUResourceBundlePointer calData(ures_open(nullptr, locale.getBaseName(), &status)); + if (U_FAILURE(status)) { return; } + ures_getByKey(calData.getAlias(), DT_DateTimeCalendarTag, calData.getAlias(), &status); + if (U_FAILURE(status)) { return; } + + LocalUResourceBundlePointer dateTimePatterns; + if (fCalendar->getType() != nullptr && *fCalendar->getType() != '\0' + && uprv_strcmp(fCalendar->getType(), DT_DateTimeGregorianTag) != 0) { + dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), fCalendar->getType(), + nullptr, &status)); + ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag, + dateTimePatterns.getAlias(), &status); + } + + if (dateTimePatterns.isNull() || status == U_MISSING_RESOURCE_ERROR) { + status = U_ZERO_ERROR; + dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), DT_DateTimeGregorianTag, + dateTimePatterns.orphan(), &status)); + ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag, + dateTimePatterns.getAlias(), &status); + } + if (U_FAILURE(status)) { return; } - if (ures_getSize(dateTimePatterns) <= DateFormat::kDateTime) + if (ures_getSize(dateTimePatterns.getAlias()) <= DateFormat::kDateTime) { status = U_INVALID_FORMAT_ERROR; return; } - resStr = ures_getStringByIndex(dateTimePatterns, (int32_t)DateFormat::kDateTime, &resStrLen, &status); + resStr = ures_getStringByIndex(dateTimePatterns.getAlias(), (int32_t)DateFormat::kDateTime, &resStrLen, &status); setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen)); - - delete fCalendar; } void @@ -890,7 +1377,12 @@ DateTimePatternGenerator::addPattern( UnicodeString &conflictingPattern, UErrorCode& status) { - return addPatternWithSkeleton(pattern, NULL, override, conflictingPattern, status); + if (U_FAILURE(internalErrorCode)) { + status = internalErrorCode; + return UDATPG_NO_CONFLICT; + } + + return addPatternWithSkeleton(pattern, nullptr, override, conflictingPattern, status); } // For DateTimePatternGenerator::addPatternWithSkeleton - @@ -911,13 +1403,17 @@ DateTimePatternGenerator::addPatternWithSkeleton( UnicodeString& conflictingPattern, UErrorCode& status) { + if (U_FAILURE(internalErrorCode)) { + status = internalErrorCode; + return UDATPG_NO_CONFLICT; + } UnicodeString basePattern; PtnSkeleton skeleton; UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT; DateTimeMatcher matcher; - if ( skeletonToUse == NULL ) { + if ( skeletonToUse == nullptr ) { matcher.set(pattern, fp, skeleton); matcher.getBasePattern(basePattern); } else { @@ -933,7 +1429,7 @@ DateTimePatternGenerator::addPatternWithSkeleton( // availableFormats items from root, which should not override any previous entry with the same base. UBool entryHadSpecifiedSkeleton; const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton); - if (duplicatePattern != NULL && (!entryHadSpecifiedSkeleton || (skeletonToUse != NULL && !override))) { + if (duplicatePattern != nullptr && (!entryHadSpecifiedSkeleton || (skeletonToUse != nullptr && !override))) { conflictingStatus = UDATPG_BASE_CONFLICT; conflictingPattern = *duplicatePattern; if (!override) { @@ -944,16 +1440,16 @@ DateTimePatternGenerator::addPatternWithSkeleton( // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for // the previously-specified conflicting item. - const PtnSkeleton* entrySpecifiedSkeleton = NULL; + const PtnSkeleton* entrySpecifiedSkeleton = nullptr; duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton); - if (duplicatePattern != NULL ) { + if (duplicatePattern != nullptr ) { conflictingStatus = UDATPG_CONFLICT; conflictingPattern = *duplicatePattern; - if (!override || (skeletonToUse != NULL && entrySpecifiedSkeleton != NULL)) { + if (!override || (skeletonToUse != nullptr && entrySpecifiedSkeleton != nullptr)) { return conflictingStatus; } } - patternMap->add(basePattern, skeleton, pattern, skeletonToUse != NULL, status); + patternMap->add(basePattern, skeleton, pattern, skeletonToUse != nullptr, status); if(U_FAILURE(status)) { return conflictingStatus; } @@ -973,9 +1469,23 @@ DateTimePatternGenerator::getAppendFormatNumber(const char* field) const { } UDateTimePatternField -DateTimePatternGenerator::getAppendNameNumber(const char* field) const { +DateTimePatternGenerator::getFieldAndWidthIndices(const char* key, UDateTimePGDisplayWidth* widthP) const { + char cldrFieldKey[UDATPG_FIELD_KEY_MAX + 1]; + uprv_strncpy(cldrFieldKey, key, UDATPG_FIELD_KEY_MAX); + cldrFieldKey[UDATPG_FIELD_KEY_MAX]=0; // ensure termination + *widthP = UDATPG_WIDE; + char* hyphenPtr = uprv_strchr(cldrFieldKey, '-'); + if (hyphenPtr) { + for (int32_t i=UDATPG_WIDTH_COUNT-1; i>0; --i) { + if (uprv_strcmp(CLDR_FIELD_WIDTH[i], hyphenPtr)==0) { + *widthP=(UDateTimePGDisplayWidth)i; + break; + } + } + *hyphenPtr = 0; // now delete width portion of key + } for (int32_t i=0; iset(pattern); @@ -1045,9 +1558,12 @@ DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern, } const dtTypeElem *row = &dtTypes[canonicalIndex]; int32_t typeValue = row->field; - if (fixFractionalSeconds && typeValue == UDATPG_SECOND_FIELD) { - UnicodeString newField=dtMatcher->skeleton.original[UDATPG_FRACTIONAL_SECOND_FIELD]; - field = field + decimal + newField; + + // handle day periods - with #13183, no longer need special handling here, integrated with normal types + + if ((flags & kDTPGFixFractionalSeconds) != 0 && typeValue == UDATPG_SECOND_FIELD) { + field += decimal; + dtMatcher->skeleton.original.appendFieldTo(UDATPG_FRACTIONAL_SECOND_FIELD, field); } else if (dtMatcher->skeleton.type[typeValue]!=0) { // Here: // - "reqField" is the field from the originally requested skeleton, with length @@ -1071,9 +1587,9 @@ DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern, // a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen. // b) The pattern field is numeric and the skeleton field is not, or vice versa. - UnicodeString reqField = dtMatcher->skeleton.original[typeValue]; - int32_t reqFieldLen = reqField.length(); - if (reqField.charAt(0) == CAP_E && reqFieldLen < 3) + UChar reqFieldChar = dtMatcher->skeleton.original.getFieldChar(typeValue); + int32_t reqFieldLen = dtMatcher->skeleton.original.getFieldLength(typeValue); + if (reqFieldChar == CAP_E && reqFieldLen < 3) reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e int32_t adjFieldLen = reqFieldLen; if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) || @@ -1081,8 +1597,7 @@ DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern, (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) { adjFieldLen = field.length(); } else if (specifiedSkeleton) { - UnicodeString skelField = specifiedSkeleton->original[typeValue]; - int32_t skelFieldLen = skelField.length(); + int32_t skelFieldLen = specifiedSkeleton->original.getFieldLength(typeValue); UBool patFieldIsNumeric = (row->type > 0); UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0); if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) { @@ -1090,12 +1605,36 @@ DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern, adjFieldLen = field.length(); } } - UChar c = (typeValue!= UDATPG_HOUR_FIELD && typeValue!= UDATPG_MONTH_FIELD && - typeValue!= UDATPG_WEEKDAY_FIELD && (typeValue!= UDATPG_YEAR_FIELD || reqField.charAt(0)==CAP_Y))? - reqField.charAt(0): field.charAt(0); + UChar c = (typeValue!= UDATPG_HOUR_FIELD + && typeValue!= UDATPG_MONTH_FIELD + && typeValue!= UDATPG_WEEKDAY_FIELD + && (typeValue!= UDATPG_YEAR_FIELD || reqFieldChar==CAP_Y)) + ? reqFieldChar + : field.charAt(0); + if (typeValue == UDATPG_HOUR_FIELD && (flags & kDTPGSkeletonUsesCapJ) != 0) { + c = fDefaultHourFormatChar; + switch (options & UADATPG_FORCE_HOUR_CYCLE_MASK) { + case UADATPG_FORCE_12_HOUR_CYCLE: + if (c == CAP_H || c == LOW_K) { + // Have 24-hour cycle, change to 12-hour cycle. + // Should have better way to pick 'h' or 'K'. + c = LOW_H; + }; + break; + case UADATPG_FORCE_24_HOUR_CYCLE: + if (c == LOW_H || c == CAP_K) { + // Have 12-hour cycle, change to 24-hour cycle. + // Should have better way to pick 'H' or 'k'. + c = CAP_H; + }; + break; + default: + break; + } + } field.remove(); - for (int32_t i=adjFieldLen; i>0; --i) { - field+=c; + for (int32_t j=adjFieldLen; j>0; --j) { + field += c; } } newPattern+=field; @@ -1105,15 +1644,22 @@ DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern, } UnicodeString -DateTimePatternGenerator::getBestAppending(int32_t missingFields, UDateTimePatternMatchOptions options) { +DateTimePatternGenerator::getBestAppending(int32_t missingFields, int32_t flags, UErrorCode &status, UDateTimePatternMatchOptions options) { + if (U_FAILURE(status)) { + return UnicodeString(); + } UnicodeString resultPattern, tempPattern; - UErrorCode err=U_ZERO_ERROR; + const UnicodeString* tempPatternPtr; int32_t lastMissingFieldMask=0; if (missingFields!=0) { resultPattern=UnicodeString(); - const PtnSkeleton* specifiedSkeleton=NULL; - tempPattern = *getBestRaw(*dtMatcher, missingFields, distanceInfo, &specifiedSkeleton); - resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options); + const PtnSkeleton* specifiedSkeleton=nullptr; + tempPatternPtr = getBestRaw(*dtMatcher, missingFields, distanceInfo, status, &specifiedSkeleton); + if (U_FAILURE(status)) { + return UnicodeString(); + } + tempPattern = *tempPatternPtr; + resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options); if ( distanceInfo->missingFieldMask==0 ) { return resultPattern; } @@ -1123,24 +1669,31 @@ DateTimePatternGenerator::getBestAppending(int32_t missingFields, UDateTimePatte } if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) && ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) { - resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, TRUE, options); + resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, flags | kDTPGFixFractionalSeconds, options); distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK; continue; } int32_t startingMask = distanceInfo->missingFieldMask; - tempPattern = *getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, &specifiedSkeleton); - tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options); + tempPatternPtr = getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, status, &specifiedSkeleton); + if (U_FAILURE(status)) { + return UnicodeString(); + } + tempPattern = *tempPatternPtr; + tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options); int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask; int32_t topField=getTopBitNumber(foundMask); - UnicodeString appendName; - getAppendName((UDateTimePatternField)topField, appendName); - const Formattable formatPattern[] = { - resultPattern, - tempPattern, - appendName - }; - UnicodeString emptyStr; - resultPattern = MessageFormat::format(appendItemFormats[topField], formatPattern, 3, emptyStr, err); + + if (appendItemFormats[topField].length() != 0) { + UnicodeString appendName; + getAppendName((UDateTimePatternField)topField, appendName); + const UnicodeString *values[3] = { + &resultPattern, + &tempPattern, + &appendName + }; + SimpleFormatter(appendItemFormats[topField], 2, 3, status). + formatAndReplace(values, 3, resultPattern, nullptr, 0, status); + } lastMissingFieldMask = distanceInfo->missingFieldMask; } } @@ -1148,7 +1701,7 @@ DateTimePatternGenerator::getBestAppending(int32_t missingFields, UDateTimePatte } int32_t -DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) { +DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) const { if ( foundMask==0 ) { return 0; } @@ -1177,22 +1730,21 @@ DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const { void DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) { - - if (other == NULL) { + if (other == nullptr || U_FAILURE(status)) { return; } - if (fAvailableFormatKeyHash != NULL) { + if (fAvailableFormatKeyHash != nullptr) { delete fAvailableFormatKeyHash; - fAvailableFormatKeyHash = NULL; + fAvailableFormatKeyHash = nullptr; } initHashtable(status); if(U_FAILURE(status)){ return; } - int32_t pos = -1; - const UHashElement* elem = NULL; + int32_t pos = UHASH_FIRST; + const UHashElement* elem = nullptr; // walk through the hash table and create a deep clone - while((elem = other->nextElement(pos))!= NULL){ + while((elem = other->nextElement(pos))!= nullptr){ const UHashTok otherKeyTok = elem->key; UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer; fAvailableFormatKeyHash->puti(*otherKey, 1, status); @@ -1204,8 +1756,17 @@ DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) { StringEnumeration* DateTimePatternGenerator::getSkeletons(UErrorCode& status) const { - StringEnumeration* skeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status); - return skeletonEnumerator; + if (U_FAILURE(status)) { + return nullptr; + } + if (U_FAILURE(internalErrorCode)) { + status = internalErrorCode; + return nullptr; + } + LocalPointer skeletonEnumerator( + new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status), status); + + return U_SUCCESS(status) ? skeletonEnumerator.orphan() : nullptr; } const UnicodeString& @@ -1216,47 +1777,70 @@ DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) c return emptyString; } curElem = patternMap->getHeader(skeleton.charAt(0)); - while ( curElem != NULL ) { + while ( curElem != nullptr ) { if ( curElem->skeleton->getSkeleton()==skeleton ) { return curElem->pattern; } - curElem=curElem->next; + curElem = curElem->next.getAlias(); } return emptyString; } StringEnumeration* DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const { - StringEnumeration* baseSkeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status); - return baseSkeletonEnumerator; + if (U_FAILURE(status)) { + return nullptr; + } + if (U_FAILURE(internalErrorCode)) { + status = internalErrorCode; + return nullptr; + } + LocalPointer baseSkeletonEnumerator( + new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status), status); + + return U_SUCCESS(status) ? baseSkeletonEnumerator.orphan() : nullptr; } StringEnumeration* DateTimePatternGenerator::getRedundants(UErrorCode& status) { - StringEnumeration* output = new DTRedundantEnumeration(); + if (U_FAILURE(status)) { return nullptr; } + if (U_FAILURE(internalErrorCode)) { + status = internalErrorCode; + return nullptr; + } + LocalPointer output(new DTRedundantEnumeration(), status); + if (U_FAILURE(status)) { return nullptr; } const UnicodeString *pattern; - PatternMapIterator it; + PatternMapIterator it(status); + if (U_FAILURE(status)) { return nullptr; } + for (it.set(*patternMap); it.hasNext(); ) { DateTimeMatcher current = it.next(); pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton())); if ( isCanonicalItem(*pattern) ) { continue; } - if ( skipMatcher == NULL ) { + if ( skipMatcher == nullptr ) { skipMatcher = new DateTimeMatcher(current); + if (skipMatcher == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } } else { *skipMatcher = current; } UnicodeString trial = getBestPattern(current.getPattern(), status); + if (U_FAILURE(status)) { return nullptr; } if (trial == *pattern) { - ((DTRedundantEnumeration *)output)->add(*pattern, status); + ((DTRedundantEnumeration *)output.getAlias())->add(*pattern, status); + if (U_FAILURE(status)) { return nullptr; } } if (current.equals(skipMatcher)) { continue; } } - return output; + return output.orphan(); } UBool @@ -1280,45 +1864,54 @@ DateTimePatternGenerator::clone() const { PatternMap::PatternMap() { for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) { - boot[i]=NULL; + boot[i] = nullptr; } isDupAllowed = TRUE; } void PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) { + if (U_FAILURE(status)) { + return; + } this->isDupAllowed = other.isDupAllowed; - for (int32_t bootIndex=0; bootIndexbasePattern, otherElem->pattern))==NULL) { - // out of memory - status = U_MEMORY_ALLOCATION_ERROR; - return; + while (otherElem != nullptr) { + LocalPointer newElem(new PtnElem(otherElem->basePattern, otherElem->pattern), status); + if (U_FAILURE(status)) { + return; // out of memory } - if ( this->boot[bootIndex]== NULL ) { - this->boot[bootIndex] = curElem; - } - if ((curElem->skeleton=new PtnSkeleton(*(otherElem->skeleton))) == NULL ) { - // out of memory - status = U_MEMORY_ALLOCATION_ERROR; - return; + newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(*(otherElem->skeleton)), status); + if (U_FAILURE(status)) { + return; // out of memory } + newElem->skeletonWasSpecified = otherElem->skeletonWasSpecified; + + // Release ownership from the LocalPointer of the PtnElem object. + // The PtnElem will now be owned by either the boot (for the first entry in the linked-list) + // or owned by the previous PtnElem object in the linked-list. + curElem = newElem.orphan(); - if (prevElem!=NULL) { - prevElem->next=curElem; + if (this->boot[bootIndex] == nullptr) { + this->boot[bootIndex] = curElem; + } else { + if (prevElem != nullptr) { + prevElem->next.adoptInstead(curElem); + } else { + UPRV_UNREACHABLE; + } } - curElem->next=NULL; prevElem = curElem; - otherElem = otherElem->next; + otherElem = otherElem->next.getAlias(); } } } PtnElem* -PatternMap::getHeader(UChar baseChar) { +PatternMap::getHeader(UChar baseChar) const { PtnElem* curElem; if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) { @@ -1329,7 +1922,7 @@ PatternMap::getHeader(UChar baseChar) { curElem = boot[26+baseChar-LOW_A]; } else { - return NULL; + return nullptr; } } return curElem; @@ -1337,9 +1930,9 @@ PatternMap::getHeader(UChar baseChar) { PatternMap::~PatternMap() { for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) { - if (boot[i]!=NULL ) { + if (boot[i] != nullptr ) { delete boot[i]; - boot[i]=NULL; + boot[i] = nullptr; } } } // PatternMap destructor @@ -1368,39 +1961,45 @@ PatternMap::add(const UnicodeString& basePattern, } } - if (baseElem == NULL) { - if ((curElem = new PtnElem(basePattern, value)) == NULL ) { - // out of memory - status = U_MEMORY_ALLOCATION_ERROR; - return; + if (baseElem == nullptr) { + LocalPointer newElem(new PtnElem(basePattern, value), status); + if (U_FAILURE(status)) { + return; // out of memory + } + newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status); + if (U_FAILURE(status)) { + return; // out of memory } + newElem->skeletonWasSpecified = skeletonWasSpecified; if (baseChar >= LOW_A) { - boot[26 + (baseChar-LOW_A)] = curElem; + boot[26 + (baseChar - LOW_A)] = newElem.orphan(); // the boot array now owns the PtnElem. } else { - boot[baseChar-CAP_A] = curElem; + boot[baseChar - CAP_A] = newElem.orphan(); // the boot array now owns the PtnElem. } - curElem->skeleton = new PtnSkeleton(skeleton); - curElem->skeletonWasSpecified = skeletonWasSpecified; } - if ( baseElem != NULL ) { + if ( baseElem != nullptr ) { curElem = getDuplicateElem(basePattern, skeleton, baseElem); - if (curElem == NULL) { + if (curElem == nullptr) { // add new element to the list. curElem = baseElem; - while( curElem -> next != NULL ) + while( curElem -> next != nullptr ) { - curElem = curElem->next; + curElem = curElem->next.getAlias(); } - if ((curElem->next = new PtnElem(basePattern, value)) == NULL ) { - // out of memory - status = U_MEMORY_ALLOCATION_ERROR; - return; + + LocalPointer newElem(new PtnElem(basePattern, value), status); + if (U_FAILURE(status)) { + return; // out of memory } - curElem=curElem->next; - curElem->skeleton = new PtnSkeleton(skeleton); - curElem->skeletonWasSpecified = skeletonWasSpecified; + newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status); + if (U_FAILURE(status)) { + return; // out of memory + } + newElem->skeletonWasSpecified = skeletonWasSpecified; + curElem->next.adoptInstead(newElem.orphan()); + curElem = curElem->next.getAlias(); } else { // Pattern exists in the list already. @@ -1418,11 +2017,11 @@ PatternMap::add(const UnicodeString& basePattern, // Find the pattern from the given basePattern string. const UnicodeString * -PatternMap::getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeletonWasSpecified) { // key to search for +PatternMap::getPatternFromBasePattern(const UnicodeString& basePattern, UBool& skeletonWasSpecified) const { // key to search for PtnElem *curElem; - if ((curElem=getHeader(basePattern.charAt(0)))==NULL) { - return NULL; // no match + if ((curElem=getHeader(basePattern.charAt(0)))==nullptr) { + return nullptr; // no match } do { @@ -1430,10 +2029,10 @@ PatternMap::getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeleto skeletonWasSpecified = curElem->skeletonWasSpecified; return &(curElem->pattern); } - curElem=curElem->next; - }while (curElem != NULL); + curElem = curElem->next.getAlias(); + } while (curElem != nullptr); - return NULL; + return nullptr; } // PatternMap::getFromBasePattern @@ -1444,86 +2043,69 @@ PatternMap::getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeleto // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL), // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily. const UnicodeString * -PatternMap::getPatternFromSkeleton(PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) { // key to search for +PatternMap::getPatternFromSkeleton(const PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) const { // key to search for PtnElem *curElem; if (specifiedSkeletonPtr) { - *specifiedSkeletonPtr = NULL; + *specifiedSkeletonPtr = nullptr; } // find boot entry - UChar baseChar='\0'; - for (int32_t i=0; iskeleton->original[i].compare(skeleton.original[i]) != 0 ) - { - break; - } - } + UBool equal; + if (specifiedSkeletonPtr != nullptr) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original + equal = curElem->skeleton->original == skeleton.original; } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal - for (i=0; iskeleton->baseOriginal[i].compare(skeleton.baseOriginal[i]) != 0 ) - { - break; - } - } + equal = curElem->skeleton->baseOriginal == skeleton.baseOriginal; } - if (i == UDATPG_FIELD_COUNT) { + if (equal) { if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) { - *specifiedSkeletonPtr = curElem->skeleton; + *specifiedSkeletonPtr = curElem->skeleton.getAlias(); } return &(curElem->pattern); } - curElem=curElem->next; - }while (curElem != NULL); + curElem = curElem->next.getAlias(); + } while (curElem != nullptr); - return NULL; + return nullptr; } UBool -PatternMap::equals(const PatternMap& other) { +PatternMap::equals(const PatternMap& other) const { if ( this==&other ) { return TRUE; } - for (int32_t bootIndex=0; bootIndexbasePattern != otherElem->basePattern) || (myElem->pattern != otherElem->pattern) ) { return FALSE; } - if ((myElem->skeleton!=otherElem->skeleton)&& + if ((myElem->skeleton.getAlias() != otherElem->skeleton.getAlias()) && !myElem->skeleton->equals(*(otherElem->skeleton))) { return FALSE; } - myElem = myElem->next; - otherElem=otherElem->next; + myElem = myElem->next.getAlias(); + otherElem = otherElem->next.getAlias(); } } return TRUE; @@ -1535,21 +2117,21 @@ PtnElem* PatternMap::getDuplicateElem( const UnicodeString &basePattern, const PtnSkeleton &skeleton, - PtnElem *baseElem) { + PtnElem *baseElem) { PtnElem *curElem; - if ( baseElem == (PtnElem *)NULL ) { - return (PtnElem*)NULL; + if ( baseElem == nullptr ) { + return nullptr; } else { curElem = baseElem; } do { if ( basePattern.compare(curElem->basePattern)==0 ) { - UBool isEqual=TRUE; - for (int32_t i=0; iskeleton->type[i] != skeleton.type[i] ) { - isEqual=FALSE; + isEqual = FALSE; break; } } @@ -1557,11 +2139,11 @@ PatternMap::getDuplicateElem( return curElem; } } - curElem = curElem->next; - } while( curElem != (PtnElem *)NULL ); + curElem = curElem->next.getAlias(); + } while( curElem != nullptr ); // end of the list - return (PtnElem*)NULL; + return nullptr; } // PatternMap::getDuplicateElem @@ -1585,37 +2167,62 @@ void DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) { int32_t i; for (i=0; iset(pattern); for (i=0; i < fp->itemNumber; i++) { - UnicodeString field = fp->items[i]; - if ( field.charAt(0) == LOW_A ) { - continue; // skip 'a' - } + const UnicodeString& value = fp->items[i]; + // don't skip 'a' anymore, dayPeriod handled specially below - if ( fp->isQuoteLiteral(field) ) { + if ( fp->isQuoteLiteral(value) ) { UnicodeString quoteLiteral; fp->getQuoteLiteral(quoteLiteral, &i); continue; } - int32_t canonicalIndex = fp->getCanonicalIndex(field); - if (canonicalIndex < 0 ) { + int32_t canonicalIndex = fp->getCanonicalIndex(value); + if (canonicalIndex < 0) { continue; } const dtTypeElem *row = &dtTypes[canonicalIndex]; - int32_t typeValue = row->field; - skeletonResult.original[typeValue]=field; + int32_t field = row->field; + skeletonResult.original.populate(field, value); UChar repeatChar = row->patternChar; - int32_t repeatCount = row->minLen; // #7930 removes cap at 3 - while (repeatCount-- > 0) { - skeletonResult.baseOriginal[typeValue] += repeatChar; - } - int16_t subTypeValue = row->type; - if ( row->type > 0) { - subTypeValue += field.length(); + int32_t repeatCount = row->minLen; + skeletonResult.baseOriginal.populate(field, repeatChar, repeatCount); + int16_t subField = row->type; + if (row->type > 0) { + U_ASSERT(value.length() < INT16_MAX); + subField += static_cast(value.length()); + } + skeletonResult.type[field] = subField; + } + // #13183, handle special behavior for day period characters (a, b, B) + if (!skeletonResult.original.isFieldEmpty(UDATPG_HOUR_FIELD)) { + if (skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==LOW_H || skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==CAP_K) { + // We have a skeleton with 12-hour-cycle format + if (skeletonResult.original.isFieldEmpty(UDATPG_DAYPERIOD_FIELD)) { + // But we do not have a day period in the skeleton; add the default DAYPERIOD (currently "a") + for (i = 0; dtTypes[i].patternChar != 0; i++) { + if ( dtTypes[i].field == UDATPG_DAYPERIOD_FIELD ) { + // first entry for UDATPG_DAYPERIOD_FIELD + skeletonResult.original.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen); + skeletonResult.baseOriginal.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen); + skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = dtTypes[i].type; + skeletonResult.addedDefaultDayPeriod = TRUE; + break; + } + } + } + } else { + // Skeleton has 24-hour-cycle hour format and has dayPeriod, delete dayPeriod (i.e. ignore it) + skeletonResult.original.clearField(UDATPG_DAYPERIOD_FIELD); + skeletonResult.baseOriginal.clearField(UDATPG_DAYPERIOD_FIELD); + skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = NONE; } - skeletonResult.type[typeValue] = subTypeValue; } copyFrom(skeletonResult); } @@ -1623,28 +2230,18 @@ DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton void DateTimeMatcher::getBasePattern(UnicodeString &result ) { result.remove(); // Reset the result first. - for (int32_t i=0; iskeleton.type[i]=newSkeleton.type[i]; - this->skeleton.original[i]=newSkeleton.original[i]; - this->skeleton.baseOriginal[i]=newSkeleton.baseOriginal[i]; - } + skeleton.copyFrom(newSkeleton); } void DateTimeMatcher::copyFrom() { // same as clear - for (int32_t i=0; iskeleton.type[i]=0; - this->skeleton.original[i].remove(); - this->skeleton.baseOriginal[i].remove(); - } + skeleton.clear(); } UBool DateTimeMatcher::equals(const DateTimeMatcher* other) const { - if (other==NULL) { - return FALSE; - } - for (int32_t i=0; iskeleton.original[i]!=other->skeleton.original[i] ) { - return FALSE; - } - } - return TRUE; + if (other==nullptr) { return FALSE; } + return skeleton.original == other->skeleton.original; } int32_t -DateTimeMatcher::getFieldMask() { - int32_t result=0; +DateTimeMatcher::getFieldMask() const { + int32_t result = 0; for (int32_t i=0; i= pattern.length()) { return DONE; } @@ -1759,10 +2341,10 @@ FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t void FormatParser::set(const UnicodeString& pattern) { - int32_t startPos=0; - TokenStatus result=START; - int32_t len=0; - itemNumber =0; + int32_t startPos = 0; + TokenStatus result = START; + int32_t len = 0; + itemNumber = 0; do { result = setTokens( pattern, startPos, &len ); @@ -1793,7 +2375,7 @@ FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) { } int32_t i = 0; int32_t bestRow = -1; - while (dtTypes[i].patternChar != '\0') { + while (dtTypes[i].patternChar != 0x0000) { if ( dtTypes[i].patternChar != ch ) { ++i; continue; @@ -1812,15 +2394,15 @@ FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) { } UBool -FormatParser::isQuoteLiteral(const UnicodeString& s) const { - return (UBool)(s.charAt(0)==SINGLE_QUOTE); +FormatParser::isQuoteLiteral(const UnicodeString& s) { + return (UBool)(s.charAt(0) == SINGLE_QUOTE); } -// This function aussumes the current itemIndex points to the quote literal. +// This function assumes the current itemIndex points to the quote literal. // Please call isQuoteLiteral prior to this function. void FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) { - int32_t i=*itemIndex; + int32_t i = *itemIndex; quote.remove(); if (items[i].charAt(0)==SINGLE_QUOTE) { @@ -1849,7 +2431,7 @@ FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) { } UBool -FormatParser::isPatternSeparator(UnicodeString& field) { +FormatParser::isPatternSeparator(const UnicodeString& field) const { for (int32_t i=0; iskeleton; + return nodePtr->skeleton.getAlias(); } } UBool -PatternMapIterator::hasNext() { - int32_t headIndex=bootIndex; - PtnElem *curPtr=nodePtr; +PatternMapIterator::hasNext() const { + int32_t headIndex = bootIndex; + PtnElem *curPtr = nodePtr; - if (patternMap==NULL) { + if (patternMap==nullptr) { return FALSE; } while ( headIndex < MAX_PATTERN_ENTRIES ) { - if ( curPtr != NULL ) { - if ( curPtr->next != NULL ) { + if ( curPtr != nullptr ) { + if ( curPtr->next != nullptr ) { return TRUE; } else { headIndex++; - curPtr=NULL; + curPtr=nullptr; continue; } } else { - if ( patternMap->boot[headIndex] != NULL ) { + if ( patternMap->boot[headIndex] != nullptr ) { return TRUE; } else { @@ -1926,7 +2506,6 @@ PatternMapIterator::hasNext() { continue; } } - } return FALSE; } @@ -1934,19 +2513,19 @@ PatternMapIterator::hasNext() { DateTimeMatcher& PatternMapIterator::next() { while ( bootIndex < MAX_PATTERN_ENTRIES ) { - if ( nodePtr != NULL ) { - if ( nodePtr->next != NULL ) { - nodePtr = nodePtr->next; + if ( nodePtr != nullptr ) { + if ( nodePtr->next != nullptr ) { + nodePtr = nodePtr->next.getAlias(); break; } else { bootIndex++; - nodePtr=NULL; + nodePtr=nullptr; continue; } } else { - if ( patternMap->boot[bootIndex] != NULL ) { + if ( patternMap->boot[bootIndex] != nullptr ) { nodePtr = patternMap->boot[bootIndex]; break; } @@ -1956,7 +2535,7 @@ PatternMapIterator::next() { } } } - if (nodePtr!=NULL) { + if (nodePtr!=nullptr) { matcher->copyFrom(*nodePtr->skeleton); } else { @@ -1965,88 +2544,160 @@ PatternMapIterator::next() { return *matcher; } -PtnSkeleton::PtnSkeleton() { + +SkeletonFields::SkeletonFields() { + // Set initial values to zero + clear(); } +void SkeletonFields::clear() { + uprv_memset(chars, 0, sizeof(chars)); + uprv_memset(lengths, 0, sizeof(lengths)); +} -PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) { - for (int32_t i=0; itype[i]=other.type[i]; - this->original[i]=other.original[i]; - this->baseOriginal[i]=other.baseOriginal[i]; +void SkeletonFields::copyFrom(const SkeletonFields& other) { + uprv_memcpy(chars, other.chars, sizeof(chars)); + uprv_memcpy(lengths, other.lengths, sizeof(lengths)); +} + +void SkeletonFields::clearField(int32_t field) { + chars[field] = 0; + lengths[field] = 0; +} + +UChar SkeletonFields::getFieldChar(int32_t field) const { + return chars[field]; +} + +int32_t SkeletonFields::getFieldLength(int32_t field) const { + return lengths[field]; +} + +void SkeletonFields::populate(int32_t field, const UnicodeString& value) { + populate(field, value.charAt(0), value.length()); +} + +void SkeletonFields::populate(int32_t field, UChar ch, int32_t length) { + chars[field] = (int8_t) ch; + lengths[field] = (int8_t) length; +} + +UBool SkeletonFields::isFieldEmpty(int32_t field) const { + return lengths[field] == 0; +} + +UnicodeString& SkeletonFields::appendTo(UnicodeString& string) const { + for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) { + appendFieldTo(i, string); } + return string; } -UBool -PtnSkeleton::equals(const PtnSkeleton& other) { - for (int32_t i=0; i= 0) { + // for backward compatibility: if DateTimeMatcher.set added a single 'a' that + // was not in the provided skeleton, remove it here before returning skeleton. + result.remove(pos, 1); } return result; } UnicodeString -PtnSkeleton::getBaseSkeleton() { +PtnSkeleton::getBaseSkeleton() const { UnicodeString result; - - for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) { - if (baseOriginal[i].length()!=0) { - result += baseOriginal[i]; - } + result = baseOriginal.appendTo(result); + int32_t pos; + if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) { + // for backward compatibility: if DateTimeMatcher.set added a single 'a' that + // was not in the provided skeleton, remove it here before returning skeleton. + result.remove(pos, 1); } return result; } +UChar +PtnSkeleton::getFirstChar() const { + return baseOriginal.getFirstChar(); +} + PtnSkeleton::~PtnSkeleton() { } PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) : -basePattern(basePat), -skeleton(NULL), -pattern(pat), -next(NULL) + basePattern(basePat), skeleton(nullptr), pattern(pat), next(nullptr) { } PtnElem::~PtnElem() { - - if (next!=NULL) { - delete next; - } - delete skeleton; } -DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum type, UErrorCode& status) { +DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap& patternMap, dtStrEnum type, UErrorCode& status) : fSkeletons(nullptr) { PtnElem *curElem; PtnSkeleton *curSkeleton; UnicodeString s; int32_t bootIndex; pos=0; - fSkeletons = new UVector(status); + fSkeletons.adoptInsteadAndCheckErrorCode(new UVector(status), status); if (U_FAILURE(status)) { - delete fSkeletons; return; } + for (bootIndex=0; bootIndexbasePattern; @@ -2055,32 +2706,36 @@ DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum t s=curElem->pattern; break; case DT_SKELETON: - curSkeleton=curElem->skeleton; + curSkeleton=curElem->skeleton.getAlias(); s=curSkeleton->getSkeleton(); break; } if ( !isCanonicalItem(s) ) { - fSkeletons->addElement(new UnicodeString(s), status); + LocalPointer newElem(new UnicodeString(s), status); + if (U_FAILURE(status)) { + return; + } + fSkeletons->addElement(newElem.getAlias(), status); if (U_FAILURE(status)) { - delete fSkeletons; - fSkeletons = NULL; + fSkeletons.adoptInstead(nullptr); return; } + newElem.orphan(); // fSkeletons vector now owns the UnicodeString. } - curElem = curElem->next; + curElem = curElem->next.getAlias(); } } - if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=NULL) ) { + if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=nullptr) ) { status = U_BUFFER_OVERFLOW_ERROR; } } const UnicodeString* DTSkeletonEnumeration::snext(UErrorCode& status) { - if (U_SUCCESS(status) && pos < fSkeletons->size()) { + if (U_SUCCESS(status) && fSkeletons.isValid() && pos < fSkeletons->size()) { return (const UnicodeString*)fSkeletons->elementAt(pos++); } - return NULL; + return nullptr; } void @@ -2090,7 +2745,7 @@ DTSkeletonEnumeration::reset(UErrorCode& /*status*/) { int32_t DTSkeletonEnumeration::count(UErrorCode& /*status*/) const { - return (fSkeletons==NULL) ? 0 : fSkeletons->size(); + return (fSkeletons.isNull()) ? 0 : fSkeletons->size(); } UBool @@ -2108,44 +2763,45 @@ DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) { DTSkeletonEnumeration::~DTSkeletonEnumeration() { UnicodeString *s; - for (int32_t i=0; isize(); ++i) { - if ((s=(UnicodeString *)fSkeletons->elementAt(i))!=NULL) { - delete s; + if (fSkeletons.isValid()) { + for (int32_t i = 0; i < fSkeletons->size(); ++i) { + if ((s = (UnicodeString *)fSkeletons->elementAt(i)) != nullptr) { + delete s; + } } } - delete fSkeletons; } -DTRedundantEnumeration::DTRedundantEnumeration() { - pos=0; - fPatterns = NULL; +DTRedundantEnumeration::DTRedundantEnumeration() : pos(0), fPatterns(nullptr) { } void DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) { - if (U_FAILURE(status)) return; - if (fPatterns == NULL) { - fPatterns = new UVector(status); + if (U_FAILURE(status)) { return; } + if (fPatterns.isNull()) { + fPatterns.adoptInsteadAndCheckErrorCode(new UVector(status), status); if (U_FAILURE(status)) { - delete fPatterns; - fPatterns = NULL; return; } } - fPatterns->addElement(new UnicodeString(pattern), status); + LocalPointer newElem(new UnicodeString(pattern), status); + if (U_FAILURE(status)) { + return; + } + fPatterns->addElement(newElem.getAlias(), status); if (U_FAILURE(status)) { - delete fPatterns; - fPatterns = NULL; + fPatterns.adoptInstead(nullptr); return; } + newElem.orphan(); // fPatterns now owns the string. } const UnicodeString* DTRedundantEnumeration::snext(UErrorCode& status) { - if (U_SUCCESS(status) && pos < fPatterns->size()) { + if (U_SUCCESS(status) && fPatterns.isValid() && pos < fPatterns->size()) { return (const UnicodeString*)fPatterns->elementAt(pos++); } - return NULL; + return nullptr; } void @@ -2155,11 +2811,11 @@ DTRedundantEnumeration::reset(UErrorCode& /*status*/) { int32_t DTRedundantEnumeration::count(UErrorCode& /*status*/) const { - return (fPatterns==NULL) ? 0 : fPatterns->size(); + return (fPatterns.isNull()) ? 0 : fPatterns->size(); } UBool -DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) { +DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) const { if ( item.length() != 1 ) { return FALSE; } @@ -2173,12 +2829,92 @@ DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) { DTRedundantEnumeration::~DTRedundantEnumeration() { UnicodeString *s; - for (int32_t i=0; isize(); ++i) { - if ((s=(UnicodeString *)fPatterns->elementAt(i))!=NULL) { - delete s; + if (fPatterns.isValid()) { + for (int32_t i = 0; i < fPatterns->size(); ++i) { + if ((s = (UnicodeString *)fPatterns->elementAt(i)) != nullptr) { + delete s; + } } + } +} + +/** + * This is a utility function used by the various date formatting classes to determine whether a particular pattern string will produce an all-numeric date. + * It does this by examining the pattern string. If the range between the first d, M, or y and the last d, M, or y in the pattern contains nothing but d, M, y, + * punctuation, whitespace, and Unicode right-to-left marks, and it doesn't contain more than two Ms in a row, it's considered to have a "numeric core"-- + * that is, the part of the pattern that generates a date (minus fields like the day of the week and the era) produces an all-numeric date. + */ +extern UBool datePatternHasNumericCore(const UnicodeString& datePattern) { + StringCharacterIterator it = StringCharacterIterator(datePattern); + int32_t coreStart = -1; + int32_t coreEnd = -1; + int32_t firstLetterAfterCoreStart = -1; + int32_t numMs = 0; + UBool sawD = FALSE, sawY = FALSE; + for (UChar c = it.first(); it.hasNext(); c = it.next()) { + switch (c) { + case u'y': case u'Y': case u'r': case u'u': + case u'M': case u'L': case u'd': + if (coreStart == -1) { + coreStart = it.getIndex(); + } + coreEnd = it.getIndex(); + switch (c) { + case u'y': case u'Y': case u'r': case u'u': + sawY = TRUE; + break; + case u'M': case u'L': + // if the pattern contains more than 2 M's, the month is a word, not a number, which means + // we don't have a numeric core + ++numMs; + if (numMs > 2) { + return FALSE; + } + break; + case 'd': + sawD = TRUE; + break; + default: + break; + } + break; + default: + if (u_isalpha(c)) { + if (coreStart != -1 && firstLetterAfterCoreStart == -1) { + firstLetterAfterCoreStart = it.getIndex(); + } + } else if (!u_isspace(c) && !u_ispunct(c) && c != u'\u200f') { + // the numeric core must contain nothing but d, M, y, whitespace, punctuation, and the Unicode right-to-left mark + return FALSE; + } + break; + } + } + + // if we didn't find d, M, or y in the pattern, return FALSE + if (coreStart < 0 || coreEnd < 0) { + return FALSE; } - delete fPatterns; + + // if there's quoted literal text anywhere in the pattern, whether in the "core" or not, treat it as though + // we don't have a numeric core + if (datePattern.indexOf(u'\'') != -1) { + return FALSE; + } + + // if we found a letter other than d, M, or y between the first d, M, or y and the last one, + // we don't have a numeric core + if (firstLetterAfterCoreStart != -1 && firstLetterAfterCoreStart < coreEnd) { + return FALSE; + } + + // if the format contains only one numeric field (out of d, M, or y), we don't count it as a numeric core + if (((numMs > 0) ? 1 : 0) + (sawY ? 1 : 0) + (sawD ? 1 : 0) <= 1) { + return FALSE; + } + + // if we get to here, we have a numeric core + return TRUE; } U_NAMESPACE_END