dtptngen.cpp (114653B)
1 // © 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 3 /* 4 ******************************************************************************* 5 * Copyright (C) 2007-2016, International Business Machines Corporation and 6 * others. All Rights Reserved. 7 ******************************************************************************* 8 * 9 * File DTPTNGEN.CPP 10 * 11 ******************************************************************************* 12 */ 13 14 #include "unicode/utypes.h" 15 #if !UCONFIG_NO_FORMATTING 16 17 #include "unicode/datefmt.h" 18 #include "unicode/decimfmt.h" 19 #include "unicode/dtfmtsym.h" 20 #include "unicode/dtptngen.h" 21 #include "unicode/localpointer.h" 22 #include "unicode/simpleformatter.h" 23 #include "unicode/smpdtfmt.h" 24 #include "unicode/udat.h" 25 #include "unicode/udatpg.h" 26 #include "unicode/uniset.h" 27 #include "unicode/uloc.h" 28 #include "unicode/ures.h" 29 #include "unicode/ustring.h" 30 #include "unicode/rep.h" 31 #include "unicode/region.h" 32 #include "cpputils.h" 33 #include "mutex.h" 34 #include "umutex.h" 35 #include "cmemory.h" 36 #include "cstring.h" 37 #include "locbased.h" 38 #include "hash.h" 39 #include "uhash.h" 40 #include "ulocimp.h" 41 #include "uresimp.h" 42 #include "ulocimp.h" 43 #include "dtptngen_impl.h" 44 #include "ucln_in.h" 45 #include "charstr.h" 46 #include "uassert.h" 47 48 #if U_CHARSET_FAMILY==U_EBCDIC_FAMILY 49 /** 50 * If we are on EBCDIC, use an iterator which will 51 * traverse the bundles in ASCII order. 52 */ 53 #define U_USE_ASCII_BUNDLE_ITERATOR 54 #define U_SORT_ASCII_BUNDLE_ITERATOR 55 #endif 56 57 #if defined(U_USE_ASCII_BUNDLE_ITERATOR) 58 59 #include "unicode/ustring.h" 60 #include "uarrsort.h" 61 62 struct UResAEntry { 63 char16_t *key; 64 UResourceBundle *item; 65 }; 66 67 struct UResourceBundleAIterator { 68 UResourceBundle *bund; 69 UResAEntry *entries; 70 int32_t num; 71 int32_t cursor; 72 }; 73 74 /* Must be C linkage to pass function pointer to the sort function */ 75 76 U_CDECL_BEGIN 77 78 static int32_t U_CALLCONV 79 ures_a_codepointSort(const void *context, const void *left, const void *right) { 80 //CompareContext *cmp=(CompareContext *)context; 81 return u_strcmp(((const UResAEntry *)left)->key, 82 ((const UResAEntry *)right)->key); 83 } 84 85 U_CDECL_END 86 87 static void ures_a_open(UResourceBundleAIterator *aiter, UResourceBundle *bund, UErrorCode *status) { 88 if(U_FAILURE(*status)) { 89 return; 90 } 91 aiter->bund = bund; 92 aiter->num = ures_getSize(aiter->bund); 93 aiter->cursor = 0; 94 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR) 95 aiter->entries = nullptr; 96 #else 97 aiter->entries = (UResAEntry*)uprv_malloc(sizeof(UResAEntry)*aiter->num); 98 for(int i=0;i<aiter->num;i++) { 99 aiter->entries[i].item = ures_getByIndex(aiter->bund, i, nullptr, status); 100 const char *akey = ures_getKey(aiter->entries[i].item); 101 int32_t len = uprv_strlen(akey)+1; 102 aiter->entries[i].key = (char16_t*)uprv_malloc(len*sizeof(char16_t)); 103 u_charsToUChars(akey, aiter->entries[i].key, len); 104 } 105 uprv_sortArray(aiter->entries, aiter->num, sizeof(UResAEntry), ures_a_codepointSort, nullptr, true, status); 106 #endif 107 } 108 109 static void ures_a_close(UResourceBundleAIterator *aiter) { 110 #if defined(U_SORT_ASCII_BUNDLE_ITERATOR) 111 for(int i=0;i<aiter->num;i++) { 112 uprv_free(aiter->entries[i].key); 113 ures_close(aiter->entries[i].item); 114 } 115 #endif 116 } 117 118 static const char16_t *ures_a_getNextString(UResourceBundleAIterator *aiter, int32_t *len, const char **key, UErrorCode *err) { 119 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR) 120 return ures_getNextString(aiter->bund, len, key, err); 121 #else 122 if(U_FAILURE(*err)) return nullptr; 123 UResourceBundle *item = aiter->entries[aiter->cursor].item; 124 const char16_t* ret = ures_getString(item, len, err); 125 *key = ures_getKey(item); 126 aiter->cursor++; 127 return ret; 128 #endif 129 } 130 131 132 #endif 133 134 135 U_NAMESPACE_BEGIN 136 137 // ***************************************************************************** 138 // class DateTimePatternGenerator 139 // ***************************************************************************** 140 static const char16_t Canonical_Items[] = { 141 // GyQMwWEDFdaHmsSv 142 CAP_G, LOW_Y, CAP_Q, CAP_M, LOW_W, CAP_W, CAP_E, 143 CAP_D, CAP_F, LOW_D, LOW_A, // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J 144 CAP_H, LOW_M, LOW_S, CAP_S, LOW_V, 0 145 }; 146 147 static const dtTypeElem dtTypes[] = { 148 // patternChar, field, type, minLen, weight 149 {CAP_G, UDATPG_ERA_FIELD, DT_SHORT, 1, 3,}, 150 {CAP_G, UDATPG_ERA_FIELD, DT_LONG, 4, 0}, 151 {CAP_G, UDATPG_ERA_FIELD, DT_NARROW, 5, 0}, 152 153 {LOW_Y, UDATPG_YEAR_FIELD, DT_NUMERIC, 1, 20}, 154 {CAP_Y, UDATPG_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 20}, 155 {LOW_U, UDATPG_YEAR_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 20}, 156 {LOW_R, UDATPG_YEAR_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20}, 157 {CAP_U, UDATPG_YEAR_FIELD, DT_SHORT, 1, 3}, 158 {CAP_U, UDATPG_YEAR_FIELD, DT_LONG, 4, 0}, 159 {CAP_U, UDATPG_YEAR_FIELD, DT_NARROW, 5, 0}, 160 161 {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2}, 162 {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0}, 163 {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0}, 164 {CAP_Q, UDATPG_QUARTER_FIELD, DT_NARROW, 5, 0}, 165 {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, 166 {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT - DT_DELTA, 3, 0}, 167 {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG - DT_DELTA, 4, 0}, 168 {LOW_Q, UDATPG_QUARTER_FIELD, DT_NARROW - DT_DELTA, 5, 0}, 169 170 {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2}, 171 {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0}, 172 {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0}, 173 {CAP_M, UDATPG_MONTH_FIELD, DT_NARROW, 5, 0}, 174 {CAP_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, 175 {CAP_L, UDATPG_MONTH_FIELD, DT_SHORT - DT_DELTA, 3, 0}, 176 {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0}, 177 {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0}, 178 {LOW_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 1}, 179 180 {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2}, 181 182 {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC, 1, 0}, 183 184 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3}, 185 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0}, 186 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0}, 187 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER, 6, 0}, 188 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2}, 189 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0}, 190 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0}, 191 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0}, 192 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORTER - 2*DT_DELTA, 6, 0}, 193 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical 194 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0}, 195 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0}, 196 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0}, 197 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER - DT_DELTA, 6, 0}, 198 199 {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2}, 200 {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 20}, // really internal use, so we don't care 201 202 {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC, 1, 3}, 203 204 {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC, 1, 0}, 205 206 {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 3}, 207 {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_LONG, 4, 0}, 208 {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_NARROW, 5, 0}, 209 {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - DT_DELTA, 1, 3}, 210 {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - DT_DELTA, 4, 0}, 211 {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - DT_DELTA, 5, 0}, 212 // b needs to be closer to a than to B, so we make this 3*DT_DELTA 213 {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - 3*DT_DELTA, 1, 3}, 214 {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - 3*DT_DELTA, 4, 0}, 215 {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - 3*DT_DELTA, 5, 0}, 216 217 {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour 218 {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2}, // 24 hour 219 {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour 220 {CAP_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // 12 hour 221 // The C code has had versions of the following 3, keep & update. Should not need these, but... 222 // Without these, certain tests using e.g. staticGetSkeleton fail because j/J in patterns 223 // get skipped instead of mapped to the right hour chars, for example in 224 // DateFormatTest::TestPatternFromSkeleton 225 // IntlTestDateTimePatternGeneratorAPI:: testStaticGetSkeleton 226 // DateIntervalFormatTest::testTicket11985 227 // Need to investigate better handling of jJC replacement e.g. in staticGetSkeleton. 228 {CAP_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 5*DT_DELTA, 1, 2}, // 12/24 hour no AM/PM 229 {LOW_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 6*DT_DELTA, 1, 6}, // 12/24 hour 230 {CAP_C, UDATPG_HOUR_FIELD, DT_NUMERIC + 7*DT_DELTA, 1, 6}, // 12/24 hour with preferred dayPeriods for 12 231 232 {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2}, 233 234 {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2}, 235 {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000}, 236 237 {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC, 1, 1000}, 238 239 {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0}, 240 {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0}, 241 {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3}, 242 {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0}, 243 {CAP_Z, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 3}, 244 {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, 245 {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 5, 0}, 246 {CAP_O, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0}, 247 {CAP_O, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, 248 {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0}, 249 {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 2, 0}, 250 {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-1 - DT_DELTA, 3, 0}, 251 {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-2 - DT_DELTA, 4, 0}, 252 {CAP_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0}, 253 {CAP_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0}, 254 {CAP_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, 255 {LOW_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0}, 256 {LOW_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0}, 257 {LOW_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0}, 258 259 {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[] 260 }; 261 262 static const char* const CLDR_FIELD_APPEND[] = { 263 "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week", 264 "*", "*", "Day", "DayPeriod", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J 265 "Hour", "Minute", "Second", "FractionalSecond", "Timezone" 266 }; 267 268 static const char* const CLDR_FIELD_NAME[UDATPG_FIELD_COUNT] = { 269 "era", "year", "quarter", "month", "week", "weekOfMonth", "weekday", 270 "dayOfYear", "weekdayOfMonth", "day", "dayperiod", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J 271 "hour", "minute", "second", "fractionalSecond", "zone" 272 }; 273 274 static const char* const CLDR_FIELD_WIDTH[] = { // [UDATPG_WIDTH_COUNT] 275 "", "-short", "-narrow" 276 }; 277 278 static constexpr UDateTimePGDisplayWidth UDATPG_WIDTH_APPENDITEM = UDATPG_WIDE; 279 static constexpr int32_t UDATPG_FIELD_KEY_MAX = 24; // max length of CLDR field tag (type + width) 280 281 // For appendItems 282 static const char16_t UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A, 283 0x20, 0x7B, 0x31, 0x7D, 0x2524, 0}; // {0} \u251C{2}: {1}\u2524 284 285 //static const char16_t repeatedPatterns[6]={CAP_G, CAP_E, LOW_Z, LOW_V, CAP_Q, 0}; // "GEzvQ" 286 287 static const char DT_DateTimePatternsTag[]="DateTimePatterns"; 288 static const char DT_DateAtTimePatternsTag[]="DateTimePatterns%atTime"; 289 static const char DT_DateTimeCalendarTag[]="calendar"; 290 static const char DT_DateTimeGregorianTag[]="gregorian"; 291 static const char DT_DateTimeAppendItemsTag[]="appendItems"; 292 static const char DT_DateTimeFieldsTag[]="fields"; 293 static const char DT_DateTimeAvailableFormatsTag[]="availableFormats"; 294 //static const UnicodeString repeatedPattern=UnicodeString(repeatedPatterns); 295 296 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateTimePatternGenerator) 297 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration) 298 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTRedundantEnumeration) 299 300 DateTimePatternGenerator* U_EXPORT2 301 DateTimePatternGenerator::createInstance(UErrorCode& status) { 302 return createInstance(Locale::getDefault(), status); 303 } 304 305 DateTimePatternGenerator* U_EXPORT2 306 DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) { 307 if (U_FAILURE(status)) { 308 return nullptr; 309 } 310 LocalPointer<DateTimePatternGenerator> result( 311 new DateTimePatternGenerator(locale, status), status); 312 return U_SUCCESS(status) ? result.orphan() : nullptr; 313 } 314 315 DateTimePatternGenerator* U_EXPORT2 316 DateTimePatternGenerator::createInstanceNoStdPat(const Locale& locale, UErrorCode& status) { 317 if (U_FAILURE(status)) { 318 return nullptr; 319 } 320 LocalPointer<DateTimePatternGenerator> result( 321 new DateTimePatternGenerator(locale, status, true), status); 322 return U_SUCCESS(status) ? result.orphan() : nullptr; 323 } 324 325 DateTimePatternGenerator* U_EXPORT2 326 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) { 327 if (U_FAILURE(status)) { 328 return nullptr; 329 } 330 LocalPointer<DateTimePatternGenerator> result( 331 new DateTimePatternGenerator(status), status); 332 return U_SUCCESS(status) ? result.orphan() : nullptr; 333 } 334 335 DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) : 336 skipMatcher(nullptr), 337 fAvailableFormatKeyHash(nullptr), 338 fDefaultHourFormatChar(0), 339 internalErrorCode(U_ZERO_ERROR) 340 { 341 fp = new FormatParser(); 342 dtMatcher = new DateTimeMatcher(); 343 distanceInfo = new DistanceInfo(); 344 patternMap = new PatternMap(); 345 if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) { 346 internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR; 347 } 348 } 349 350 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status, UBool skipStdPatterns) : 351 skipMatcher(nullptr), 352 fAvailableFormatKeyHash(nullptr), 353 fDefaultHourFormatChar(0), 354 internalErrorCode(U_ZERO_ERROR) 355 { 356 fp = new FormatParser(); 357 dtMatcher = new DateTimeMatcher(); 358 distanceInfo = new DistanceInfo(); 359 patternMap = new PatternMap(); 360 if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) { 361 internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR; 362 } 363 else { 364 initData(locale, status, skipStdPatterns); 365 } 366 } 367 368 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) : 369 UObject(), 370 skipMatcher(nullptr), 371 fAvailableFormatKeyHash(nullptr), 372 fDefaultHourFormatChar(0), 373 internalErrorCode(U_ZERO_ERROR) 374 { 375 fp = new FormatParser(); 376 dtMatcher = new DateTimeMatcher(); 377 distanceInfo = new DistanceInfo(); 378 patternMap = new PatternMap(); 379 if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) { 380 internalErrorCode = U_MEMORY_ALLOCATION_ERROR; 381 } 382 *this=other; 383 } 384 385 DateTimePatternGenerator& 386 DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) { 387 // reflexive case 388 if (&other == this) { 389 return *this; 390 } 391 internalErrorCode = other.internalErrorCode; 392 pLocale = other.pLocale; 393 fDefaultHourFormatChar = other.fDefaultHourFormatChar; 394 *fp = *(other.fp); 395 dtMatcher->copyFrom(other.dtMatcher->skeleton); 396 *distanceInfo = *(other.distanceInfo); 397 for (int32_t style = UDAT_FULL; style <= UDAT_SHORT; style++) { 398 dateTimeFormat[style] = other.dateTimeFormat[style]; 399 } 400 decimal = other.decimal; 401 for (int32_t style = UDAT_FULL; style <= UDAT_SHORT; style++) { 402 dateTimeFormat[style].getTerminatedBuffer(); // NUL-terminate for the C API. 403 } 404 decimal.getTerminatedBuffer(); 405 delete skipMatcher; 406 if ( other.skipMatcher == nullptr ) { 407 skipMatcher = nullptr; 408 } 409 else { 410 skipMatcher = new DateTimeMatcher(*other.skipMatcher); 411 if (skipMatcher == nullptr) 412 { 413 internalErrorCode = U_MEMORY_ALLOCATION_ERROR; 414 return *this; 415 } 416 } 417 for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) { 418 appendItemFormats[i] = other.appendItemFormats[i]; 419 appendItemFormats[i].getTerminatedBuffer(); // NUL-terminate for the C API. 420 for (int32_t j=0; j< UDATPG_WIDTH_COUNT; ++j ) { 421 fieldDisplayNames[i][j] = other.fieldDisplayNames[i][j]; 422 fieldDisplayNames[i][j].getTerminatedBuffer(); // NUL-terminate for the C API. 423 } 424 } 425 patternMap->copyFrom(*other.patternMap, internalErrorCode); 426 copyHashtable(other.fAvailableFormatKeyHash, internalErrorCode); 427 return *this; 428 } 429 430 431 bool 432 DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) const { 433 if (this == &other) { 434 return true; 435 } 436 if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) && 437 (decimal==other.decimal)) { 438 for (int32_t style = UDAT_FULL; style <= UDAT_SHORT; style++) { 439 if (dateTimeFormat[style] != other.dateTimeFormat[style]) { 440 return false; 441 } 442 } 443 for ( int32_t i=0 ; i<UDATPG_FIELD_COUNT; ++i ) { 444 if (appendItemFormats[i] != other.appendItemFormats[i]) { 445 return false; 446 } 447 for (int32_t j=0; j< UDATPG_WIDTH_COUNT; ++j ) { 448 if (fieldDisplayNames[i][j] != other.fieldDisplayNames[i][j]) { 449 return false; 450 } 451 } 452 } 453 return true; 454 } 455 else { 456 return false; 457 } 458 } 459 460 bool 461 DateTimePatternGenerator::operator!=(const DateTimePatternGenerator& other) const { 462 return !operator==(other); 463 } 464 465 DateTimePatternGenerator::~DateTimePatternGenerator() { 466 delete fAvailableFormatKeyHash; 467 delete fp; 468 delete dtMatcher; 469 delete distanceInfo; 470 delete patternMap; 471 delete skipMatcher; 472 } 473 474 namespace { 475 476 UInitOnce initOnce {}; 477 UHashtable *localeToAllowedHourFormatsMap = nullptr; 478 479 // Value deleter for hashmap. 480 U_CFUNC void U_CALLCONV deleteAllowedHourFormats(void *ptr) { 481 uprv_free(ptr); 482 } 483 484 // Close hashmap at cleanup. 485 U_CFUNC UBool U_CALLCONV allowedHourFormatsCleanup() { 486 uhash_close(localeToAllowedHourFormatsMap); 487 return true; 488 } 489 490 enum AllowedHourFormat{ 491 ALLOWED_HOUR_FORMAT_UNKNOWN = -1, 492 ALLOWED_HOUR_FORMAT_h, 493 ALLOWED_HOUR_FORMAT_H, 494 ALLOWED_HOUR_FORMAT_K, // Added ICU-20383, used by JP 495 ALLOWED_HOUR_FORMAT_k, // Added ICU-20383, not currently used 496 ALLOWED_HOUR_FORMAT_hb, 497 ALLOWED_HOUR_FORMAT_hB, 498 ALLOWED_HOUR_FORMAT_Kb, // Added ICU-20383, not currently used 499 ALLOWED_HOUR_FORMAT_KB, // Added ICU-20383, not currently used 500 // ICU-20383 The following are unlikely and not currently used 501 ALLOWED_HOUR_FORMAT_Hb, 502 ALLOWED_HOUR_FORMAT_HB 503 }; 504 505 } // namespace 506 507 void 508 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status, UBool skipStdPatterns) { 509 //const char *baseLangName = locale.getBaseName(); // unused 510 if (U_FAILURE(status)) { return; } 511 if (locale.isBogus()) { 512 status = U_ILLEGAL_ARGUMENT_ERROR; 513 return; 514 } 515 516 skipMatcher = nullptr; 517 fAvailableFormatKeyHash=nullptr; 518 addCanonicalItems(status); 519 if (!skipStdPatterns) { // skip to prevent circular dependency when called from SimpleDateFormat::construct 520 addICUPatterns(locale, status); 521 } 522 addCLDRData(locale, status); 523 setDateTimeFromCalendar(locale, status); 524 setDecimalSymbols(locale, status); 525 umtx_initOnce(initOnce, loadAllowedHourFormatsData, status); 526 getAllowedHourFormats(locale, status); 527 // If any of the above methods failed then the object is in an invalid state. 528 internalErrorCode = status; 529 } // DateTimePatternGenerator::initData 530 531 namespace { 532 533 struct AllowedHourFormatsSink : public ResourceSink { 534 // Initialize sub-sinks. 535 AllowedHourFormatsSink() {} 536 virtual ~AllowedHourFormatsSink(); 537 538 virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/, 539 UErrorCode &errorCode) override { 540 ResourceTable timeData = value.getTable(errorCode); 541 if (U_FAILURE(errorCode)) { return; } 542 for (int32_t i = 0; timeData.getKeyAndValue(i, key, value); ++i) { 543 const char *regionOrLocale = key; 544 ResourceTable formatList = value.getTable(errorCode); 545 if (U_FAILURE(errorCode)) { return; } 546 // below we construct a list[] that has an entry for the "preferred" value at [0], 547 // followed by 1 or more entries for the "allowed" values, terminated with an 548 // entry for ALLOWED_HOUR_FORMAT_UNKNOWN (not included in length below) 549 LocalMemory<int32_t> list; 550 int32_t length = 0; 551 int32_t preferredFormat = ALLOWED_HOUR_FORMAT_UNKNOWN; 552 for (int32_t j = 0; formatList.getKeyAndValue(j, key, value); ++j) { 553 if (uprv_strcmp(key, "allowed") == 0) { 554 if (value.getType() == URES_STRING) { 555 length = 2; // 1 preferred to add later, 1 allowed to add now 556 if (list.allocateInsteadAndReset(length + 1) == nullptr) { 557 errorCode = U_MEMORY_ALLOCATION_ERROR; 558 return; 559 } 560 list[1] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode)); 561 } 562 else { 563 ResourceArray allowedFormats = value.getArray(errorCode); 564 length = allowedFormats.getSize() + 1; // 1 preferred, getSize allowed 565 if (list.allocateInsteadAndReset(length + 1) == nullptr) { 566 errorCode = U_MEMORY_ALLOCATION_ERROR; 567 return; 568 } 569 for (int32_t k = 1; k < length; ++k) { 570 allowedFormats.getValue(k-1, value); 571 list[k] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode)); 572 } 573 } 574 } else if (uprv_strcmp(key, "preferred") == 0) { 575 preferredFormat = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode)); 576 } 577 } 578 if (length > 1) { 579 list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: list[1]; 580 } else { 581 // fallback handling for missing data 582 length = 2; // 1 preferred, 1 allowed 583 if (list.allocateInsteadAndReset(length + 1) == nullptr) { 584 errorCode = U_MEMORY_ALLOCATION_ERROR; 585 return; 586 } 587 list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: ALLOWED_HOUR_FORMAT_H; 588 list[1] = list[0]; 589 } 590 list[length] = ALLOWED_HOUR_FORMAT_UNKNOWN; 591 // At this point list[] will have at least two non-ALLOWED_HOUR_FORMAT_UNKNOWN entries, 592 // followed by ALLOWED_HOUR_FORMAT_UNKNOWN. 593 uhash_put(localeToAllowedHourFormatsMap, const_cast<char *>(regionOrLocale), list.orphan(), &errorCode); 594 if (U_FAILURE(errorCode)) { return; } 595 } 596 } 597 598 AllowedHourFormat getHourFormatFromUnicodeString(const UnicodeString &s) { 599 if (s.length() == 1) { 600 if (s[0] == LOW_H) { return ALLOWED_HOUR_FORMAT_h; } 601 if (s[0] == CAP_H) { return ALLOWED_HOUR_FORMAT_H; } 602 if (s[0] == CAP_K) { return ALLOWED_HOUR_FORMAT_K; } 603 if (s[0] == LOW_K) { return ALLOWED_HOUR_FORMAT_k; } 604 } else if (s.length() == 2) { 605 if (s[0] == LOW_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_hb; } 606 if (s[0] == LOW_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_hB; } 607 if (s[0] == CAP_K && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Kb; } 608 if (s[0] == CAP_K && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_KB; } 609 if (s[0] == CAP_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Hb; } 610 if (s[0] == CAP_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_HB; } 611 } 612 613 return ALLOWED_HOUR_FORMAT_UNKNOWN; 614 } 615 }; 616 617 } // namespace 618 619 AllowedHourFormatsSink::~AllowedHourFormatsSink() {} 620 621 U_CFUNC void U_CALLCONV DateTimePatternGenerator::loadAllowedHourFormatsData(UErrorCode &status) { 622 if (U_FAILURE(status)) { return; } 623 localeToAllowedHourFormatsMap = uhash_open( 624 uhash_hashChars, uhash_compareChars, nullptr, &status); 625 if (U_FAILURE(status)) { return; } 626 627 uhash_setValueDeleter(localeToAllowedHourFormatsMap, deleteAllowedHourFormats); 628 ucln_i18n_registerCleanup(UCLN_I18N_ALLOWED_HOUR_FORMATS, allowedHourFormatsCleanup); 629 630 LocalUResourceBundlePointer rb(ures_openDirect(nullptr, "supplementalData", &status)); 631 if (U_FAILURE(status)) { return; } 632 633 AllowedHourFormatsSink sink; 634 // TODO: Currently in the enumeration each table allocates a new array. 635 // Try to reduce the number of memory allocations. Consider storing a 636 // UVector32 with the concatenation of all of the sub-arrays, put the start index 637 // into the hashmap, store 6 single-value sub-arrays right at the beginning of the 638 // vector (at index enum*2) for easy data sharing, copy sub-arrays into runtime 639 // object. Remember to clean up the vector, too. 640 ures_getAllItemsWithFallback(rb.getAlias(), "timeData", sink, status); 641 } 642 643 static int32_t* getAllowedHourFormatsLangCountry(const char* language, const char* country, UErrorCode& status) { 644 CharString langCountry; 645 langCountry.append(language, status); 646 langCountry.append('_', status); 647 langCountry.append(country, status); 648 649 int32_t* allowedFormats; 650 allowedFormats = static_cast<int32_t*>(uhash_get(localeToAllowedHourFormatsMap, langCountry.data())); 651 if (allowedFormats == nullptr) { 652 allowedFormats = static_cast<int32_t*>(uhash_get(localeToAllowedHourFormatsMap, const_cast<char*>(country))); 653 } 654 655 return allowedFormats; 656 } 657 658 void DateTimePatternGenerator::getAllowedHourFormats(const Locale &locale, UErrorCode &status) { 659 if (U_FAILURE(status)) { return; } 660 661 const char *language = locale.getLanguage(); 662 CharString baseCountry = ulocimp_getRegionForSupplementalData(locale.getName(), false, status); 663 const char* country = baseCountry.data(); 664 665 Locale maxLocale; // must be here for correct lifetime 666 if (*language == '\0' || *country == '\0') { 667 maxLocale = locale; 668 UErrorCode localStatus = U_ZERO_ERROR; 669 maxLocale.addLikelySubtags(localStatus); 670 if (U_SUCCESS(localStatus)) { 671 language = maxLocale.getLanguage(); 672 country = maxLocale.getCountry(); 673 } 674 } 675 if (*language == '\0') { 676 // Unexpected, but fail gracefully 677 language = "und"; 678 } 679 if (*country == '\0') { 680 country = "001"; 681 } 682 683 int32_t* allowedFormats = getAllowedHourFormatsLangCountry(language, country, status); 684 685 // We need to check if there is an hour cycle on locale 686 char buffer[8]; 687 int32_t count = locale.getKeywordValue("hours", buffer, sizeof(buffer), status); 688 689 fDefaultHourFormatChar = 0; 690 if (U_SUCCESS(status) && count > 0) { 691 if(uprv_strcmp(buffer, "h24") == 0) { 692 fDefaultHourFormatChar = LOW_K; 693 } else if(uprv_strcmp(buffer, "h23") == 0) { 694 fDefaultHourFormatChar = CAP_H; 695 } else if(uprv_strcmp(buffer, "h12") == 0) { 696 fDefaultHourFormatChar = LOW_H; 697 } else if(uprv_strcmp(buffer, "h11") == 0) { 698 fDefaultHourFormatChar = CAP_K; 699 } 700 } 701 702 // Check if the region has an alias 703 if (allowedFormats == nullptr) { 704 UErrorCode localStatus = U_ZERO_ERROR; 705 const Region* region = Region::getInstance(country, localStatus); 706 if (U_SUCCESS(localStatus)) { 707 country = region->getRegionCode(); // the real region code 708 allowedFormats = getAllowedHourFormatsLangCountry(language, country, status); 709 } 710 } 711 712 if (allowedFormats != nullptr) { // Lookup is successful 713 // Here allowedFormats points to a list consisting of key for preferredFormat, 714 // followed by one or more keys for allowedFormats, then followed by ALLOWED_HOUR_FORMAT_UNKNOWN. 715 if (!fDefaultHourFormatChar) { 716 switch (allowedFormats[0]) { 717 case ALLOWED_HOUR_FORMAT_h: fDefaultHourFormatChar = LOW_H; break; 718 case ALLOWED_HOUR_FORMAT_H: fDefaultHourFormatChar = CAP_H; break; 719 case ALLOWED_HOUR_FORMAT_K: fDefaultHourFormatChar = CAP_K; break; 720 case ALLOWED_HOUR_FORMAT_k: fDefaultHourFormatChar = LOW_K; break; 721 default: fDefaultHourFormatChar = CAP_H; break; 722 } 723 } 724 725 for (int32_t i = 0; i < UPRV_LENGTHOF(fAllowedHourFormats); ++i) { 726 fAllowedHourFormats[i] = allowedFormats[i + 1]; 727 if (fAllowedHourFormats[i] == ALLOWED_HOUR_FORMAT_UNKNOWN) { 728 break; 729 } 730 } 731 } else { // Lookup failed, twice 732 if (!fDefaultHourFormatChar) { 733 fDefaultHourFormatChar = CAP_H; 734 } 735 fAllowedHourFormats[0] = ALLOWED_HOUR_FORMAT_H; 736 fAllowedHourFormats[1] = ALLOWED_HOUR_FORMAT_UNKNOWN; 737 } 738 } 739 740 UDateFormatHourCycle 741 DateTimePatternGenerator::getDefaultHourCycle(UErrorCode& status) const { 742 if (U_FAILURE(status)) { 743 return UDAT_HOUR_CYCLE_23; 744 } 745 if (fDefaultHourFormatChar == 0) { 746 // We need to return something, but the caller should ignore it 747 // anyways since the returned status is a failure. 748 status = U_UNSUPPORTED_ERROR; 749 return UDAT_HOUR_CYCLE_23; 750 } 751 switch (fDefaultHourFormatChar) { 752 case CAP_K: 753 return UDAT_HOUR_CYCLE_11; 754 case LOW_H: 755 return UDAT_HOUR_CYCLE_12; 756 case CAP_H: 757 return UDAT_HOUR_CYCLE_23; 758 case LOW_K: 759 return UDAT_HOUR_CYCLE_24; 760 default: 761 UPRV_UNREACHABLE_EXIT; 762 } 763 } 764 765 UnicodeString 766 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode& 767 /*status*/) { 768 FormatParser fp2; 769 DateTimeMatcher matcher; 770 PtnSkeleton localSkeleton; 771 matcher.set(pattern, &fp2, localSkeleton); 772 return localSkeleton.getSkeleton(); 773 } 774 775 UnicodeString 776 DateTimePatternGenerator::staticGetSkeleton( 777 const UnicodeString& pattern, UErrorCode& /*status*/) { 778 FormatParser fp; 779 DateTimeMatcher matcher; 780 PtnSkeleton localSkeleton; 781 matcher.set(pattern, &fp, localSkeleton); 782 return localSkeleton.getSkeleton(); 783 } 784 785 UnicodeString 786 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) { 787 FormatParser fp2; 788 DateTimeMatcher matcher; 789 PtnSkeleton localSkeleton; 790 matcher.set(pattern, &fp2, localSkeleton); 791 return localSkeleton.getBaseSkeleton(); 792 } 793 794 UnicodeString 795 DateTimePatternGenerator::staticGetBaseSkeleton( 796 const UnicodeString& pattern, UErrorCode& /*status*/) { 797 FormatParser fp; 798 DateTimeMatcher matcher; 799 PtnSkeleton localSkeleton; 800 matcher.set(pattern, &fp, localSkeleton); 801 return localSkeleton.getBaseSkeleton(); 802 } 803 804 void 805 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) { 806 if (U_FAILURE(status)) { 807 return; 808 } 809 810 LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getBaseName(), &status)); 811 CharString calendarTypeToUse; // to be filled in with the type to use, if all goes well 812 getCalendarTypeToUse(locale, calendarTypeToUse, status); 813 814 // HACK to get around the fact that the old SimpleDateFormat code (actually, Calendar::getCalendarTypeForLocale() ) 815 // returns "gregorian" for ja_JP_TRADITIONAL instead of "japanese" 816 if (uprv_strcmp(locale.getBaseName(), "ja_JP_TRADITIONAL") == 0) { 817 calendarTypeToUse.clear().append("gregorian", status); 818 } 819 820 if (U_FAILURE(status)) { 821 return; 822 } 823 824 // TODO: See ICU-22867 825 CharString patternResourcePath; 826 patternResourcePath.append(DT_DateTimeCalendarTag, status) 827 .append('/', status) 828 .append(calendarTypeToUse, status) 829 .append('/', status) 830 .append(DT_DateTimePatternsTag, status); 831 832 LocalUResourceBundlePointer dateTimePatterns(ures_getByKeyWithFallback(rb.getAlias(), patternResourcePath.data(), 833 nullptr, &status)); 834 if (ures_getType(dateTimePatterns.getAlias()) != URES_ARRAY || ures_getSize(dateTimePatterns.getAlias()) < 8) { 835 status = U_INVALID_FORMAT_ERROR; 836 return; 837 } 838 839 for (int32_t i = 0; U_SUCCESS(status) && i < DateFormat::kDateTime; i++) { 840 LocalUResourceBundlePointer patternRes(ures_getByIndex(dateTimePatterns.getAlias(), i, nullptr, &status)); 841 UnicodeString pattern; 842 switch (ures_getType(patternRes.getAlias())) { 843 case URES_STRING: 844 pattern = ures_getUnicodeString(patternRes.getAlias(), &status); 845 break; 846 case URES_ARRAY: 847 pattern = ures_getUnicodeStringByIndex(patternRes.getAlias(), 0, &status); 848 break; 849 default: 850 status = U_INVALID_FORMAT_ERROR; 851 return; 852 } 853 854 if (U_SUCCESS(status)) { 855 UnicodeString conflictingPattern; 856 addPatternWithOptionalSkeleton(pattern, nullptr, false, conflictingPattern, status); 857 } 858 } 859 } 860 861 void 862 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status) { 863 UnicodeString conflictingString; 864 865 fp->set(hackPattern); 866 UnicodeString mmss; 867 UBool gotMm=false; 868 for (int32_t i=0; i<fp->itemNumber; ++i) { 869 UnicodeString field = fp->items[i]; 870 if ( fp->isQuoteLiteral(field) ) { 871 if ( gotMm ) { 872 UnicodeString quoteLiteral; 873 fp->getQuoteLiteral(quoteLiteral, &i); 874 mmss += quoteLiteral; 875 } 876 } 877 else { 878 if (fp->isPatternSeparator(field) && gotMm) { 879 mmss+=field; 880 } 881 else { 882 char16_t ch=field.charAt(0); 883 if (ch==LOW_M) { 884 gotMm=true; 885 mmss+=field; 886 } 887 else { 888 if (ch==LOW_S) { 889 if (!gotMm) { 890 break; 891 } 892 mmss+= field; 893 addPattern(mmss, false, conflictingString, status); 894 break; 895 } 896 else { 897 if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) { 898 break; 899 } 900 } 901 } 902 } 903 } 904 } 905 } 906 907 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY) 908 909 void 910 DateTimePatternGenerator::getCalendarTypeToUse(const Locale& locale, CharString& destination, UErrorCode& err) { 911 destination.clear().append(DT_DateTimeGregorianTag, -1, err); // initial default 912 if ( U_SUCCESS(err) ) { 913 UErrorCode localStatus = U_ZERO_ERROR; 914 char localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY]; 915 // obtain a locale that always has the calendar key value that should be used 916 ures_getFunctionalEquivalent( 917 localeWithCalendarKey, 918 ULOC_LOCALE_IDENTIFIER_CAPACITY, 919 nullptr, 920 "calendar", 921 "calendar", 922 locale.getName(), 923 nullptr, 924 false, 925 &localStatus); 926 localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination 927 // now get the calendar key value from that locale 928 // (the call to ures_getFunctionalEquivalent() above might fail, and if it does, localeWithCalendarKey 929 // won't contain a `calendar` keyword. If this happens, the line below will stomp on `destination`, 930 // so we have to check the return code before overwriting `destination`.) 931 if (U_SUCCESS(localStatus)) { 932 destination = ulocimp_getKeywordValue(localeWithCalendarKey, "calendar", localStatus); 933 } 934 // If the input locale was invalid, don't fail with missing resource error, instead 935 // continue with default of Gregorian. 936 if (U_FAILURE(localStatus) && localStatus != U_MISSING_RESOURCE_ERROR) { 937 err = localStatus; 938 } 939 } 940 } 941 942 void 943 DateTimePatternGenerator::consumeShortTimePattern(const UnicodeString& shortTimePattern, 944 UErrorCode& status) { 945 if (U_FAILURE(status)) { return; } 946 // ICU-20383 No longer set fDefaultHourFormatChar to the hour format character from 947 // this pattern; instead it is set from localeToAllowedHourFormatsMap which now 948 // includes entries for both preferred and allowed formats. 949 950 // HACK for hh:ss 951 hackTimes(shortTimePattern, status); 952 } 953 954 struct DateTimePatternGenerator::AppendItemFormatsSink : public ResourceSink { 955 956 // Destination for data, modified via setters. 957 DateTimePatternGenerator& dtpg; 958 959 AppendItemFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {} 960 virtual ~AppendItemFormatsSink(); 961 962 virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/, 963 UErrorCode &errorCode) override { 964 UDateTimePatternField field = dtpg.getAppendFormatNumber(key); 965 if (field == UDATPG_FIELD_COUNT) { return; } 966 const UnicodeString& valueStr = value.getUnicodeString(errorCode); 967 if (dtpg.getAppendItemFormat(field).isEmpty() && !valueStr.isEmpty()) { 968 dtpg.setAppendItemFormat(field, valueStr); 969 } 970 } 971 972 void fillInMissing() { 973 UnicodeString defaultItemFormat(true, UDATPG_ItemFormat, UPRV_LENGTHOF(UDATPG_ItemFormat)-1); // Read-only alias. 974 for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) { 975 UDateTimePatternField field = static_cast<UDateTimePatternField>(i); 976 if (dtpg.getAppendItemFormat(field).isEmpty()) { 977 dtpg.setAppendItemFormat(field, defaultItemFormat); 978 } 979 } 980 } 981 }; 982 983 struct DateTimePatternGenerator::AppendItemNamesSink : public ResourceSink { 984 985 // Destination for data, modified via setters. 986 DateTimePatternGenerator& dtpg; 987 988 AppendItemNamesSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {} 989 virtual ~AppendItemNamesSink(); 990 991 virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/, 992 UErrorCode &errorCode) override { 993 UDateTimePGDisplayWidth width; 994 UDateTimePatternField field = dtpg.getFieldAndWidthIndices(key, &width); 995 if (field == UDATPG_FIELD_COUNT) { return; } 996 ResourceTable detailsTable = value.getTable(errorCode); 997 if (U_FAILURE(errorCode)) { return; } 998 if (!detailsTable.findValue("dn", value)) { return; } 999 const UnicodeString& valueStr = value.getUnicodeString(errorCode); 1000 if (U_SUCCESS(errorCode) && dtpg.getFieldDisplayName(field,width).isEmpty() && !valueStr.isEmpty()) { 1001 dtpg.setFieldDisplayName(field,width,valueStr); 1002 } 1003 } 1004 1005 void fillInMissing() { 1006 for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) { 1007 UnicodeString& valueStr = dtpg.getMutableFieldDisplayName(static_cast<UDateTimePatternField>(i), UDATPG_WIDE); 1008 if (valueStr.isEmpty()) { 1009 valueStr = CAP_F; 1010 U_ASSERT(i < 20); 1011 if (i < 10) { 1012 // F0, F1, ..., F9 1013 valueStr += static_cast<char16_t>(i + 0x30); 1014 } else { 1015 // F10, F11, ... 1016 valueStr += static_cast<char16_t>(0x31); 1017 valueStr += static_cast<char16_t>(i - 10 + 0x30); 1018 } 1019 // NUL-terminate for the C API. 1020 valueStr.getTerminatedBuffer(); 1021 } 1022 for (int32_t j = 1; j < UDATPG_WIDTH_COUNT; j++) { 1023 UnicodeString& valueStr2 = dtpg.getMutableFieldDisplayName(static_cast<UDateTimePatternField>(i), static_cast<UDateTimePGDisplayWidth>(j)); 1024 if (valueStr2.isEmpty()) { 1025 valueStr2 = dtpg.getFieldDisplayName(static_cast<UDateTimePatternField>(i), static_cast<UDateTimePGDisplayWidth>(j - 1)); 1026 } 1027 } 1028 } 1029 } 1030 }; 1031 1032 struct DateTimePatternGenerator::AvailableFormatsSink : public ResourceSink { 1033 1034 // Destination for data, modified via setters. 1035 DateTimePatternGenerator& dtpg; 1036 1037 // Temporary variable, required for calling addPatternWithSkeleton. 1038 UnicodeString conflictingPattern; 1039 1040 AvailableFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {} 1041 virtual ~AvailableFormatsSink(); 1042 1043 virtual void put(const char *key, ResourceValue &value, UBool /*isRoot*/, 1044 UErrorCode &errorCode) override { 1045 const UnicodeString formatKey(key, -1, US_INV); 1046 if (!dtpg.isAvailableFormatSet(formatKey) ) { 1047 dtpg.setAvailableFormat(formatKey, errorCode); 1048 // Add pattern with its associated skeleton. Override any duplicate 1049 // derived from std patterns, but not a previous availableFormats entry: 1050 const UnicodeString& formatValue = value.getUnicodeString(errorCode); 1051 conflictingPattern.remove(); 1052 dtpg.addPatternWithSkeleton(formatValue, formatKey, true, conflictingPattern, errorCode); 1053 } 1054 } 1055 }; 1056 1057 // Virtual destructors must be defined out of line. 1058 DateTimePatternGenerator::AppendItemFormatsSink::~AppendItemFormatsSink() {} 1059 DateTimePatternGenerator::AppendItemNamesSink::~AppendItemNamesSink() {} 1060 DateTimePatternGenerator::AvailableFormatsSink::~AvailableFormatsSink() {} 1061 1062 void 1063 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& errorCode) { 1064 if (U_FAILURE(errorCode)) { return; } 1065 UnicodeString rbPattern, value, field; 1066 CharString path; 1067 1068 LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &errorCode)); 1069 if (U_FAILURE(errorCode)) { return; } 1070 1071 CharString calendarTypeToUse; // to be filled in with the type to use, if all goes well 1072 getCalendarTypeToUse(locale, calendarTypeToUse, errorCode); 1073 if (U_FAILURE(errorCode)) { return; } 1074 1075 // Local err to ignore resource not found exceptions 1076 UErrorCode err = U_ZERO_ERROR; 1077 1078 // Load append item formats. 1079 AppendItemFormatsSink appendItemFormatsSink(*this); 1080 path.clear() 1081 .append(DT_DateTimeCalendarTag, errorCode) 1082 .append('/', errorCode) 1083 .append(calendarTypeToUse, errorCode) 1084 .append('/', errorCode) 1085 .append(DT_DateTimeAppendItemsTag, errorCode); // i.e., calendar/xxx/appendItems 1086 if (U_FAILURE(errorCode)) { return; } 1087 ures_getAllChildrenWithFallback(rb.getAlias(), path.data(), appendItemFormatsSink, err); 1088 appendItemFormatsSink.fillInMissing(); 1089 1090 // Load CLDR item names. 1091 err = U_ZERO_ERROR; 1092 AppendItemNamesSink appendItemNamesSink(*this); 1093 ures_getAllChildrenWithFallback(rb.getAlias(), DT_DateTimeFieldsTag, appendItemNamesSink, err); 1094 appendItemNamesSink.fillInMissing(); 1095 1096 // Load the available formats from CLDR. 1097 err = U_ZERO_ERROR; 1098 initHashtable(errorCode); 1099 if (U_FAILURE(errorCode)) { return; } 1100 AvailableFormatsSink availableFormatsSink(*this); 1101 path.clear() 1102 .append(DT_DateTimeCalendarTag, errorCode) 1103 .append('/', errorCode) 1104 .append(calendarTypeToUse, errorCode) 1105 .append('/', errorCode) 1106 .append(DT_DateTimeAvailableFormatsTag, errorCode); // i.e., calendar/xxx/availableFormats 1107 if (U_FAILURE(errorCode)) { return; } 1108 ures_getAllChildrenWithFallback(rb.getAlias(), path.data(), availableFormatsSink, err); 1109 } 1110 1111 void 1112 DateTimePatternGenerator::initHashtable(UErrorCode& err) { 1113 if (U_FAILURE(err)) { return; } 1114 if (fAvailableFormatKeyHash!=nullptr) { 1115 return; 1116 } 1117 LocalPointer<Hashtable> hash(new Hashtable(false, err), err); 1118 if (U_SUCCESS(err)) { 1119 fAvailableFormatKeyHash = hash.orphan(); 1120 } 1121 } 1122 1123 void 1124 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) { 1125 appendItemFormats[field] = value; 1126 // NUL-terminate for the C API. 1127 appendItemFormats[field].getTerminatedBuffer(); 1128 } 1129 1130 const UnicodeString& 1131 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const { 1132 return appendItemFormats[field]; 1133 } 1134 1135 void 1136 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) { 1137 setFieldDisplayName(field, UDATPG_WIDTH_APPENDITEM, value); 1138 } 1139 1140 const UnicodeString& 1141 DateTimePatternGenerator::getAppendItemName(UDateTimePatternField field) const { 1142 return fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM]; 1143 } 1144 1145 void 1146 DateTimePatternGenerator::setFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width, const UnicodeString& value) { 1147 fieldDisplayNames[field][width] = value; 1148 // NUL-terminate for the C API. 1149 fieldDisplayNames[field][width].getTerminatedBuffer(); 1150 } 1151 1152 UnicodeString 1153 DateTimePatternGenerator::getFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) const { 1154 return fieldDisplayNames[field][width]; 1155 } 1156 1157 UnicodeString& 1158 DateTimePatternGenerator::getMutableFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) { 1159 return fieldDisplayNames[field][width]; 1160 } 1161 1162 void 1163 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) { 1164 value = SINGLE_QUOTE; 1165 value += fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM]; 1166 value += SINGLE_QUOTE; 1167 } 1168 1169 UnicodeString 1170 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) { 1171 return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status); 1172 } 1173 1174 UnicodeString 1175 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) { 1176 if (U_FAILURE(status)) { 1177 return {}; 1178 } 1179 if (U_FAILURE(internalErrorCode)) { 1180 status = internalErrorCode; 1181 return {}; 1182 } 1183 const UnicodeString *bestPattern = nullptr; 1184 UnicodeString dtFormat; 1185 UnicodeString resultPattern; 1186 int32_t flags = kDTPGNoFlags; 1187 1188 int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1; 1189 int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask; 1190 1191 // Replace hour metacharacters 'j', 'C' and 'J', set flags as necessary 1192 UnicodeString patternFormMapped = mapSkeletonMetacharacters(patternForm, &flags, status); 1193 if (U_FAILURE(status)) { 1194 return {}; 1195 } 1196 1197 resultPattern.remove(); 1198 dtMatcher->set(patternFormMapped, fp); 1199 const PtnSkeleton* specifiedSkeleton = nullptr; 1200 bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, status, &specifiedSkeleton); 1201 if (U_FAILURE(status)) { 1202 return {}; 1203 } 1204 1205 if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) { 1206 resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, flags, options); 1207 1208 return resultPattern; 1209 } 1210 int32_t neededFields = dtMatcher->getFieldMask(); 1211 UnicodeString datePattern=getBestAppending(neededFields & dateMask, flags, status, options); 1212 UnicodeString timePattern=getBestAppending(neededFields & timeMask, flags, status, options); 1213 if (U_FAILURE(status)) { 1214 return {}; 1215 } 1216 if (datePattern.length()==0) { 1217 if (timePattern.length()==0) { 1218 resultPattern.remove(); 1219 } 1220 else { 1221 return timePattern; 1222 } 1223 } 1224 if (timePattern.length()==0) { 1225 return datePattern; 1226 } 1227 resultPattern.remove(); 1228 status = U_ZERO_ERROR; 1229 // determine which dateTimeFormat to use 1230 PtnSkeleton* reqSkeleton = dtMatcher->getSkeletonPtr(); 1231 UDateFormatStyle style = UDAT_SHORT; 1232 int32_t monthFieldLen = reqSkeleton->baseOriginal.getFieldLength(UDATPG_MONTH_FIELD); 1233 if (monthFieldLen == 4) { 1234 if (reqSkeleton->baseOriginal.getFieldLength(UDATPG_WEEKDAY_FIELD) > 0) { 1235 style = UDAT_FULL; 1236 } else { 1237 style = UDAT_LONG; 1238 } 1239 } else if (monthFieldLen == 3) { 1240 style = UDAT_MEDIUM; 1241 } 1242 // and now use it to compose date and time 1243 dtFormat=getDateTimeFormat(style, status); 1244 SimpleFormatter(dtFormat, 2, 2, status).format(timePattern, datePattern, resultPattern, status); 1245 return resultPattern; 1246 } 1247 1248 /* 1249 * Map a skeleton that may have metacharacters jJC to one without, by replacing 1250 * the metacharacters with locale-appropriate fields of h/H/k/K and of a/b/B 1251 * (depends on fDefaultHourFormatChar and fAllowedHourFormats being set, which in 1252 * turn depends on initData having been run). This method also updates the flags 1253 * as necessary. Returns the updated skeleton. 1254 */ 1255 UnicodeString 1256 DateTimePatternGenerator::mapSkeletonMetacharacters(const UnicodeString& patternForm, int32_t* flags, UErrorCode& status) { 1257 UnicodeString patternFormMapped; 1258 patternFormMapped.remove(); 1259 UBool inQuoted = false; 1260 int32_t patPos, patLen = patternForm.length(); 1261 for (patPos = 0; patPos < patLen; patPos++) { 1262 char16_t patChr = patternForm.charAt(patPos); 1263 if (patChr == SINGLE_QUOTE) { 1264 inQuoted = !inQuoted; 1265 } else if (!inQuoted) { 1266 // Handle special mappings for 'j' and 'C' in which fields lengths 1267 // 1,3,5 => hour field length 1 1268 // 2,4,6 => hour field length 2 1269 // 1,2 => abbreviated dayPeriod (field length 1..3) 1270 // 3,4 => long dayPeriod (field length 4) 1271 // 5,6 => narrow dayPeriod (field length 5) 1272 if (patChr == LOW_J || patChr == CAP_C) { 1273 int32_t extraLen = 0; // 1 less than total field length 1274 while (patPos+1 < patLen && patternForm.charAt(patPos+1)==patChr) { 1275 extraLen++; 1276 patPos++; 1277 } 1278 int32_t hourLen = 1 + (extraLen & 1); 1279 int32_t dayPeriodLen = (extraLen < 2)? 1: 3 + (extraLen >> 1); 1280 char16_t hourChar = LOW_H; 1281 char16_t dayPeriodChar = LOW_A; 1282 if (patChr == LOW_J) { 1283 hourChar = fDefaultHourFormatChar; 1284 } else { 1285 AllowedHourFormat bestAllowed; 1286 if (fAllowedHourFormats[0] != ALLOWED_HOUR_FORMAT_UNKNOWN) { 1287 bestAllowed = static_cast<AllowedHourFormat>(fAllowedHourFormats[0]); 1288 } else { 1289 status = U_INVALID_FORMAT_ERROR; 1290 return {}; 1291 } 1292 if (bestAllowed == ALLOWED_HOUR_FORMAT_H || bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_Hb) { 1293 hourChar = CAP_H; 1294 } else if (bestAllowed == ALLOWED_HOUR_FORMAT_K || bestAllowed == ALLOWED_HOUR_FORMAT_KB || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) { 1295 hourChar = CAP_K; 1296 } else if (bestAllowed == ALLOWED_HOUR_FORMAT_k) { 1297 hourChar = LOW_K; 1298 } 1299 // in #13183 just add b/B to skeleton, no longer need to set special flags 1300 if (bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_hB || bestAllowed == ALLOWED_HOUR_FORMAT_KB) { 1301 dayPeriodChar = CAP_B; 1302 } else if (bestAllowed == ALLOWED_HOUR_FORMAT_Hb || bestAllowed == ALLOWED_HOUR_FORMAT_hb || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) { 1303 dayPeriodChar = LOW_B; 1304 } 1305 } 1306 if (hourChar==CAP_H || hourChar==LOW_K) { 1307 dayPeriodLen = 0; 1308 } 1309 while (dayPeriodLen-- > 0) { 1310 patternFormMapped.append(dayPeriodChar); 1311 } 1312 while (hourLen-- > 0) { 1313 patternFormMapped.append(hourChar); 1314 } 1315 } else if (patChr == CAP_J) { 1316 // Get pattern for skeleton with H, then replace H or k 1317 // with fDefaultHourFormatChar (if different) 1318 patternFormMapped.append(CAP_H); 1319 *flags |= kDTPGSkeletonUsesCapJ; 1320 } else { 1321 patternFormMapped.append(patChr); 1322 } 1323 } 1324 } 1325 return patternFormMapped; 1326 } 1327 1328 UnicodeString 1329 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern, 1330 const UnicodeString& skeleton, 1331 UErrorCode& status) { 1332 return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status); 1333 } 1334 1335 UnicodeString 1336 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern, 1337 const UnicodeString& skeleton, 1338 UDateTimePatternMatchOptions options, 1339 UErrorCode& status) { 1340 if (U_FAILURE(status)) { 1341 return {}; 1342 } 1343 if (U_FAILURE(internalErrorCode)) { 1344 status = internalErrorCode; 1345 return {}; 1346 } 1347 dtMatcher->set(skeleton, fp); 1348 UnicodeString result = adjustFieldTypes(pattern, nullptr, kDTPGNoFlags, options); 1349 return result; 1350 } 1351 1352 void 1353 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) { 1354 this->decimal = newDecimal; 1355 // NUL-terminate for the C API. 1356 this->decimal.getTerminatedBuffer(); 1357 } 1358 1359 const UnicodeString& 1360 DateTimePatternGenerator::getDecimal() const { 1361 return decimal; 1362 } 1363 1364 void 1365 DateTimePatternGenerator::addCanonicalItems(UErrorCode& status) { 1366 if (U_FAILURE(status)) { return; } 1367 UnicodeString conflictingPattern; 1368 1369 for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) { 1370 if (Canonical_Items[i] > 0) { 1371 addPattern(UnicodeString(Canonical_Items[i]), false, conflictingPattern, status); 1372 } 1373 if (U_FAILURE(status)) { return; } 1374 } 1375 } 1376 1377 void 1378 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) { 1379 UErrorCode status = U_ZERO_ERROR; 1380 for (int32_t style = UDAT_FULL; style <= UDAT_SHORT; style++) { 1381 setDateTimeFormat(static_cast<UDateFormatStyle>(style), dtFormat, status); 1382 } 1383 } 1384 1385 const UnicodeString& 1386 DateTimePatternGenerator::getDateTimeFormat() const { 1387 UErrorCode status = U_ZERO_ERROR; 1388 return getDateTimeFormat(UDAT_MEDIUM, status); 1389 } 1390 1391 void 1392 DateTimePatternGenerator::setDateTimeFormat(UDateFormatStyle style, const UnicodeString& dtFormat, UErrorCode& status) { 1393 if (U_FAILURE(status)) { 1394 return; 1395 } 1396 if (style < UDAT_FULL || style > UDAT_SHORT) { 1397 status = U_ILLEGAL_ARGUMENT_ERROR; 1398 return; 1399 } 1400 dateTimeFormat[style] = dtFormat; 1401 // Note for the following: getTerminatedBuffer() can re-allocate the UnicodeString 1402 // buffer so we do this here before clients request a const ref to the UnicodeString 1403 // or its buffer. 1404 dateTimeFormat[style].getTerminatedBuffer(); // NUL-terminate for the C API. 1405 } 1406 1407 const UnicodeString& 1408 DateTimePatternGenerator::getDateTimeFormat(UDateFormatStyle style, UErrorCode& status) const { 1409 static const UnicodeString emptyString = UNICODE_STRING_SIMPLE(""); 1410 if (U_FAILURE(status)) { 1411 return emptyString; 1412 } 1413 if (style < UDAT_FULL || style > UDAT_SHORT) { 1414 status = U_ILLEGAL_ARGUMENT_ERROR; 1415 return emptyString; 1416 } 1417 return dateTimeFormat[style]; 1418 } 1419 1420 static const int32_t cTypeBufMax = 32; 1421 1422 void 1423 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) { 1424 if (U_FAILURE(status)) { return; } 1425 1426 const char16_t *resStr; 1427 int32_t resStrLen = 0; 1428 1429 LocalUResourceBundlePointer calData(ures_open(nullptr, locale.getBaseName(), &status)); 1430 if (U_FAILURE(status)) { return; } 1431 ures_getByKey(calData.getAlias(), DT_DateTimeCalendarTag, calData.getAlias(), &status); 1432 if (U_FAILURE(status)) { return; } 1433 1434 char cType[cTypeBufMax + 1]; 1435 Calendar::getCalendarTypeFromLocale(locale, cType, cTypeBufMax, status); 1436 cType[cTypeBufMax] = 0; 1437 if (U_FAILURE(status) || cType[0] == 0) { 1438 status = U_ZERO_ERROR; 1439 uprv_strcpy(cType, DT_DateTimeGregorianTag); 1440 } 1441 UBool cTypeIsGregorian = (uprv_strcmp(cType, DT_DateTimeGregorianTag) == 0); 1442 1443 // Currently, for compatibility with pre-CLDR-42 data, we default to the "atTime" 1444 // combining patterns. Depending on guidance in CLDR 42 spec and on DisplayOptions, 1445 // we may change this. 1446 LocalUResourceBundlePointer specificCalBundle; 1447 LocalUResourceBundlePointer dateTimePatterns; 1448 int32_t dateTimeOffset = 0; // initially for DateTimePatterns%atTime 1449 if (!cTypeIsGregorian) { 1450 specificCalBundle.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), cType, 1451 nullptr, &status)); 1452 dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(specificCalBundle.getAlias(), DT_DateAtTimePatternsTag, // the %atTime variant, 4 entries 1453 nullptr, &status)); 1454 } 1455 if (dateTimePatterns.isNull() || status == U_MISSING_RESOURCE_ERROR) { 1456 status = U_ZERO_ERROR; 1457 specificCalBundle.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), DT_DateTimeGregorianTag, 1458 nullptr, &status)); 1459 dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(specificCalBundle.getAlias(), DT_DateAtTimePatternsTag, // the %atTime variant, 4 entries 1460 nullptr, &status)); 1461 } 1462 if (U_SUCCESS(status) && (ures_getSize(dateTimePatterns.getAlias()) < 4)) { 1463 status = U_INVALID_FORMAT_ERROR; 1464 } 1465 if (status == U_MISSING_RESOURCE_ERROR) { 1466 // Try again with standard variant 1467 status = U_ZERO_ERROR; 1468 dateTimePatterns.orphan(); 1469 dateTimeOffset = static_cast<int32_t>(DateFormat::kDateTimeOffset); 1470 if (!cTypeIsGregorian) { 1471 specificCalBundle.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), cType, 1472 nullptr, &status)); 1473 dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(specificCalBundle.getAlias(), DT_DateTimePatternsTag, // the standard variant, 13 entries 1474 nullptr, &status)); 1475 } 1476 if (dateTimePatterns.isNull() || status == U_MISSING_RESOURCE_ERROR) { 1477 status = U_ZERO_ERROR; 1478 specificCalBundle.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), DT_DateTimeGregorianTag, 1479 nullptr, &status)); 1480 dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(specificCalBundle.getAlias(), DT_DateTimePatternsTag, // the standard variant, 13 entries 1481 nullptr, &status)); 1482 } 1483 if (U_SUCCESS(status) && (ures_getSize(dateTimePatterns.getAlias()) <= DateFormat::kDateTimeOffset + DateFormat::kShort)) { 1484 status = U_INVALID_FORMAT_ERROR; 1485 } 1486 } 1487 if (U_FAILURE(status)) { return; } 1488 for (int32_t style = UDAT_FULL; style <= UDAT_SHORT; style++) { 1489 resStr = ures_getStringByIndex(dateTimePatterns.getAlias(), dateTimeOffset + style, &resStrLen, &status); 1490 setDateTimeFormat(static_cast<UDateFormatStyle>(style), UnicodeString(true, resStr, resStrLen), status); 1491 } 1492 } 1493 1494 void 1495 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) { 1496 DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status); 1497 if(U_SUCCESS(status)) { 1498 decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol); 1499 // NUL-terminate for the C API. 1500 decimal.getTerminatedBuffer(); 1501 } 1502 } 1503 1504 UDateTimePatternConflict 1505 DateTimePatternGenerator::addPattern( 1506 const UnicodeString& pattern, 1507 UBool override, 1508 UnicodeString &conflictingPattern, 1509 UErrorCode& status) 1510 { 1511 if (U_FAILURE(internalErrorCode)) { 1512 status = internalErrorCode; 1513 return UDATPG_NO_CONFLICT; 1514 } 1515 1516 return addPatternWithOptionalSkeleton(pattern, nullptr, override, conflictingPattern, status); 1517 } 1518 1519 UDateTimePatternConflict 1520 DateTimePatternGenerator::addPatternWithSkeleton( 1521 const UnicodeString& pattern, 1522 const UnicodeString& skeletonToUse, 1523 UBool override, 1524 UnicodeString& conflictingPattern, 1525 UErrorCode& status) 1526 { 1527 return addPatternWithOptionalSkeleton(pattern, &skeletonToUse, override, conflictingPattern, status); 1528 } 1529 1530 // For DateTimePatternGenerator::addPatternWithSkeleton - 1531 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case: 1532 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern. 1533 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified 1534 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override 1535 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual 1536 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was 1537 // derived (i.e. entries derived from the standard date/time patters for the specified locale). 1538 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a 1539 // specified skeleton (which sets a new field in the PtnElem in the PatternMap). 1540 UDateTimePatternConflict 1541 DateTimePatternGenerator::addPatternWithOptionalSkeleton( 1542 const UnicodeString& pattern, 1543 const UnicodeString* skeletonToUse, 1544 UBool override, 1545 UnicodeString& conflictingPattern, 1546 UErrorCode& status) 1547 { 1548 if (U_FAILURE(internalErrorCode)) { 1549 status = internalErrorCode; 1550 return UDATPG_NO_CONFLICT; 1551 } 1552 1553 UnicodeString basePattern; 1554 PtnSkeleton skeleton; 1555 UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT; 1556 1557 DateTimeMatcher matcher; 1558 if ( skeletonToUse == nullptr ) { 1559 matcher.set(pattern, fp, skeleton); 1560 matcher.getBasePattern(basePattern); 1561 } else { 1562 matcher.set(*skeletonToUse, fp, skeleton); // no longer trims skeleton fields to max len 3, per #7930 1563 matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse; 1564 } 1565 // We only care about base conflicts - and replacing the pattern associated with a base - if: 1566 // 1. the conflicting previous base pattern did *not* have an explicit skeleton; in that case the previous 1567 // base + pattern combination was derived from either (a) a canonical item, (b) a standard format, or 1568 // (c) a pattern specified programmatically with a previous call to addPattern (which would only happen 1569 // if we are getting here from a subsequent call to addPattern). 1570 // 2. a skeleton is specified for the current pattern, but override=false; in that case we are checking 1571 // availableFormats items from root, which should not override any previous entry with the same base. 1572 UBool entryHadSpecifiedSkeleton; 1573 const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton); 1574 if (duplicatePattern != nullptr && (!entryHadSpecifiedSkeleton || (skeletonToUse != nullptr && !override))) { 1575 conflictingStatus = UDATPG_BASE_CONFLICT; 1576 conflictingPattern = *duplicatePattern; 1577 if (!override) { 1578 return conflictingStatus; 1579 } 1580 } 1581 // The only time we get here with override=true and skeletonToUse!=null is when adding availableFormats 1582 // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with 1583 // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for 1584 // the previously-specified conflicting item. 1585 const PtnSkeleton* entrySpecifiedSkeleton = nullptr; 1586 duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton); 1587 if (duplicatePattern != nullptr ) { 1588 conflictingStatus = UDATPG_CONFLICT; 1589 conflictingPattern = *duplicatePattern; 1590 if (!override || (skeletonToUse != nullptr && entrySpecifiedSkeleton != nullptr)) { 1591 return conflictingStatus; 1592 } 1593 } 1594 patternMap->add(basePattern, skeleton, pattern, skeletonToUse != nullptr, status); 1595 if(U_FAILURE(status)) { 1596 return conflictingStatus; 1597 } 1598 1599 return UDATPG_NO_CONFLICT; 1600 } 1601 1602 1603 UDateTimePatternField 1604 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const { 1605 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) { 1606 if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) { 1607 return static_cast<UDateTimePatternField>(i); 1608 } 1609 } 1610 return UDATPG_FIELD_COUNT; 1611 } 1612 1613 UDateTimePatternField 1614 DateTimePatternGenerator::getFieldAndWidthIndices(const char* key, UDateTimePGDisplayWidth* widthP) const { 1615 char cldrFieldKey[UDATPG_FIELD_KEY_MAX + 1]; 1616 uprv_strncpy(cldrFieldKey, key, UDATPG_FIELD_KEY_MAX); 1617 cldrFieldKey[UDATPG_FIELD_KEY_MAX]=0; // ensure termination 1618 *widthP = UDATPG_WIDE; 1619 char* hyphenPtr = uprv_strchr(cldrFieldKey, '-'); 1620 if (hyphenPtr) { 1621 for (int32_t i=UDATPG_WIDTH_COUNT-1; i>0; --i) { 1622 if (uprv_strcmp(CLDR_FIELD_WIDTH[i], hyphenPtr)==0) { 1623 *widthP = static_cast<UDateTimePGDisplayWidth>(i); 1624 break; 1625 } 1626 } 1627 *hyphenPtr = 0; // now delete width portion of key 1628 } 1629 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) { 1630 if (uprv_strcmp(CLDR_FIELD_NAME[i],cldrFieldKey)==0) { 1631 return static_cast<UDateTimePatternField>(i); 1632 } 1633 } 1634 return UDATPG_FIELD_COUNT; 1635 } 1636 1637 const UnicodeString* 1638 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source, 1639 int32_t includeMask, 1640 DistanceInfo* missingFields, 1641 UErrorCode &status, 1642 const PtnSkeleton** specifiedSkeletonPtr) { 1643 int32_t bestDistance = 0x7fffffff; 1644 int32_t bestMissingFieldMask = -1; 1645 DistanceInfo tempInfo; 1646 const UnicodeString *bestPattern=nullptr; 1647 const PtnSkeleton* specifiedSkeleton=nullptr; 1648 1649 PatternMapIterator it(status); 1650 if (U_FAILURE(status)) { return nullptr; } 1651 1652 for (it.set(*patternMap); it.hasNext(); ) { 1653 DateTimeMatcher trial = it.next(); 1654 if (trial.equals(skipMatcher)) { 1655 continue; 1656 } 1657 int32_t distance=source.getDistance(trial, includeMask, tempInfo); 1658 // Because we iterate over a map the order is undefined. Can change between implementations, 1659 // versions, and will very likely be different between Java and C/C++. 1660 // So if we have patterns with the same distance we also look at the missingFieldMask, 1661 // and we favour the smallest one. Because the field is a bitmask this technically means we 1662 // favour differences in the "least significant fields". For example we prefer the one with differences 1663 // in seconds field vs one with difference in the hours field. 1664 if (distance<bestDistance || (distance==bestDistance && bestMissingFieldMask<tempInfo.missingFieldMask)) { 1665 bestDistance=distance; 1666 bestMissingFieldMask=tempInfo.missingFieldMask; 1667 bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton); 1668 missingFields->setTo(tempInfo); 1669 if (distance==0) { 1670 break; 1671 } 1672 } 1673 } 1674 1675 // If the best raw match had a specified skeleton and that skeleton was requested by the caller, 1676 // then return it too. This generally happens when the caller needs to pass that skeleton 1677 // through to adjustFieldTypes so the latter can do a better job. 1678 if (bestPattern && specifiedSkeletonPtr) { 1679 *specifiedSkeletonPtr = specifiedSkeleton; 1680 } 1681 return bestPattern; 1682 } 1683 1684 UnicodeString 1685 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern, 1686 const PtnSkeleton* specifiedSkeleton, 1687 int32_t flags, 1688 UDateTimePatternMatchOptions options) { 1689 UnicodeString newPattern; 1690 fp->set(pattern); 1691 for (int32_t i=0; i < fp->itemNumber; i++) { 1692 UnicodeString field = fp->items[i]; 1693 if ( fp->isQuoteLiteral(field) ) { 1694 1695 UnicodeString quoteLiteral; 1696 fp->getQuoteLiteral(quoteLiteral, &i); 1697 newPattern += quoteLiteral; 1698 } 1699 else { 1700 if (fp->isPatternSeparator(field)) { 1701 newPattern+=field; 1702 continue; 1703 } 1704 int32_t canonicalIndex = fp->getCanonicalIndex(field); 1705 if (canonicalIndex < 0) { 1706 newPattern+=field; 1707 continue; // don't adjust 1708 } 1709 const dtTypeElem *row = &dtTypes[canonicalIndex]; 1710 int32_t typeValue = row->field; 1711 1712 // handle day periods - with #13183, no longer need special handling here, integrated with normal types 1713 1714 if ((flags & kDTPGFixFractionalSeconds) != 0 && typeValue == UDATPG_SECOND_FIELD) { 1715 field += decimal; 1716 dtMatcher->skeleton.original.appendFieldTo(UDATPG_FRACTIONAL_SECOND_FIELD, field); 1717 } else if (dtMatcher->skeleton.type[typeValue]!=0) { 1718 // Here: 1719 // - "reqField" is the field from the originally requested skeleton after replacement 1720 // of metacharacters 'j', 'C' and 'J', with length "reqFieldLen". 1721 // - "field" is the field from the found pattern. 1722 // 1723 // The adjusted field should consist of characters from the originally requested 1724 // skeleton, except in the case of UDATPG_MONTH_FIELD or 1725 // UDATPG_WEEKDAY_FIELD or UDATPG_YEAR_FIELD, in which case it should consist 1726 // of characters from the found pattern. In some cases of UDATPG_HOUR_FIELD, 1727 // there is adjustment following the "defaultHourFormatChar". There is explanation 1728 // how it is done below. 1729 // 1730 // The length of the adjusted field (adjFieldLen) should match that in the originally 1731 // requested skeleton, except that in the following cases the length of the adjusted field 1732 // should match that in the found pattern (i.e. the length of this pattern field should 1733 // not be adjusted): 1734 // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is 1735 // not set (ticket #7180). Note, we may want to implement a similar change for other 1736 // numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for 1737 // field length, but options bits can be used to override this. 1738 // 2. There is a specified skeleton for the found pattern and one of the following is true: 1739 // a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen. 1740 // b) The pattern field is numeric and the requested field is not, or vice versa. 1741 1742 char16_t reqFieldChar = dtMatcher->skeleton.original.getFieldChar(typeValue); 1743 int32_t reqFieldLen = dtMatcher->skeleton.original.getFieldLength(typeValue); 1744 if (reqFieldChar == CAP_E && reqFieldLen < 3) 1745 reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e 1746 int32_t adjFieldLen = reqFieldLen; 1747 if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) || 1748 (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) || 1749 (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) { 1750 adjFieldLen = field.length(); 1751 } else if (specifiedSkeleton && reqFieldChar != LOW_C && reqFieldChar != LOW_E) { 1752 // (we skip this section for 'c' and 'e' because unlike the other characters considered in this function, 1753 // they have no minimum field length-- 'E' and 'EE' are equivalent to 'EEE', but 'e' and 'ee' are not 1754 // equivalent to 'eee' -- see the entries for "week day" in 1755 // https://www.unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table for more info) 1756 int32_t skelFieldLen = specifiedSkeleton->original.getFieldLength(typeValue); 1757 UBool patFieldIsNumeric = (row->type > 0); 1758 UBool reqFieldIsNumeric = (dtMatcher->skeleton.type[typeValue] > 0); 1759 if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !reqFieldIsNumeric) || (reqFieldIsNumeric && !patFieldIsNumeric)) { 1760 // don't adjust the field length in the found pattern 1761 adjFieldLen = field.length(); 1762 } 1763 } 1764 char16_t c = (typeValue!= UDATPG_HOUR_FIELD 1765 && typeValue!= UDATPG_MONTH_FIELD 1766 && typeValue!= UDATPG_WEEKDAY_FIELD 1767 && (typeValue!= UDATPG_YEAR_FIELD || reqFieldChar==CAP_Y)) 1768 ? reqFieldChar 1769 : field.charAt(0); 1770 if (c == CAP_E && adjFieldLen < 3) { 1771 c = LOW_E; 1772 } 1773 if (typeValue == UDATPG_HOUR_FIELD && fDefaultHourFormatChar != 0) { 1774 // The adjustment here is required to match spec (https://www.unicode.org/reports/tr35/tr35-dates.html#dfst-hour). 1775 // It is necessary to match the hour-cycle preferred by the Locale. 1776 // Given that, we need to do the following adjustments: 1777 // 1. When hour-cycle is h11 it should replace 'h' by 'K'. 1778 // 2. When hour-cycle is h23 it should replace 'H' by 'k'. 1779 // 3. When hour-cycle is h24 it should replace 'k' by 'H'. 1780 // 4. When hour-cycle is h12 it should replace 'K' by 'h'. 1781 1782 if ((flags & kDTPGSkeletonUsesCapJ) != 0 || reqFieldChar == fDefaultHourFormatChar) { 1783 c = fDefaultHourFormatChar; 1784 } else if (reqFieldChar == LOW_H && fDefaultHourFormatChar == CAP_K) { 1785 c = CAP_K; 1786 } else if (reqFieldChar == CAP_H && fDefaultHourFormatChar == LOW_K) { 1787 c = LOW_K; 1788 } else if (reqFieldChar == LOW_K && fDefaultHourFormatChar == CAP_H) { 1789 c = CAP_H; 1790 } else if (reqFieldChar == CAP_K && fDefaultHourFormatChar == LOW_H) { 1791 c = LOW_H; 1792 } 1793 } 1794 1795 field.remove(); 1796 for (int32_t j=adjFieldLen; j>0; --j) { 1797 field += c; 1798 } 1799 } 1800 newPattern+=field; 1801 } 1802 } 1803 return newPattern; 1804 } 1805 1806 UnicodeString 1807 DateTimePatternGenerator::getBestAppending(int32_t missingFields, int32_t flags, UErrorCode &status, UDateTimePatternMatchOptions options) { 1808 if (U_FAILURE(status)) { 1809 return {}; 1810 } 1811 UnicodeString resultPattern, tempPattern; 1812 const UnicodeString* tempPatternPtr; 1813 int32_t lastMissingFieldMask=0; 1814 if (missingFields!=0) { 1815 resultPattern=UnicodeString(); 1816 const PtnSkeleton* specifiedSkeleton=nullptr; 1817 tempPatternPtr = getBestRaw(*dtMatcher, missingFields, distanceInfo, status, &specifiedSkeleton); 1818 if (U_FAILURE(status)) { 1819 return {}; 1820 } 1821 tempPattern = *tempPatternPtr; 1822 resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options); 1823 if ( distanceInfo->missingFieldMask==0 ) { 1824 return resultPattern; 1825 } 1826 while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work! 1827 if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) { 1828 break; // cannot find the proper missing field 1829 } 1830 if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) && 1831 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) { 1832 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, flags | kDTPGFixFractionalSeconds, options); 1833 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK; 1834 continue; 1835 } 1836 int32_t startingMask = distanceInfo->missingFieldMask; 1837 tempPatternPtr = getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, status, &specifiedSkeleton); 1838 if (U_FAILURE(status)) { 1839 return {}; 1840 } 1841 tempPattern = *tempPatternPtr; 1842 tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options); 1843 int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask; 1844 int32_t topField=getTopBitNumber(foundMask); 1845 1846 if (appendItemFormats[topField].length() != 0) { 1847 UnicodeString appendName; 1848 getAppendName(static_cast<UDateTimePatternField>(topField), appendName); 1849 const UnicodeString *values[3] = { 1850 &resultPattern, 1851 &tempPattern, 1852 &appendName 1853 }; 1854 SimpleFormatter(appendItemFormats[topField], 2, 3, status). 1855 formatAndReplace(values, 3, resultPattern, nullptr, 0, status); 1856 } 1857 lastMissingFieldMask = distanceInfo->missingFieldMask; 1858 } 1859 } 1860 return resultPattern; 1861 } 1862 1863 int32_t 1864 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) const { 1865 if ( foundMask==0 ) { 1866 return 0; 1867 } 1868 int32_t i=0; 1869 while (foundMask!=0) { 1870 foundMask >>=1; 1871 ++i; 1872 } 1873 if (i-1 >UDATPG_ZONE_FIELD) { 1874 return UDATPG_ZONE_FIELD; 1875 } 1876 else 1877 return i-1; 1878 } 1879 1880 void 1881 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err) 1882 { 1883 fAvailableFormatKeyHash->puti(key, 1, err); 1884 } 1885 1886 UBool 1887 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const { 1888 return fAvailableFormatKeyHash->geti(key) == 1; 1889 } 1890 1891 void 1892 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) { 1893 if (other == nullptr || U_FAILURE(status)) { 1894 return; 1895 } 1896 if (fAvailableFormatKeyHash != nullptr) { 1897 delete fAvailableFormatKeyHash; 1898 fAvailableFormatKeyHash = nullptr; 1899 } 1900 initHashtable(status); 1901 if(U_FAILURE(status)){ 1902 return; 1903 } 1904 int32_t pos = UHASH_FIRST; 1905 const UHashElement* elem = nullptr; 1906 // walk through the hash table and create a deep clone 1907 while((elem = other->nextElement(pos))!= nullptr){ 1908 const UHashTok otherKeyTok = elem->key; 1909 UnicodeString* otherKey = static_cast<UnicodeString*>(otherKeyTok.pointer); 1910 fAvailableFormatKeyHash->puti(*otherKey, 1, status); 1911 if(U_FAILURE(status)){ 1912 return; 1913 } 1914 } 1915 } 1916 1917 StringEnumeration* 1918 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const { 1919 if (U_FAILURE(status)) { 1920 return nullptr; 1921 } 1922 if (U_FAILURE(internalErrorCode)) { 1923 status = internalErrorCode; 1924 return nullptr; 1925 } 1926 LocalPointer<StringEnumeration> skeletonEnumerator( 1927 new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status), status); 1928 1929 return U_SUCCESS(status) ? skeletonEnumerator.orphan() : nullptr; 1930 } 1931 1932 const UnicodeString& 1933 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const { 1934 PtnElem *curElem; 1935 1936 if (skeleton.length() ==0) { 1937 return emptyString; 1938 } 1939 curElem = patternMap->getHeader(skeleton.charAt(0)); 1940 while ( curElem != nullptr ) { 1941 if ( curElem->skeleton->getSkeleton()==skeleton ) { 1942 return curElem->pattern; 1943 } 1944 curElem = curElem->next.getAlias(); 1945 } 1946 return emptyString; 1947 } 1948 1949 StringEnumeration* 1950 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const { 1951 if (U_FAILURE(status)) { 1952 return nullptr; 1953 } 1954 if (U_FAILURE(internalErrorCode)) { 1955 status = internalErrorCode; 1956 return nullptr; 1957 } 1958 LocalPointer<StringEnumeration> baseSkeletonEnumerator( 1959 new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status), status); 1960 1961 return U_SUCCESS(status) ? baseSkeletonEnumerator.orphan() : nullptr; 1962 } 1963 1964 StringEnumeration* 1965 DateTimePatternGenerator::getRedundants(UErrorCode& status) { 1966 if (U_FAILURE(status)) { return nullptr; } 1967 if (U_FAILURE(internalErrorCode)) { 1968 status = internalErrorCode; 1969 return nullptr; 1970 } 1971 LocalPointer<StringEnumeration> output(new DTRedundantEnumeration(), status); 1972 if (U_FAILURE(status)) { return nullptr; } 1973 const UnicodeString *pattern; 1974 PatternMapIterator it(status); 1975 if (U_FAILURE(status)) { return nullptr; } 1976 1977 for (it.set(*patternMap); it.hasNext(); ) { 1978 DateTimeMatcher current = it.next(); 1979 pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton())); 1980 if ( isCanonicalItem(*pattern) ) { 1981 continue; 1982 } 1983 if ( skipMatcher == nullptr ) { 1984 skipMatcher = new DateTimeMatcher(current); 1985 if (skipMatcher == nullptr) { 1986 status = U_MEMORY_ALLOCATION_ERROR; 1987 return nullptr; 1988 } 1989 } 1990 else { 1991 *skipMatcher = current; 1992 } 1993 UnicodeString trial = getBestPattern(current.getPattern(), status); 1994 if (U_FAILURE(status)) { return nullptr; } 1995 if (trial == *pattern) { 1996 ((DTRedundantEnumeration *)output.getAlias())->add(*pattern, status); 1997 if (U_FAILURE(status)) { return nullptr; } 1998 } 1999 if (current.equals(skipMatcher)) { 2000 continue; 2001 } 2002 } 2003 return output.orphan(); 2004 } 2005 2006 UBool 2007 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const { 2008 if ( item.length() != 1 ) { 2009 return false; 2010 } 2011 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) { 2012 if (item.charAt(0)==Canonical_Items[i]) { 2013 return true; 2014 } 2015 } 2016 return false; 2017 } 2018 2019 2020 DateTimePatternGenerator* 2021 DateTimePatternGenerator::clone() const { 2022 return new DateTimePatternGenerator(*this); 2023 } 2024 2025 PatternMap::PatternMap() { 2026 for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) { 2027 boot[i] = nullptr; 2028 } 2029 isDupAllowed = true; 2030 } 2031 2032 void 2033 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) { 2034 if (U_FAILURE(status)) { 2035 return; 2036 } 2037 this->isDupAllowed = other.isDupAllowed; 2038 for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) { 2039 PtnElem *curElem, *otherElem, *prevElem=nullptr; 2040 otherElem = other.boot[bootIndex]; 2041 while (otherElem != nullptr) { 2042 LocalPointer<PtnElem> newElem(new PtnElem(otherElem->basePattern, otherElem->pattern), status); 2043 if (U_FAILURE(status)) { 2044 return; // out of memory 2045 } 2046 newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(*(otherElem->skeleton)), status); 2047 if (U_FAILURE(status)) { 2048 return; // out of memory 2049 } 2050 newElem->skeletonWasSpecified = otherElem->skeletonWasSpecified; 2051 2052 // Release ownership from the LocalPointer of the PtnElem object. 2053 // The PtnElem will now be owned by either the boot (for the first entry in the linked-list) 2054 // or owned by the previous PtnElem object in the linked-list. 2055 curElem = newElem.orphan(); 2056 2057 if (this->boot[bootIndex] == nullptr) { 2058 this->boot[bootIndex] = curElem; 2059 } else { 2060 if (prevElem != nullptr) { 2061 prevElem->next.adoptInstead(curElem); 2062 } else { 2063 UPRV_UNREACHABLE_EXIT; 2064 } 2065 } 2066 prevElem = curElem; 2067 otherElem = otherElem->next.getAlias(); 2068 } 2069 2070 } 2071 } 2072 2073 PtnElem* 2074 PatternMap::getHeader(char16_t baseChar) const { 2075 PtnElem* curElem; 2076 2077 if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) { 2078 curElem = boot[baseChar-CAP_A]; 2079 } 2080 else { 2081 if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) { 2082 curElem = boot[26+baseChar-LOW_A]; 2083 } 2084 else { 2085 return nullptr; 2086 } 2087 } 2088 return curElem; 2089 } 2090 2091 PatternMap::~PatternMap() { 2092 for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) { 2093 if (boot[i] != nullptr ) { 2094 delete boot[i]; 2095 boot[i] = nullptr; 2096 } 2097 } 2098 } // PatternMap destructor 2099 2100 void 2101 PatternMap::add(const UnicodeString& basePattern, 2102 const PtnSkeleton& skeleton, 2103 const UnicodeString& value,// mapped pattern value 2104 UBool skeletonWasSpecified, 2105 UErrorCode &status) { 2106 char16_t baseChar = basePattern.charAt(0); 2107 PtnElem *curElem, *baseElem; 2108 status = U_ZERO_ERROR; 2109 2110 // the baseChar must be A-Z or a-z 2111 if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) { 2112 baseElem = boot[baseChar-CAP_A]; 2113 } 2114 else { 2115 if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) { 2116 baseElem = boot[26+baseChar-LOW_A]; 2117 } 2118 else { 2119 status = U_ILLEGAL_CHARACTER; 2120 return; 2121 } 2122 } 2123 2124 if (baseElem == nullptr) { 2125 LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status); 2126 if (U_FAILURE(status)) { 2127 return; // out of memory 2128 } 2129 newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status); 2130 if (U_FAILURE(status)) { 2131 return; // out of memory 2132 } 2133 newElem->skeletonWasSpecified = skeletonWasSpecified; 2134 if (baseChar >= LOW_A) { 2135 boot[26 + (baseChar - LOW_A)] = newElem.orphan(); // the boot array now owns the PtnElem. 2136 } 2137 else { 2138 boot[baseChar - CAP_A] = newElem.orphan(); // the boot array now owns the PtnElem. 2139 } 2140 } 2141 if ( baseElem != nullptr ) { 2142 curElem = getDuplicateElem(basePattern, skeleton, baseElem); 2143 2144 if (curElem == nullptr) { 2145 // add new element to the list. 2146 curElem = baseElem; 2147 while( curElem -> next != nullptr ) 2148 { 2149 curElem = curElem->next.getAlias(); 2150 } 2151 2152 LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status); 2153 if (U_FAILURE(status)) { 2154 return; // out of memory 2155 } 2156 newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status); 2157 if (U_FAILURE(status)) { 2158 return; // out of memory 2159 } 2160 newElem->skeletonWasSpecified = skeletonWasSpecified; 2161 curElem->next.adoptInstead(newElem.orphan()); 2162 curElem = curElem->next.getAlias(); 2163 } 2164 else { 2165 // Pattern exists in the list already. 2166 if ( !isDupAllowed ) { 2167 return; 2168 } 2169 // Overwrite the value. 2170 curElem->pattern = value; 2171 // It was a bug that we were not doing the following previously, 2172 // though that bug hid other problems by making things partly work. 2173 curElem->skeletonWasSpecified = skeletonWasSpecified; 2174 } 2175 } 2176 } // PatternMap::add 2177 2178 // Find the pattern from the given basePattern string. 2179 const UnicodeString * 2180 PatternMap::getPatternFromBasePattern(const UnicodeString& basePattern, UBool& skeletonWasSpecified) const { // key to search for 2181 PtnElem *curElem; 2182 2183 if ((curElem=getHeader(basePattern.charAt(0)))==nullptr) { 2184 return nullptr; // no match 2185 } 2186 2187 do { 2188 if ( basePattern.compare(curElem->basePattern)==0 ) { 2189 skeletonWasSpecified = curElem->skeletonWasSpecified; 2190 return &(curElem->pattern); 2191 } 2192 curElem = curElem->next.getAlias(); 2193 } while (curElem != nullptr); 2194 2195 return nullptr; 2196 } // PatternMap::getFromBasePattern 2197 2198 2199 // Find the pattern from the given skeleton. 2200 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-nullptr), 2201 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw) 2202 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the 2203 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is nullptr), 2204 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily. 2205 const UnicodeString * 2206 PatternMap::getPatternFromSkeleton(const PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) const { // key to search for 2207 PtnElem *curElem; 2208 2209 if (specifiedSkeletonPtr) { 2210 *specifiedSkeletonPtr = nullptr; 2211 } 2212 2213 // find boot entry 2214 char16_t baseChar = skeleton.getFirstChar(); 2215 if ((curElem=getHeader(baseChar))==nullptr) { 2216 return nullptr; // no match 2217 } 2218 2219 do { 2220 UBool equal; 2221 if (specifiedSkeletonPtr != nullptr) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original 2222 equal = curElem->skeleton->original == skeleton.original; 2223 } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal 2224 equal = curElem->skeleton->baseOriginal == skeleton.baseOriginal; 2225 } 2226 if (equal) { 2227 if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) { 2228 *specifiedSkeletonPtr = curElem->skeleton.getAlias(); 2229 } 2230 return &(curElem->pattern); 2231 } 2232 curElem = curElem->next.getAlias(); 2233 } while (curElem != nullptr); 2234 2235 return nullptr; 2236 } 2237 2238 UBool 2239 PatternMap::equals(const PatternMap& other) const { 2240 if ( this==&other ) { 2241 return true; 2242 } 2243 for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) { 2244 if (boot[bootIndex] == other.boot[bootIndex]) { 2245 continue; 2246 } 2247 if ((boot[bootIndex] == nullptr) || (other.boot[bootIndex] == nullptr)) { 2248 return false; 2249 } 2250 PtnElem *otherElem = other.boot[bootIndex]; 2251 PtnElem *myElem = boot[bootIndex]; 2252 while ((otherElem != nullptr) || (myElem != nullptr)) { 2253 if ( myElem == otherElem ) { 2254 break; 2255 } 2256 if ((otherElem == nullptr) || (myElem == nullptr)) { 2257 return false; 2258 } 2259 if ( (myElem->basePattern != otherElem->basePattern) || 2260 (myElem->pattern != otherElem->pattern) ) { 2261 return false; 2262 } 2263 if ((myElem->skeleton.getAlias() != otherElem->skeleton.getAlias()) && 2264 !myElem->skeleton->equals(*(otherElem->skeleton))) { 2265 return false; 2266 } 2267 myElem = myElem->next.getAlias(); 2268 otherElem = otherElem->next.getAlias(); 2269 } 2270 } 2271 return true; 2272 } 2273 2274 // find any key existing in the mapping table already. 2275 // return true if there is an existing key, otherwise return false. 2276 PtnElem* 2277 PatternMap::getDuplicateElem( 2278 const UnicodeString &basePattern, 2279 const PtnSkeleton &skeleton, 2280 PtnElem *baseElem) { 2281 PtnElem *curElem; 2282 2283 if ( baseElem == nullptr ) { 2284 return nullptr; 2285 } 2286 else { 2287 curElem = baseElem; 2288 } 2289 do { 2290 if ( basePattern.compare(curElem->basePattern)==0 ) { 2291 UBool isEqual = true; 2292 for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) { 2293 if (curElem->skeleton->type[i] != skeleton.type[i] ) { 2294 isEqual = false; 2295 break; 2296 } 2297 } 2298 if (isEqual) { 2299 return curElem; 2300 } 2301 } 2302 curElem = curElem->next.getAlias(); 2303 } while( curElem != nullptr ); 2304 2305 // end of the list 2306 return nullptr; 2307 2308 } // PatternMap::getDuplicateElem 2309 2310 DateTimeMatcher::DateTimeMatcher() { 2311 } 2312 2313 DateTimeMatcher::~DateTimeMatcher() {} 2314 2315 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) { 2316 copyFrom(other.skeleton); 2317 } 2318 2319 DateTimeMatcher& DateTimeMatcher::operator=(const DateTimeMatcher& other) { 2320 if (this != &other) { 2321 copyFrom(other.skeleton); 2322 } 2323 return *this; 2324 } 2325 2326 2327 void 2328 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) { 2329 PtnSkeleton localSkeleton; 2330 return set(pattern, fp, localSkeleton); 2331 } 2332 2333 void 2334 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) { 2335 int32_t i; 2336 for (i=0; i<UDATPG_FIELD_COUNT; ++i) { 2337 skeletonResult.type[i] = NONE; 2338 } 2339 skeletonResult.original.clear(); 2340 skeletonResult.baseOriginal.clear(); 2341 skeletonResult.addedDefaultDayPeriod = false; 2342 2343 fp->set(pattern); 2344 for (i=0; i < fp->itemNumber; i++) { 2345 const UnicodeString& value = fp->items[i]; 2346 // don't skip 'a' anymore, dayPeriod handled specially below 2347 2348 if ( fp->isQuoteLiteral(value) ) { 2349 UnicodeString quoteLiteral; 2350 fp->getQuoteLiteral(quoteLiteral, &i); 2351 continue; 2352 } 2353 int32_t canonicalIndex = fp->getCanonicalIndex(value); 2354 if (canonicalIndex < 0) { 2355 continue; 2356 } 2357 const dtTypeElem *row = &dtTypes[canonicalIndex]; 2358 int32_t field = row->field; 2359 skeletonResult.original.populate(field, value); 2360 char16_t repeatChar = row->patternChar; 2361 int32_t repeatCount = row->minLen; 2362 skeletonResult.baseOriginal.populate(field, repeatChar, repeatCount); 2363 int16_t subField = row->type; 2364 if (row->type > 0) { 2365 U_ASSERT(value.length() < INT16_MAX); 2366 subField += static_cast<int16_t>(value.length()); 2367 } 2368 skeletonResult.type[field] = subField; 2369 } 2370 2371 // #20739, we have a skeleton with minutes and milliseconds, but no seconds 2372 // 2373 // Theoretically we would need to check and fix all fields with "gaps": 2374 // for example year-day (no month), month-hour (no day), and so on, All the possible field combinations. 2375 // Plus some smartness: year + hour => should we add month, or add day-of-year? 2376 // What about month + day-of-week, or month + am/pm indicator. 2377 // I think beyond a certain point we should not try to fix bad developer input and try guessing what they mean. 2378 // Garbage in, garbage out. 2379 if (!skeletonResult.original.isFieldEmpty(UDATPG_MINUTE_FIELD) 2380 && !skeletonResult.original.isFieldEmpty(UDATPG_FRACTIONAL_SECOND_FIELD) 2381 && skeletonResult.original.isFieldEmpty(UDATPG_SECOND_FIELD)) { 2382 // Force the use of seconds 2383 for (i = 0; dtTypes[i].patternChar != 0; i++) { 2384 if (dtTypes[i].field == UDATPG_SECOND_FIELD) { 2385 // first entry for UDATPG_SECOND_FIELD 2386 skeletonResult.original.populate(UDATPG_SECOND_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen); 2387 skeletonResult.baseOriginal.populate(UDATPG_SECOND_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen); 2388 // We add value.length, same as above, when type is first initialized. 2389 // The value we want to "fake" here is "s", and 1 means "s".length() 2390 int16_t subField = dtTypes[i].type; 2391 skeletonResult.type[UDATPG_SECOND_FIELD] = (subField > 0) ? subField + 1 : subField; 2392 break; 2393 } 2394 } 2395 } 2396 2397 // #13183, handle special behavior for day period characters (a, b, B) 2398 if (!skeletonResult.original.isFieldEmpty(UDATPG_HOUR_FIELD)) { 2399 if (skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==LOW_H || skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==CAP_K) { 2400 // We have a skeleton with 12-hour-cycle format 2401 if (skeletonResult.original.isFieldEmpty(UDATPG_DAYPERIOD_FIELD)) { 2402 // But we do not have a day period in the skeleton; add the default DAYPERIOD (currently "a") 2403 for (i = 0; dtTypes[i].patternChar != 0; i++) { 2404 if ( dtTypes[i].field == UDATPG_DAYPERIOD_FIELD ) { 2405 // first entry for UDATPG_DAYPERIOD_FIELD 2406 skeletonResult.original.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen); 2407 skeletonResult.baseOriginal.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen); 2408 skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = dtTypes[i].type; 2409 skeletonResult.addedDefaultDayPeriod = true; 2410 break; 2411 } 2412 } 2413 } 2414 } else { 2415 // Skeleton has 24-hour-cycle hour format and has dayPeriod, delete dayPeriod (i.e. ignore it) 2416 skeletonResult.original.clearField(UDATPG_DAYPERIOD_FIELD); 2417 skeletonResult.baseOriginal.clearField(UDATPG_DAYPERIOD_FIELD); 2418 skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = NONE; 2419 } 2420 } 2421 copyFrom(skeletonResult); 2422 } 2423 2424 void 2425 DateTimeMatcher::getBasePattern(UnicodeString &result ) { 2426 result.remove(); // Reset the result first. 2427 skeleton.baseOriginal.appendTo(result); 2428 } 2429 2430 UnicodeString 2431 DateTimeMatcher::getPattern() { 2432 UnicodeString result; 2433 return skeleton.original.appendTo(result); 2434 } 2435 2436 int32_t 2437 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) const { 2438 int32_t result = 0; 2439 distanceInfo.clear(); 2440 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) { 2441 int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i]; 2442 int32_t otherType = other.skeleton.type[i]; 2443 if (myType==otherType) { 2444 continue; 2445 } 2446 if (myType==0) {// and other is not 2447 result += EXTRA_FIELD; 2448 distanceInfo.addExtra(i); 2449 } 2450 else { 2451 if (otherType==0) { 2452 result += MISSING_FIELD; 2453 distanceInfo.addMissing(i); 2454 } 2455 else { 2456 result += abs(myType - otherType); 2457 } 2458 } 2459 2460 } 2461 return result; 2462 } 2463 2464 void 2465 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) { 2466 skeleton.copyFrom(newSkeleton); 2467 } 2468 2469 void 2470 DateTimeMatcher::copyFrom() { 2471 // same as clear 2472 skeleton.clear(); 2473 } 2474 2475 UBool 2476 DateTimeMatcher::equals(const DateTimeMatcher* other) const { 2477 if (other==nullptr) { return false; } 2478 return skeleton.original == other->skeleton.original; 2479 } 2480 2481 int32_t 2482 DateTimeMatcher::getFieldMask() const { 2483 int32_t result = 0; 2484 2485 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) { 2486 if (skeleton.type[i]!=0) { 2487 result |= (1<<i); 2488 } 2489 } 2490 return result; 2491 } 2492 2493 PtnSkeleton* 2494 DateTimeMatcher::getSkeletonPtr() { 2495 return &skeleton; 2496 } 2497 2498 FormatParser::FormatParser () { 2499 status = START; 2500 itemNumber = 0; 2501 } 2502 2503 2504 FormatParser::~FormatParser () { 2505 } 2506 2507 2508 // Find the next token with the starting position and length 2509 // Note: the startPos may 2510 FormatParser::TokenStatus 2511 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) { 2512 int32_t curLoc = startPos; 2513 if ( curLoc >= pattern.length()) { 2514 return DONE; 2515 } 2516 // check the current char is between A-Z or a-z 2517 do { 2518 char16_t c=pattern.charAt(curLoc); 2519 if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) { 2520 curLoc++; 2521 } 2522 else { 2523 startPos = curLoc; 2524 *len=1; 2525 return ADD_TOKEN; 2526 } 2527 2528 if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) { 2529 break; // not the same token 2530 } 2531 } while(curLoc <= pattern.length()); 2532 *len = curLoc-startPos; 2533 return ADD_TOKEN; 2534 } 2535 2536 void 2537 FormatParser::set(const UnicodeString& pattern) { 2538 int32_t startPos = 0; 2539 TokenStatus result = START; 2540 int32_t len = 0; 2541 itemNumber = 0; 2542 2543 do { 2544 result = setTokens( pattern, startPos, &len ); 2545 if ( result == ADD_TOKEN ) 2546 { 2547 items[itemNumber++] = UnicodeString(pattern, startPos, len ); 2548 startPos += len; 2549 } 2550 else { 2551 break; 2552 } 2553 } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN); 2554 } 2555 2556 int32_t 2557 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) { 2558 int32_t len = s.length(); 2559 if (len == 0) { 2560 return -1; 2561 } 2562 char16_t ch = s.charAt(0); 2563 2564 // Verify that all are the same character. 2565 for (int32_t l = 1; l < len; l++) { 2566 if (ch != s.charAt(l)) { 2567 return -1; 2568 } 2569 } 2570 int32_t i = 0; 2571 int32_t bestRow = -1; 2572 while (dtTypes[i].patternChar != 0x0000) { 2573 if ( dtTypes[i].patternChar != ch ) { 2574 ++i; 2575 continue; 2576 } 2577 bestRow = i; 2578 if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) { 2579 return i; 2580 } 2581 if (dtTypes[i+1].minLen <= len) { 2582 ++i; 2583 continue; 2584 } 2585 return i; 2586 } 2587 return strict ? -1 : bestRow; 2588 } 2589 2590 UBool 2591 FormatParser::isQuoteLiteral(const UnicodeString& s) { 2592 return s.charAt(0) == SINGLE_QUOTE; 2593 } 2594 2595 // This function assumes the current itemIndex points to the quote literal. 2596 // Please call isQuoteLiteral prior to this function. 2597 void 2598 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) { 2599 int32_t i = *itemIndex; 2600 2601 quote.remove(); 2602 if (items[i].charAt(0)==SINGLE_QUOTE) { 2603 quote += items[i]; 2604 ++i; 2605 } 2606 while ( i < itemNumber ) { 2607 if ( items[i].charAt(0)==SINGLE_QUOTE ) { 2608 if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) { 2609 // two single quotes e.g. 'o''clock' 2610 quote += items[i++]; 2611 quote += items[i++]; 2612 continue; 2613 } 2614 else { 2615 quote += items[i]; 2616 break; 2617 } 2618 } 2619 else { 2620 quote += items[i]; 2621 } 2622 ++i; 2623 } 2624 *itemIndex=i; 2625 } 2626 2627 UBool 2628 FormatParser::isPatternSeparator(const UnicodeString& field) const { 2629 for (int32_t i=0; i<field.length(); ++i ) { 2630 char16_t c= field.charAt(i); 2631 if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) || 2632 (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) { 2633 continue; 2634 } 2635 else { 2636 return false; 2637 } 2638 } 2639 return true; 2640 } 2641 2642 DistanceInfo::~DistanceInfo() {} 2643 2644 void 2645 DistanceInfo::setTo(const DistanceInfo& other) { 2646 missingFieldMask = other.missingFieldMask; 2647 extraFieldMask= other.extraFieldMask; 2648 } 2649 2650 PatternMapIterator::PatternMapIterator(UErrorCode& status) : 2651 bootIndex(0), nodePtr(nullptr), matcher(nullptr), patternMap(nullptr) 2652 { 2653 if (U_FAILURE(status)) { return; } 2654 matcher.adoptInsteadAndCheckErrorCode(new DateTimeMatcher(), status); 2655 } 2656 2657 PatternMapIterator::~PatternMapIterator() { 2658 } 2659 2660 void 2661 PatternMapIterator::set(PatternMap& newPatternMap) { 2662 this->patternMap=&newPatternMap; 2663 } 2664 2665 PtnSkeleton* 2666 PatternMapIterator::getSkeleton() const { 2667 if ( nodePtr == nullptr ) { 2668 return nullptr; 2669 } 2670 else { 2671 return nodePtr->skeleton.getAlias(); 2672 } 2673 } 2674 2675 UBool 2676 PatternMapIterator::hasNext() const { 2677 int32_t headIndex = bootIndex; 2678 PtnElem *curPtr = nodePtr; 2679 2680 if (patternMap==nullptr) { 2681 return false; 2682 } 2683 while ( headIndex < MAX_PATTERN_ENTRIES ) { 2684 if ( curPtr != nullptr ) { 2685 if ( curPtr->next != nullptr ) { 2686 return true; 2687 } 2688 else { 2689 headIndex++; 2690 curPtr=nullptr; 2691 continue; 2692 } 2693 } 2694 else { 2695 if ( patternMap->boot[headIndex] != nullptr ) { 2696 return true; 2697 } 2698 else { 2699 headIndex++; 2700 continue; 2701 } 2702 } 2703 } 2704 return false; 2705 } 2706 2707 DateTimeMatcher& 2708 PatternMapIterator::next() { 2709 while ( bootIndex < MAX_PATTERN_ENTRIES ) { 2710 if ( nodePtr != nullptr ) { 2711 if ( nodePtr->next != nullptr ) { 2712 nodePtr = nodePtr->next.getAlias(); 2713 break; 2714 } 2715 else { 2716 bootIndex++; 2717 nodePtr=nullptr; 2718 continue; 2719 } 2720 } 2721 else { 2722 if ( patternMap->boot[bootIndex] != nullptr ) { 2723 nodePtr = patternMap->boot[bootIndex]; 2724 break; 2725 } 2726 else { 2727 bootIndex++; 2728 continue; 2729 } 2730 } 2731 } 2732 if (nodePtr!=nullptr) { 2733 matcher->copyFrom(*nodePtr->skeleton); 2734 } 2735 else { 2736 matcher->copyFrom(); 2737 } 2738 return *matcher; 2739 } 2740 2741 2742 SkeletonFields::SkeletonFields() { 2743 // Set initial values to zero 2744 clear(); 2745 } 2746 2747 void SkeletonFields::clear() { 2748 uprv_memset(chars, 0, sizeof(chars)); 2749 uprv_memset(lengths, 0, sizeof(lengths)); 2750 } 2751 2752 void SkeletonFields::copyFrom(const SkeletonFields& other) { 2753 uprv_memcpy(chars, other.chars, sizeof(chars)); 2754 uprv_memcpy(lengths, other.lengths, sizeof(lengths)); 2755 } 2756 2757 void SkeletonFields::clearField(int32_t field) { 2758 chars[field] = 0; 2759 lengths[field] = 0; 2760 } 2761 2762 char16_t SkeletonFields::getFieldChar(int32_t field) const { 2763 return chars[field]; 2764 } 2765 2766 int32_t SkeletonFields::getFieldLength(int32_t field) const { 2767 return lengths[field]; 2768 } 2769 2770 void SkeletonFields::populate(int32_t field, const UnicodeString& value) { 2771 populate(field, value.charAt(0), value.length()); 2772 } 2773 2774 void SkeletonFields::populate(int32_t field, char16_t ch, int32_t length) { 2775 chars[field] = static_cast<int8_t>(ch); 2776 lengths[field] = static_cast<int8_t>(length); 2777 } 2778 2779 UBool SkeletonFields::isFieldEmpty(int32_t field) const { 2780 return lengths[field] == 0; 2781 } 2782 2783 UnicodeString& SkeletonFields::appendTo(UnicodeString& string) const { 2784 for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) { 2785 appendFieldTo(i, string); 2786 } 2787 return string; 2788 } 2789 2790 UnicodeString& SkeletonFields::appendFieldTo(int32_t field, UnicodeString& string) const { 2791 char16_t ch(chars[field]); 2792 int32_t length = static_cast<int32_t>(lengths[field]); 2793 2794 for (int32_t i=0; i<length; i++) { 2795 string += ch; 2796 } 2797 return string; 2798 } 2799 2800 char16_t SkeletonFields::getFirstChar() const { 2801 for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) { 2802 if (lengths[i] != 0) { 2803 return chars[i]; 2804 } 2805 } 2806 return '\0'; 2807 } 2808 2809 2810 PtnSkeleton::PtnSkeleton() 2811 : addedDefaultDayPeriod(false) { 2812 } 2813 2814 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) { 2815 copyFrom(other); 2816 } 2817 2818 void PtnSkeleton::copyFrom(const PtnSkeleton& other) { 2819 uprv_memcpy(type, other.type, sizeof(type)); 2820 original.copyFrom(other.original); 2821 baseOriginal.copyFrom(other.baseOriginal); 2822 addedDefaultDayPeriod = other.addedDefaultDayPeriod; 2823 } 2824 2825 void PtnSkeleton::clear() { 2826 uprv_memset(type, 0, sizeof(type)); 2827 original.clear(); 2828 baseOriginal.clear(); 2829 } 2830 2831 UBool 2832 PtnSkeleton::equals(const PtnSkeleton& other) const { 2833 return (original == other.original) 2834 && (baseOriginal == other.baseOriginal) 2835 && (uprv_memcmp(type, other.type, sizeof(type)) == 0); 2836 } 2837 2838 UnicodeString 2839 PtnSkeleton::getSkeleton() const { 2840 UnicodeString result; 2841 result = original.appendTo(result); 2842 int32_t pos; 2843 if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) { 2844 // for backward compatibility: if DateTimeMatcher.set added a single 'a' that 2845 // was not in the provided skeleton, remove it here before returning skeleton. 2846 result.remove(pos, 1); 2847 } 2848 return result; 2849 } 2850 2851 UnicodeString 2852 PtnSkeleton::getBaseSkeleton() const { 2853 UnicodeString result; 2854 result = baseOriginal.appendTo(result); 2855 int32_t pos; 2856 if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) { 2857 // for backward compatibility: if DateTimeMatcher.set added a single 'a' that 2858 // was not in the provided skeleton, remove it here before returning skeleton. 2859 result.remove(pos, 1); 2860 } 2861 return result; 2862 } 2863 2864 char16_t 2865 PtnSkeleton::getFirstChar() const { 2866 return baseOriginal.getFirstChar(); 2867 } 2868 2869 PtnSkeleton::~PtnSkeleton() { 2870 } 2871 2872 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) : 2873 basePattern(basePat), skeleton(nullptr), pattern(pat), next(nullptr) 2874 { 2875 } 2876 2877 PtnElem::~PtnElem() { 2878 } 2879 2880 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap& patternMap, dtStrEnum type, UErrorCode& status) : fSkeletons(nullptr) { 2881 PtnElem *curElem; 2882 PtnSkeleton *curSkeleton; 2883 UnicodeString s; 2884 int32_t bootIndex; 2885 2886 pos=0; 2887 fSkeletons.adoptInsteadAndCheckErrorCode(new UVector(status), status); 2888 if (U_FAILURE(status)) { 2889 return; 2890 } 2891 2892 for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) { 2893 curElem = patternMap.boot[bootIndex]; 2894 while (curElem!=nullptr) { 2895 switch(type) { 2896 case DT_BASESKELETON: 2897 s=curElem->basePattern; 2898 break; 2899 case DT_PATTERN: 2900 s=curElem->pattern; 2901 break; 2902 case DT_SKELETON: 2903 curSkeleton=curElem->skeleton.getAlias(); 2904 s=curSkeleton->getSkeleton(); 2905 break; 2906 } 2907 if ( !isCanonicalItem(s) ) { 2908 LocalPointer<UnicodeString> newElem(s.clone(), status); 2909 if (U_FAILURE(status)) { 2910 return; 2911 } 2912 fSkeletons->addElement(newElem.getAlias(), status); 2913 if (U_FAILURE(status)) { 2914 fSkeletons.adoptInstead(nullptr); 2915 return; 2916 } 2917 newElem.orphan(); // fSkeletons vector now owns the UnicodeString (although it 2918 // does not use a deleter function to manage the ownership). 2919 } 2920 curElem = curElem->next.getAlias(); 2921 } 2922 } 2923 if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=nullptr) ) { 2924 status = U_BUFFER_OVERFLOW_ERROR; 2925 } 2926 } 2927 2928 const UnicodeString* 2929 DTSkeletonEnumeration::snext(UErrorCode& status) { 2930 if (U_SUCCESS(status) && fSkeletons.isValid() && pos < fSkeletons->size()) { 2931 return static_cast<const UnicodeString*>(fSkeletons->elementAt(pos++)); 2932 } 2933 return nullptr; 2934 } 2935 2936 void 2937 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) { 2938 pos=0; 2939 } 2940 2941 int32_t 2942 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const { 2943 return (fSkeletons.isNull()) ? 0 : fSkeletons->size(); 2944 } 2945 2946 UBool 2947 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) { 2948 if ( item.length() != 1 ) { 2949 return false; 2950 } 2951 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) { 2952 if (item.charAt(0)==Canonical_Items[i]) { 2953 return true; 2954 } 2955 } 2956 return false; 2957 } 2958 2959 DTSkeletonEnumeration::~DTSkeletonEnumeration() { 2960 UnicodeString *s; 2961 if (fSkeletons.isValid()) { 2962 for (int32_t i = 0; i < fSkeletons->size(); ++i) { 2963 if ((s = static_cast<UnicodeString*>(fSkeletons->elementAt(i))) != nullptr) { 2964 delete s; 2965 } 2966 } 2967 } 2968 } 2969 2970 DTRedundantEnumeration::DTRedundantEnumeration() : pos(0), fPatterns(nullptr) { 2971 } 2972 2973 void 2974 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) { 2975 if (U_FAILURE(status)) { return; } 2976 if (fPatterns.isNull()) { 2977 fPatterns.adoptInsteadAndCheckErrorCode(new UVector(status), status); 2978 if (U_FAILURE(status)) { 2979 return; 2980 } 2981 } 2982 LocalPointer<UnicodeString> newElem(new UnicodeString(pattern), status); 2983 if (U_FAILURE(status)) { 2984 return; 2985 } 2986 fPatterns->addElement(newElem.getAlias(), status); 2987 if (U_FAILURE(status)) { 2988 fPatterns.adoptInstead(nullptr); 2989 return; 2990 } 2991 newElem.orphan(); // fPatterns now owns the string, although a UVector 2992 // deleter function is not used to manage that ownership. 2993 } 2994 2995 const UnicodeString* 2996 DTRedundantEnumeration::snext(UErrorCode& status) { 2997 if (U_SUCCESS(status) && fPatterns.isValid() && pos < fPatterns->size()) { 2998 return static_cast<const UnicodeString*>(fPatterns->elementAt(pos++)); 2999 } 3000 return nullptr; 3001 } 3002 3003 void 3004 DTRedundantEnumeration::reset(UErrorCode& /*status*/) { 3005 pos=0; 3006 } 3007 3008 int32_t 3009 DTRedundantEnumeration::count(UErrorCode& /*status*/) const { 3010 return (fPatterns.isNull()) ? 0 : fPatterns->size(); 3011 } 3012 3013 UBool 3014 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) const { 3015 if ( item.length() != 1 ) { 3016 return false; 3017 } 3018 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) { 3019 if (item.charAt(0)==Canonical_Items[i]) { 3020 return true; 3021 } 3022 } 3023 return false; 3024 } 3025 3026 DTRedundantEnumeration::~DTRedundantEnumeration() { 3027 UnicodeString *s; 3028 if (fPatterns.isValid()) { 3029 for (int32_t i = 0; i < fPatterns->size(); ++i) { 3030 if ((s = static_cast<UnicodeString*>(fPatterns->elementAt(i))) != nullptr) { 3031 delete s; 3032 } 3033 } 3034 } 3035 } 3036 3037 U_NAMESPACE_END 3038 3039 3040 #endif /* #if !UCONFIG_NO_FORMATTING */ 3041 3042 //eof