icuSources/i18n/dtptngen.cpp

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 2007-2011, International Business Machines Corporation and
   4 * others. All Rights Reserved.
   5 *******************************************************************************
   6 *
   7 * File DTPTNGEN.CPP
   8 *
   9 *******************************************************************************
  10 */
  11
  12 #include "unicode/utypes.h"
  13 #if !UCONFIG_NO_FORMATTING
  14
  15 #include "unicode/datefmt.h"
  16 #include "unicode/decimfmt.h"
  17 #include "unicode/dtfmtsym.h"
  18 #include "unicode/dtptngen.h"
  19 #include "unicode/msgfmt.h"
  20 #include "unicode/smpdtfmt.h"
  21 #include "unicode/udat.h"
  22 #include "unicode/udatpg.h"
  23 #include "unicode/uniset.h"
  24 #include "unicode/uloc.h"
  25 #include "unicode/ures.h"
  26 #include "unicode/ustring.h"
  27 #include "unicode/rep.h"
  28 #include "cpputils.h"
  29 #include "ucln_in.h"
  30 #include "mutex.h"
  31 #include "cmemory.h"
  32 #include "cstring.h"
  33 #include "locbased.h"
  34 #include "gregoimp.h"
  35 #include "hash.h"
  36 #include "uresimp.h"
  37 #include "dtptngen_impl.h"
  38
  39 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
  40
  41 #if U_CHARSET_FAMILY==U_EBCDIC_FAMILY
  42 /**
  43  * If we are on EBCDIC, use an iterator which will
  44  * traverse the bundles in ASCII order.
  45  */
  46 #define U_USE_ASCII_BUNDLE_ITERATOR
  47 #define U_SORT_ASCII_BUNDLE_ITERATOR
  48 #endif
  49
  50 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
  51
  52 #include "unicode/ustring.h"
  53 #include "uarrsort.h"
  54
  55 struct UResAEntry {
  56     UChar *key;
  57     UResourceBundle *item;
  58 };
  59
  60 struct UResourceBundleAIterator {
  61     UResourceBundle  *bund;
  62     UResAEntry *entries;
  63     int32_t num;
  64     int32_t cursor;
  65 };
  66
  67 /* Must be C linkage to pass function pointer to the sort function */
  68
  69 #if !defined (OS390) && !defined (OS400)
  70 extern "C"
  71 #endif
  72 static int32_t U_CALLCONV
  73 ures_a_codepointSort(const void *context, const void *left, const void *right) {
  74     //CompareContext *cmp=(CompareContext *)context;
  75     return u_strcmp(((const UResAEntry *)left)->key,
  76                     ((const UResAEntry *)right)->key);
  77 }
  78
  79
  80 static void ures_a_open(UResourceBundleAIterator *aiter, UResourceBundle *bund, UErrorCode *status) {
  81     if(U_FAILURE(*status)) {
  82         return;
  83     }
  84     aiter->bund = bund;
  85     aiter->num = ures_getSize(aiter->bund);
  86     aiter->cursor = 0;
  87 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
  88     aiter->entries = NULL;
  89 #else
  90     aiter->entries = (UResAEntry*)uprv_malloc(sizeof(UResAEntry)*aiter->num);
  91     for(int i=0;i<aiter->num;i++) {
  92         aiter->entries[i].item = ures_getByIndex(aiter->bund, i, NULL, status);
  93         const char *akey = ures_getKey(aiter->entries[i].item);
  94         int32_t len = uprv_strlen(akey)+1;
  95         aiter->entries[i].key = (UChar*)uprv_malloc(len*sizeof(UChar));
  96         u_charsToUChars(akey, aiter->entries[i].key, len);
  97     }
  98     uprv_sortArray(aiter->entries, aiter->num, sizeof(UResAEntry), ures_a_codepointSort, NULL, TRUE, status);
  99 #endif
 100 }
 101
 102 static void ures_a_close(UResourceBundleAIterator *aiter) {
 103 #if defined(U_SORT_ASCII_BUNDLE_ITERATOR)
 104     for(int i=0;i<aiter->num;i++) {
 105         uprv_free(aiter->entries[i].key);
 106         ures_close(aiter->entries[i].item);
 107     }
 108 #endif
 109 }
 110
 111 static const UChar *ures_a_getNextString(UResourceBundleAIterator *aiter, int32_t *len, const char **key, UErrorCode *err) {
 112 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
 113     return ures_getNextString(aiter->bund, len, key, err);
 114 #else
 115     if(U_FAILURE(*err)) return NULL;
 116     UResourceBundle *item = aiter->entries[aiter->cursor].item;
 117     const UChar* ret = ures_getString(item, len, err);
 118     *key = ures_getKey(item);
 119     aiter->cursor++;
 120     return ret;
 121 #endif
 122 }
 123
 124
 125 #endif
 126
 127
 128 U_NAMESPACE_BEGIN
 129
 130
 131 // *****************************************************************************
 132 // class DateTimePatternGenerator
 133 // *****************************************************************************
 134 static const UChar Canonical_Items[] = {
 135     // GyQMwWEdDFHmsSv
 136     CAP_G, LOW_Y, CAP_Q, CAP_M, LOW_W, CAP_W, CAP_E, LOW_D, CAP_D, CAP_F,
 137     CAP_H, LOW_M, LOW_S, CAP_S, LOW_V, 0
 138 };
 139
 140 static const dtTypeElem dtTypes[] = {
 141     // patternChar, field, type, minLen, weight
 142     {CAP_G, UDATPG_ERA_FIELD, DT_SHORT, 1, 3,},
 143     {CAP_G, UDATPG_ERA_FIELD, DT_LONG, 4, 0},
 144     {LOW_Y, UDATPG_YEAR_FIELD, DT_NUMERIC, 1, 20},
 145     {CAP_Y, UDATPG_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 20},
 146     {LOW_U, UDATPG_YEAR_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 20},
 147     {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2},
 148     {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0},
 149     {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0},
 150     {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
 151     {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT + DT_DELTA, 3, 0},
 152     {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG + DT_DELTA, 4, 0},
 153     {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2},
 154     {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0},
 155     {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0},
 156     {CAP_M, UDATPG_MONTH_FIELD, DT_NARROW, 5, 0},
 157     {CAP_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
 158     {CAP_L, UDATPG_MONTH_FIELD, DT_SHORT - DT_DELTA, 3, 0},
 159     {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0},
 160     {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0},
 161     {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2},
 162     {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 0},
 163     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3},
 164     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0},
 165     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0},
 166     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2},
 167     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0},
 168     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
 169     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0},
 170     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical
 171     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0},
 172     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0},
 173     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0},
 174     {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2},
 175     {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 3},
 176     {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 0},
 177     {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20}, // really internal use, so we don't care
 178     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 0},
 179     {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour
 180     {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2},
 181     {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour
 182     {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
 183     {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2},
 184     {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2},
 185     {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000},
 186     {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 1000},
 187     {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0},
 188     {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
 189     {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3},
 190     {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0},
 191     {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3},
 192     {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
 193     {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3},
 194     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
 195     {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[]
 196  };
 197
 198 static const char* const CLDR_FIELD_APPEND[] = {
 199     "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week", "Day", "*", "*", "*",
 200     "Hour", "Minute", "Second", "*", "Timezone"
 201 };
 202
 203 static const char* const CLDR_FIELD_NAME[] = {
 204     "era", "year", "quarter", "month", "week", "*", "weekday", "day", "*", "*", "dayperiod",
 205     "hour", "minute", "second", "*", "zone"
 206 };
 207
 208 static const char* const Resource_Fields[] = {
 209     "day", "dayperiod", "era", "hour", "minute", "month", "second", "week",
 210     "weekday", "year", "zone", "quarter" };
 211
 212 // For appendItems
 213 static const UChar UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A,
 214     0x20, 0x7B, 0x31, 0x7D, 0x2524, 0};  // {0} \u251C{2}: {1}\u2524
 215
 216 static const UChar repeatedPatterns[6]={CAP_G, CAP_E, LOW_Z, LOW_V, CAP_Q, 0}; // "GEzvQ"
 217
 218 static const char DT_DateTimePatternsTag[]="DateTimePatterns";
 219 static const char DT_DateTimeCalendarTag[]="calendar";
 220 static const char DT_DateTimeGregorianTag[]="gregorian";
 221 static const char DT_DateTimeAppendItemsTag[]="appendItems";
 222 static const char DT_DateTimeFieldsTag[]="fields";
 223 static const char DT_DateTimeAvailableFormatsTag[]="availableFormats";
 224 //static const UnicodeString repeatedPattern=UnicodeString(repeatedPatterns);
 225
 226 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateTimePatternGenerator)
 227 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)
 228 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTRedundantEnumeration)
 229
 230 DateTimePatternGenerator*  U_EXPORT2
 231 DateTimePatternGenerator::createInstance(UErrorCode& status) {
 232     return createInstance(Locale::getDefault(), status);
 233 }
 234
 235 DateTimePatternGenerator* U_EXPORT2
 236 DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) {
 237     DateTimePatternGenerator *result = new DateTimePatternGenerator(locale, status);
 238     if (result == NULL) {
 239         status = U_MEMORY_ALLOCATION_ERROR;
 240     }
 241     if (U_FAILURE(status)) {
 242         delete result;
 243         result = NULL;
 244     }
 245     return result;
 246 }
 247
 248 DateTimePatternGenerator*  U_EXPORT2
 249 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) {
 250     DateTimePatternGenerator *result = new DateTimePatternGenerator(status);
 251     if (result == NULL) {
 252         status = U_MEMORY_ALLOCATION_ERROR;
 253     }
 254     if (U_FAILURE(status)) {
 255         delete result;
 256         result = NULL;
 257     }
 258     return result;
 259 }
 260
 261 DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) :
 262     skipMatcher(NULL),
 263     fAvailableFormatKeyHash(NULL)
 264 {
 265     fp = new FormatParser();
 266     dtMatcher = new DateTimeMatcher();
 267     distanceInfo = new DistanceInfo();
 268     patternMap = new PatternMap();
 269     if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) {
 270         status = U_MEMORY_ALLOCATION_ERROR;
 271     }
 272 }
 273
 274 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status) :
 275     skipMatcher(NULL),
 276     fAvailableFormatKeyHash(NULL)
 277 {
 278     fp = new FormatParser();
 279     dtMatcher = new DateTimeMatcher();
 280     distanceInfo = new DistanceInfo();
 281     patternMap = new PatternMap();
 282     if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) {
 283         status = U_MEMORY_ALLOCATION_ERROR;
 284     }
 285     else {
 286         initData(locale, status);
 287     }
 288 }
 289
 290 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) :
 291     UObject(),
 292     skipMatcher(NULL),
 293     fAvailableFormatKeyHash(NULL)
 294 {
 295     fp = new FormatParser();
 296     dtMatcher = new DateTimeMatcher();
 297     distanceInfo = new DistanceInfo();
 298     patternMap = new PatternMap();
 299     *this=other;
 300 }
 301
 302 DateTimePatternGenerator&
 303 DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) {
 304     pLocale = other.pLocale;
 305     fDefaultHourFormatChar = other.fDefaultHourFormatChar;
 306     *fp = *(other.fp);
 307     dtMatcher->copyFrom(other.dtMatcher->skeleton);
 308     *distanceInfo = *(other.distanceInfo);
 309     dateTimeFormat = other.dateTimeFormat;
 310     decimal = other.decimal;
 311     // NUL-terminate for the C API.
 312     dateTimeFormat.getTerminatedBuffer();
 313     decimal.getTerminatedBuffer();
 314     delete skipMatcher;
 315     if ( other.skipMatcher == NULL ) {
 316         skipMatcher = NULL;
 317     }
 318     else {
 319         skipMatcher = new DateTimeMatcher(*other.skipMatcher);
 320     }
 321     for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) {
 322         appendItemFormats[i] = other.appendItemFormats[i];
 323         appendItemNames[i] = other.appendItemNames[i];
 324         // NUL-terminate for the C API.
 325         appendItemFormats[i].getTerminatedBuffer();
 326         appendItemNames[i].getTerminatedBuffer();
 327     }
 328     UErrorCode status = U_ZERO_ERROR;
 329     patternMap->copyFrom(*other.patternMap, status);
 330     copyHashtable(other.fAvailableFormatKeyHash, status);
 331     return *this;
 332 }
 333
 334
 335 UBool
 336 DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) const {
 337     if (this == &other) {
 338         return TRUE;
 339     }
 340     if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) &&
 341         (dateTimeFormat==other.dateTimeFormat) && (decimal==other.decimal)) {
 342         for ( int32_t i=0 ; i<UDATPG_FIELD_COUNT; ++i ) {
 343            if ((appendItemFormats[i] != other.appendItemFormats[i]) ||
 344                (appendItemNames[i] != other.appendItemNames[i]) ) {
 345                return FALSE;
 346            }
 347         }
 348         return TRUE;
 349     }
 350     else {
 351         return FALSE;
 352     }
 353 }
 354
 355 UBool
 356 DateTimePatternGenerator::operator!=(const DateTimePatternGenerator& other) const {
 357     return  !operator==(other);
 358 }
 359
 360 DateTimePatternGenerator::~DateTimePatternGenerator() {
 361     if (fAvailableFormatKeyHash!=NULL) {
 362         delete fAvailableFormatKeyHash;
 363     }
 364
 365     if (fp != NULL) delete fp;
 366     if (dtMatcher != NULL) delete dtMatcher;
 367     if (distanceInfo != NULL) delete distanceInfo;
 368     if (patternMap != NULL) delete patternMap;
 369     if (skipMatcher != NULL) delete skipMatcher;
 370 }
 371
 372 void
 373 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status) {
 374     //const char *baseLangName = locale.getBaseName(); // unused
 375
 376     skipMatcher = NULL;
 377     fAvailableFormatKeyHash=NULL;
 378     addCanonicalItems();
 379     addICUPatterns(locale, status);
 380     if (U_FAILURE(status)) {
 381         return;
 382     }
 383     addCLDRData(locale, status);
 384     setDateTimeFromCalendar(locale, status);
 385     setDecimalSymbols(locale, status);
 386 } // DateTimePatternGenerator::initData
 387
 388 UnicodeString
 389 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode&
 390 /*status*/) {
 391     dtMatcher->set(pattern, fp);
 392     return dtMatcher->getSkeletonPtr()->getSkeleton();
 393 }
 394
 395 UnicodeString
 396 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) {
 397     dtMatcher->set(pattern, fp);
 398     return dtMatcher->getSkeletonPtr()->getBaseSkeleton();
 399 }
 400
 401 void
 402 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) {
 403     UnicodeString dfPattern;
 404     UnicodeString conflictingString;
 405     UDateTimePatternConflict conflictingStatus;
 406     DateFormat* df;
 407
 408     if (U_FAILURE(status)) {
 409         return;
 410     }
 411
 412     // Load with ICU patterns
 413     for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) {
 414         DateFormat::EStyle style = (DateFormat::EStyle)i;
 415         df = DateFormat::createDateInstance(style, locale);
 416         SimpleDateFormat* sdf;
 417         if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
 418             conflictingStatus = addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
 419         }
 420         // TODO Maybe we should return an error when the date format isn't simple.
 421         delete df;
 422         if (U_FAILURE(status)) {
 423             return;
 424         }
 425
 426         df = DateFormat::createTimeInstance(style, locale);
 427         if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
 428             conflictingStatus = addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
 429             // HACK for hh:ss
 430             if ( i==DateFormat::kMedium ) {
 431                 hackPattern = dfPattern;
 432             }
 433         }
 434         // TODO Maybe we should return an error when the date format isn't simple.
 435         delete df;
 436         if (U_FAILURE(status)) {
 437             return;
 438         }
 439     }
 440 }
 441
 442 void
 443 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status)  {
 444     UDateTimePatternConflict conflictingStatus;
 445     UnicodeString conflictingString;
 446
 447     fp->set(hackPattern);
 448     UnicodeString mmss;
 449     UBool gotMm=FALSE;
 450     for (int32_t i=0; i<fp->itemNumber; ++i) {
 451         UnicodeString field = fp->items[i];
 452         if ( fp->isQuoteLiteral(field) ) {
 453             if ( gotMm ) {
 454                UnicodeString quoteLiteral;
 455                fp->getQuoteLiteral(quoteLiteral, &i);
 456                mmss += quoteLiteral;
 457             }
 458         }
 459         else {
 460             if (fp->isPatternSeparator(field) && gotMm) {
 461                 mmss+=field;
 462             }
 463             else {
 464                 UChar ch=field.charAt(0);
 465                 if (ch==LOW_M) {
 466                     gotMm=TRUE;
 467                     mmss+=field;
 468                 }
 469                 else {
 470                     if (ch==LOW_S) {
 471                         if (!gotMm) {
 472                             break;
 473                         }
 474                         mmss+= field;
 475                         conflictingStatus = addPattern(mmss, FALSE, conflictingString, status);
 476                         break;
 477                     }
 478                     else {
 479                         if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) {
 480                             break;
 481                         }
 482                     }
 483                 }
 484             }
 485         }
 486     }
 487 }
 488
 489 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY)
 490
 491 static const UChar hourFormatChars[] = { CAP_H, LOW_H, CAP_K, LOW_K, 0 }; // HhKk, the hour format characters
 492
 493 void
 494 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& err) {
 495     UResourceBundle *rb, *calTypeBundle, *calBundle;
 496     UResourceBundle *patBundle, *fieldBundle, *fBundle;
 497     UnicodeString rbPattern, value, field;
 498     UnicodeString conflictingPattern;
 499     UDateTimePatternConflict conflictingStatus;
 500     const char *key=NULL;
 501     int32_t i;
 502
 503     UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, LENGTHOF(UDATPG_ItemFormat)-1);  // Read-only alias.
 504
 505     err = U_ZERO_ERROR;
 506
 507     fDefaultHourFormatChar = 0;
 508     for (i=0; i<UDATPG_FIELD_COUNT; ++i ) {
 509         appendItemNames[i]=CAP_F;
 510         if (i<10) {
 511             appendItemNames[i]+=(UChar)(i+0x30);
 512         }
 513         else {
 514             appendItemNames[i]+=(UChar)0x31;
 515             appendItemNames[i]+=(UChar)(i-10 + 0x30);
 516         }
 517         // NUL-terminate for the C API.
 518         appendItemNames[i].getTerminatedBuffer();
 519     }
 520
 521     rb = ures_open(NULL, locale.getName(), &err);
 522     if (rb == NULL || U_FAILURE(err)) {
 523         return;
 524     }
 525     const char *curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
 526     const char * calendarTypeToUse = DT_DateTimeGregorianTag; // initial default
 527     char         calendarType[ULOC_KEYWORDS_CAPACITY]; // to be filled in with the type to use, if all goes well
 528     if ( U_SUCCESS(err) ) {
 529         char    localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
 530         // obtain a locale that always has the calendar key value that should be used
 531         (void)ures_getFunctionalEquivalent(localeWithCalendarKey, ULOC_LOCALE_IDENTIFIER_CAPACITY, NULL,
 532                                             "calendar", "calendar", locale.getName(), NULL, FALSE, &err);
 533         localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination
 534         // now get the calendar key value from that locale
 535         int32_t calendarTypeLen = uloc_getKeywordValue(localeWithCalendarKey, "calendar", calendarType, ULOC_KEYWORDS_CAPACITY, &err);
 536         if (U_SUCCESS(err) && calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
 537             calendarTypeToUse = calendarType;
 538         }
 539         err = U_ZERO_ERROR;
 540     }
 541     calBundle = ures_getByKeyWithFallback(rb, DT_DateTimeCalendarTag, NULL, &err);
 542     calTypeBundle = ures_getByKeyWithFallback(calBundle, calendarTypeToUse, NULL, &err);
 543
 544     key=NULL;
 545     int32_t dtCount=0;
 546     patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimePatternsTag, NULL, &err);
 547     while (U_SUCCESS(err)) {
 548         rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
 549         dtCount++;
 550         if (rbPattern.length()==0 ) {
 551             break;  // no more pattern
 552         }
 553         else {
 554             if (dtCount==9) {
 555                 setDateTimeFormat(rbPattern);
 556             } else if (dtCount==4) { // short time format
 557                 // set fDefaultHourFormatChar to the hour format character from this pattern
 558                 int32_t tfIdx, tfLen = rbPattern.length();
 559                 UBool ignoreChars = FALSE;
 560                 for (tfIdx = 0; tfIdx < tfLen; tfIdx++) {
 561                     UChar tfChar = rbPattern.charAt(tfIdx);
 562                     if ( tfChar == SINGLE_QUOTE ) {
 563                         ignoreChars = !ignoreChars; // toggle (handle quoted literals & '' for single quote)
 564                     } else if ( !ignoreChars && u_strchr(hourFormatChars, tfChar) != NULL ) {
 565                         fDefaultHourFormatChar = tfChar;
 566                         break;
 567                     }
 568                 }
 569             }
 570         }
 571     }
 572     ures_close(patBundle);
 573
 574     err = U_ZERO_ERROR;
 575     patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAppendItemsTag, NULL, &err);
 576     key=NULL;
 577     UnicodeString itemKey;
 578     while (U_SUCCESS(err)) {
 579         rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
 580         if (rbPattern.length()==0 ) {
 581             break;  // no more pattern
 582         }
 583         else {
 584             setAppendItemFormat(getAppendFormatNumber(key), rbPattern);
 585         }
 586     }
 587     ures_close(patBundle);
 588
 589     key=NULL;
 590     err = U_ZERO_ERROR;
 591     fBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeFieldsTag, NULL, &err);
 592     for (i=0; i<MAX_RESOURCE_FIELD; ++i) {
 593         err = U_ZERO_ERROR;
 594         patBundle = ures_getByKeyWithFallback(fBundle, Resource_Fields[i], NULL, &err);
 595         fieldBundle = ures_getByKeyWithFallback(patBundle, "dn", NULL, &err);
 596         rbPattern = ures_getNextUnicodeString(fieldBundle, &key, &err);
 597         ures_close(fieldBundle);
 598         ures_close(patBundle);
 599         if (rbPattern.length()==0 ) {
 600             continue;
 601         }
 602         else {
 603             setAppendItemName(getAppendNameNumber(Resource_Fields[i]), rbPattern);
 604         }
 605     }
 606     ures_close(fBundle);
 607
 608     // add available formats
 609     err = U_ZERO_ERROR;
 610     initHashtable(err);
 611     patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAvailableFormatsTag, NULL, &err);
 612     if (U_SUCCESS(err)) {
 613         int32_t numberKeys = ures_getSize(patBundle);
 614         int32_t len;
 615         const UChar *retPattern;
 616         key=NULL;
 617 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 618         UResourceBundleAIterator aiter;
 619         ures_a_open(&aiter, patBundle, &err);
 620 #endif
 621         for(i=0; i<numberKeys; ++i) {
 622 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 623             retPattern=ures_a_getNextString(&aiter, &len, &key, &err);
 624 #else
 625             retPattern=ures_getNextString(patBundle, &len, &key, &err);
 626 #endif
 627             UnicodeString format=UnicodeString(retPattern);
 628             UnicodeString retKey=UnicodeString(key, -1, US_INV);
 629             setAvailableFormat(retKey, err);
 630             // Add pattern with its associated skeleton. Override any duplicate derived from std patterns,
 631             // but not a previous availableFormats entry:
 632             conflictingStatus = addPatternWithSkeleton(format, &retKey, TRUE, conflictingPattern, err);
 633         }
 634 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 635         ures_a_close(&aiter);
 636 #endif
 637     }
 638     ures_close(patBundle);
 639     ures_close(calTypeBundle);
 640     ures_close(calBundle);
 641     ures_close(rb);
 642
 643     err = U_ZERO_ERROR;
 644     char parentLocale[50];
 645     int32_t localeNameLen=0;
 646     uprv_strcpy(parentLocale, curLocaleName);
 647     while((localeNameLen=uloc_getParent(parentLocale, parentLocale, 50, &err))>=0 ) {
 648         rb = ures_open(NULL, parentLocale, &err);
 649         curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
 650         uprv_strcpy(parentLocale, curLocaleName);
 651         calBundle = ures_getByKey(rb, DT_DateTimeCalendarTag, NULL, &err);
 652         calTypeBundle = ures_getByKey(calBundle, calendarTypeToUse, NULL, &err);
 653         patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAvailableFormatsTag, NULL, &err);
 654         if (U_SUCCESS(err)) {
 655             int32_t numberKeys = ures_getSize(patBundle);
 656             int32_t len;
 657             const UChar *retPattern;
 658             key=NULL;
 659 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 660             UResourceBundleAIterator aiter;
 661             ures_a_open(&aiter, patBundle, &err);
 662 #endif
 663             for(i=0; i<numberKeys; ++i) {
 664 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 665                 retPattern=ures_a_getNextString(&aiter, &len, &key, &err);
 666 #else
 667                 retPattern=ures_getNextString(patBundle, &len, &key, &err);
 668 #endif
 669                 UnicodeString format=UnicodeString(retPattern);
 670                 UnicodeString retKey=UnicodeString(key, -1, US_INV);
 671                 if ( !isAvailableFormatSet(retKey) ) {
 672                     setAvailableFormat(retKey, err);
 673                     // Add pattern with its associated skeleton. Override any duplicate derived from std patterns,
 674                     // but not a previous availableFormats entry:
 675                     conflictingStatus = addPatternWithSkeleton(format, &retKey, TRUE, conflictingPattern, err);
 676                 }
 677             }
 678 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 679             ures_a_close(&aiter);
 680 #endif
 681         }
 682         err = U_ZERO_ERROR; // reset; if this locale lacks the necessary data, need to keep checking up to root.
 683         ures_close(patBundle);
 684         ures_close(calTypeBundle);
 685         ures_close(calBundle);
 686         ures_close(rb);
 687         if (localeNameLen==0) {
 688             break;
 689         }
 690     }
 691
 692     if (hackPattern.length()>0) {
 693         hackTimes(hackPattern, err);
 694     }
 695 }
 696
 697 void
 698 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
 699     if (fAvailableFormatKeyHash!=NULL) {
 700         return;
 701     }
 702     if ((fAvailableFormatKeyHash = new Hashtable(FALSE, err))==NULL) {
 703         err=U_MEMORY_ALLOCATION_ERROR;
 704         return;
 705     }
 706 }
 707
 708
 709 void
 710 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
 711     appendItemFormats[field] = value;
 712     // NUL-terminate for the C API.
 713     appendItemFormats[field].getTerminatedBuffer();
 714 }
 715
 716 const UnicodeString&
 717 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
 718     return appendItemFormats[field];
 719 }
 720
 721 void
 722 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
 723     appendItemNames[field] = value;
 724     // NUL-terminate for the C API.
 725     appendItemNames[field].getTerminatedBuffer();
 726 }
 727
 728 const UnicodeString&
 729 DateTimePatternGenerator:: getAppendItemName(UDateTimePatternField field) const {
 730     return appendItemNames[field];
 731 }
 732
 733 void
 734 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
 735     value = SINGLE_QUOTE;
 736     value += appendItemNames[field];
 737     value += SINGLE_QUOTE;
 738 }
 739
 740 UnicodeString
 741 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
 742     return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
 743 }
 744
 745 UnicodeString
 746 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
 747     const UnicodeString *bestPattern=NULL;
 748     UnicodeString dtFormat;
 749     UnicodeString resultPattern;
 750
 751     int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
 752     int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
 753
 754     UnicodeString patternFormCopy = UnicodeString(patternForm);
 755     patternFormCopy.findAndReplace(UnicodeString(LOW_J), UnicodeString(fDefaultHourFormatChar));
 756
 757     resultPattern.remove();
 758     dtMatcher->set(patternFormCopy, fp);
 759     const PtnSkeleton* specifiedSkeleton=NULL;
 760     bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, &specifiedSkeleton);
 761     if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
 762         resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, FALSE, options);
 763
 764         return resultPattern;
 765     }
 766     int32_t neededFields = dtMatcher->getFieldMask();
 767     UnicodeString datePattern=getBestAppending(neededFields & dateMask, options);
 768     UnicodeString timePattern=getBestAppending(neededFields & timeMask, options);
 769     if (datePattern.length()==0) {
 770         if (timePattern.length()==0) {
 771             resultPattern.remove();
 772         }
 773         else {
 774             return timePattern;
 775         }
 776     }
 777     if (timePattern.length()==0) {
 778         return datePattern;
 779     }
 780     resultPattern.remove();
 781     status = U_ZERO_ERROR;
 782     dtFormat=getDateTimeFormat();
 783     Formattable dateTimeObject[] = { timePattern, datePattern };
 784     resultPattern = MessageFormat::format(dtFormat, dateTimeObject, 2, resultPattern, status );
 785     return resultPattern;
 786 }
 787
 788 UnicodeString
 789 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
 790                                             const UnicodeString& skeleton,
 791                                             UErrorCode& status) {
 792     return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
 793 }
 794
 795 UnicodeString
 796 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
 797                                             const UnicodeString& skeleton,
 798                                             UDateTimePatternMatchOptions options,
 799                                             UErrorCode& /*status*/) {
 800     dtMatcher->set(skeleton, fp);
 801     UnicodeString result = adjustFieldTypes(pattern, NULL, FALSE, options);
 802     return result;
 803 }
 804
 805 void
 806 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
 807     this->decimal = newDecimal;
 808     // NUL-terminate for the C API.
 809     this->decimal.getTerminatedBuffer();
 810 }
 811
 812 const UnicodeString&
 813 DateTimePatternGenerator::getDecimal() const {
 814     return decimal;
 815 }
 816
 817 void
 818 DateTimePatternGenerator::addCanonicalItems() {
 819     UnicodeString  conflictingPattern;
 820     UDateTimePatternConflict conflictingStatus;
 821     UErrorCode status = U_ZERO_ERROR;
 822
 823     for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
 824         conflictingStatus = addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
 825     }
 826 }
 827
 828 void
 829 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
 830     dateTimeFormat = dtFormat;
 831     // NUL-terminate for the C API.
 832     dateTimeFormat.getTerminatedBuffer();
 833 }
 834
 835 const UnicodeString&
 836 DateTimePatternGenerator::getDateTimeFormat() const {
 837     return dateTimeFormat;
 838 }
 839
 840 void
 841 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
 842     const UChar *resStr;
 843     int32_t resStrLen = 0;
 844
 845     Calendar* fCalendar = Calendar::createInstance(locale, status);
 846     CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
 847     UResourceBundle *dateTimePatterns = calData.getByKey(DT_DateTimePatternsTag, status);
 848     if (U_FAILURE(status)) return;
 849
 850     if (ures_getSize(dateTimePatterns) <= DateFormat::kDateTime)
 851     {
 852         status = U_INVALID_FORMAT_ERROR;
 853         return;
 854     }
 855     resStr = ures_getStringByIndex(dateTimePatterns, (int32_t)DateFormat::kDateTime, &resStrLen, &status);
 856     setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
 857
 858     delete fCalendar;
 859 }
 860
 861 void
 862 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
 863     DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
 864     if(U_SUCCESS(status)) {
 865         decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
 866         // NUL-terminate for the C API.
 867         decimal.getTerminatedBuffer();
 868     }
 869 }
 870
 871 UDateTimePatternConflict
 872 DateTimePatternGenerator::addPattern(
 873     const UnicodeString& pattern,
 874     UBool override,
 875     UnicodeString &conflictingPattern,
 876     UErrorCode& status)
 877 {
 878     return addPatternWithSkeleton(pattern, NULL, override, conflictingPattern, status);
 879 }
 880
 881 // For DateTimePatternGenerator::addPatternWithSkeleton -
 882 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
 883 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
 884 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
 885 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
 886 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
 887 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
 888 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
 889 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
 890 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
 891 UDateTimePatternConflict
 892 DateTimePatternGenerator::addPatternWithSkeleton(
 893     const UnicodeString& pattern,
 894     const UnicodeString* skeletonToUse,
 895     UBool override,
 896     UnicodeString& conflictingPattern,
 897     UErrorCode& status)
 898 {
 899
 900     UnicodeString basePattern;
 901     PtnSkeleton   skeleton;
 902     UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
 903
 904     DateTimeMatcher matcher;
 905     if ( skeletonToUse == NULL ) {
 906         matcher.set(pattern, fp, skeleton);
 907         matcher.getBasePattern(basePattern);
 908     } else {
 909         matcher.set(*skeletonToUse, fp, skeleton); // this still trims skeleton fields to max len 3, may need to change it.
 910         matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
 911     }
 912     UBool entryHadSpecifiedSkeleton;
 913     const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
 914     if (duplicatePattern != NULL ) {
 915         conflictingStatus = UDATPG_BASE_CONFLICT;
 916         conflictingPattern = *duplicatePattern;
 917         if (!override || (skeletonToUse != NULL && entryHadSpecifiedSkeleton)) {
 918             return conflictingStatus;
 919         }
 920     }
 921     const PtnSkeleton* entrySpecifiedSkeleton = NULL;
 922     duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
 923     if (duplicatePattern != NULL ) {
 924         conflictingStatus = UDATPG_CONFLICT;
 925         conflictingPattern = *duplicatePattern;
 926         if (!override || (skeletonToUse != NULL && entrySpecifiedSkeleton != NULL)) {
 927             return conflictingStatus;
 928         }
 929     }
 930     patternMap->add(basePattern, skeleton, pattern, skeletonToUse != NULL, status);
 931     if(U_FAILURE(status)) {
 932         return conflictingStatus;
 933     }
 934
 935     return UDATPG_NO_CONFLICT;
 936 }
 937
 938
 939 UDateTimePatternField
 940 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
 941     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
 942         if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
 943             return (UDateTimePatternField)i;
 944         }
 945     }
 946     return UDATPG_FIELD_COUNT;
 947 }
 948
 949 UDateTimePatternField
 950 DateTimePatternGenerator::getAppendNameNumber(const char* field) const {
 951     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
 952         if (uprv_strcmp(CLDR_FIELD_NAME[i],field)==0) {
 953             return (UDateTimePatternField)i;
 954         }
 955     }
 956     return UDATPG_FIELD_COUNT;
 957 }
 958
 959 const UnicodeString*
 960 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
 961                                      int32_t includeMask,
 962                                      DistanceInfo* missingFields,
 963                                      const PtnSkeleton** specifiedSkeletonPtr) {
 964     int32_t bestDistance = 0x7fffffff;
 965     DistanceInfo tempInfo;
 966     const UnicodeString *bestPattern=NULL;
 967     const PtnSkeleton* specifiedSkeleton=NULL;
 968
 969     PatternMapIterator it;
 970     for (it.set(*patternMap); it.hasNext(); ) {
 971         DateTimeMatcher trial = it.next();
 972         if (trial.equals(skipMatcher)) {
 973             continue;
 974         }
 975         int32_t distance=source.getDistance(trial, includeMask, tempInfo);
 976         if (distance<bestDistance) {
 977             bestDistance=distance;
 978             bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
 979             missingFields->setTo(tempInfo);
 980             if (distance==0) {
 981                 break;
 982             }
 983         }
 984     }
 985
 986     // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
 987     // then return it too. This generally happens when the caller needs to pass that skeleton
 988     // through to adjustFieldTypes so the latter can do a better job.
 989     if (bestPattern && specifiedSkeletonPtr) {
 990         *specifiedSkeletonPtr = specifiedSkeleton;
 991     }
 992     return bestPattern;
 993 }
 994
 995 UnicodeString
 996 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
 997                                            const PtnSkeleton* specifiedSkeleton,
 998                                            UBool fixFractionalSeconds,
 999                                            UDateTimePatternMatchOptions options) {
1000     UnicodeString newPattern;
1001     fp->set(pattern);
1002     for (int32_t i=0; i < fp->itemNumber; i++) {
1003         UnicodeString field = fp->items[i];
1004         if ( fp->isQuoteLiteral(field) ) {
1005
1006             UnicodeString quoteLiteral;
1007             fp->getQuoteLiteral(quoteLiteral, &i);
1008             newPattern += quoteLiteral;
1009         }
1010         else {
1011             if (fp->isPatternSeparator(field)) {
1012                 newPattern+=field;
1013                 continue;
1014             }
1015             int32_t canonicalIndex = fp->getCanonicalIndex(field);
1016             if (canonicalIndex < 0) {
1017                 newPattern+=field;
1018                 continue;  // don't adjust
1019             }
1020             const dtTypeElem *row = &dtTypes[canonicalIndex];
1021             int32_t typeValue = row->field;
1022             if (fixFractionalSeconds && typeValue == UDATPG_SECOND_FIELD) {
1023                 UnicodeString newField=dtMatcher->skeleton.original[UDATPG_FRACTIONAL_SECOND_FIELD];
1024                 field = field + decimal + newField;
1025             } else if (dtMatcher->skeleton.type[typeValue]!=0) {
1026                     // Here:
1027                     // - "reqField" is the field from the originally requested skeleton, with length
1028                     // "reqFieldLen".
1029                     // - "field" is the field from the found pattern.
1030                     //
1031                     // The adjusted field should consist of characters from the originally requested
1032                     // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1033                     // UDATPG_WEEKDAY_FIELD, in which case it should consist of characters from the
1034                     // found pattern.
1035                     //
1036                     // The length of the adjusted field (adjFieldLen) should match that in the originally
1037                     // requested skeleton, except that in the following cases the length of the adjusted field
1038                     // should match that in the found pattern (i.e. the length of this pattern field should
1039                     // not be adjusted):
1040                     // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1041                     //    not set (ticket #7180). Note, we may want to implement a similar change for other
1042                     //    numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1043                     //    field length, but options bits can be used to override this.
1044                     // 2. There is a specified skeleton for the found pattern and one of the following is true:
1045                     //    a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1046                     //    b) The pattern field is numeric and the skeleton field is not, or vice versa.
1047
1048                     UnicodeString reqField = dtMatcher->skeleton.original[typeValue];
1049                     int32_t reqFieldLen = reqField.length();
1050                     if (reqField.charAt(0) == CAP_E && reqFieldLen < 3)
1051                         reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1052                     int32_t adjFieldLen = reqFieldLen;
1053                     if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1054                          (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1055                          (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1056                          adjFieldLen = field.length();
1057                     } else if (specifiedSkeleton) {
1058                         UnicodeString skelField = specifiedSkeleton->original[typeValue];
1059                         int32_t skelFieldLen = skelField.length();
1060                         UBool patFieldIsNumeric = (row->type > 0);
1061                         UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1062                         if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1063                             // don't adjust the field length in the found pattern
1064                             adjFieldLen = field.length();
1065                         }
1066                     }
1067                     UChar c = (typeValue!= UDATPG_HOUR_FIELD && typeValue!= UDATPG_MONTH_FIELD && typeValue!= UDATPG_WEEKDAY_FIELD)?
1068                         reqField.charAt(0): field.charAt(0);
1069                     field.remove();
1070                     for (int32_t i=adjFieldLen; i>0; --i) {
1071                         field+=c;
1072                     }
1073             }
1074             newPattern+=field;
1075         }
1076     }
1077     return newPattern;
1078 }
1079
1080 UnicodeString
1081 DateTimePatternGenerator::getBestAppending(int32_t missingFields, UDateTimePatternMatchOptions options) {
1082     UnicodeString  resultPattern, tempPattern;
1083     UErrorCode err=U_ZERO_ERROR;
1084     int32_t lastMissingFieldMask=0;
1085     if (missingFields!=0) {
1086         resultPattern=UnicodeString();
1087         const PtnSkeleton* specifiedSkeleton=NULL;
1088         tempPattern = *getBestRaw(*dtMatcher, missingFields, distanceInfo, &specifiedSkeleton);
1089         resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options);
1090         if ( distanceInfo->missingFieldMask==0 ) {
1091             return resultPattern;
1092         }
1093         while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1094             if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1095                 break;  // cannot find the proper missing field
1096             }
1097             if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1098                 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1099                 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, TRUE, options);
1100                 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1101                 continue;
1102             }
1103             int32_t startingMask = distanceInfo->missingFieldMask;
1104             tempPattern = *getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, &specifiedSkeleton);
1105             tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options);
1106             int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1107             int32_t topField=getTopBitNumber(foundMask);
1108             UnicodeString appendName;
1109             getAppendName((UDateTimePatternField)topField, appendName);
1110             const Formattable formatPattern[] = {
1111                 resultPattern,
1112                 tempPattern,
1113                 appendName
1114             };
1115             UnicodeString emptyStr;
1116             resultPattern = MessageFormat::format(appendItemFormats[topField], formatPattern, 3, emptyStr, err);
1117             lastMissingFieldMask = distanceInfo->missingFieldMask;
1118         }
1119     }
1120     return resultPattern;
1121 }
1122
1123 int32_t
1124 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) {
1125     if ( foundMask==0 ) {
1126         return 0;
1127     }
1128     int32_t i=0;
1129     while (foundMask!=0) {
1130         foundMask >>=1;
1131         ++i;
1132     }
1133     if (i-1 >UDATPG_ZONE_FIELD) {
1134         return UDATPG_ZONE_FIELD;
1135     }
1136     else
1137         return i-1;
1138 }
1139
1140 void
1141 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1142 {
1143     fAvailableFormatKeyHash->puti(key, 1, err);
1144 }
1145
1146 UBool
1147 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1148     return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1149 }
1150
1151 void
1152 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1153
1154     if (other == NULL) {
1155         return;
1156     }
1157     if (fAvailableFormatKeyHash != NULL) {
1158         delete fAvailableFormatKeyHash;
1159         fAvailableFormatKeyHash = NULL;
1160     }
1161     initHashtable(status);
1162     if(U_FAILURE(status)){
1163         return;
1164     }
1165     int32_t pos = -1;
1166     const UHashElement* elem = NULL;
1167     // walk through the hash table and create a deep clone
1168     while((elem = other->nextElement(pos))!= NULL){
1169         const UHashTok otherKeyTok = elem->key;
1170         UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1171         fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1172         if(U_FAILURE(status)){
1173             return;
1174         }
1175     }
1176 }
1177
1178 StringEnumeration*
1179 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1180     StringEnumeration* skeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status);
1181     return skeletonEnumerator;
1182 }
1183
1184 const UnicodeString&
1185 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1186     PtnElem *curElem;
1187
1188     if (skeleton.length() ==0) {
1189         return emptyString;
1190     }
1191     curElem = patternMap->getHeader(skeleton.charAt(0));
1192     while ( curElem != NULL ) {
1193         if ( curElem->skeleton->getSkeleton()==skeleton ) {
1194             return curElem->pattern;
1195         }
1196         curElem=curElem->next;
1197     }
1198     return emptyString;
1199 }
1200
1201 StringEnumeration*
1202 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1203     StringEnumeration* baseSkeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status);
1204     return baseSkeletonEnumerator;
1205 }
1206
1207 StringEnumeration*
1208 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1209     StringEnumeration* output = new DTRedundantEnumeration();
1210     const UnicodeString *pattern;
1211     PatternMapIterator it;
1212     for (it.set(*patternMap); it.hasNext(); ) {
1213         DateTimeMatcher current = it.next();
1214         pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1215         if ( isCanonicalItem(*pattern) ) {
1216             continue;
1217         }
1218         if ( skipMatcher == NULL ) {
1219             skipMatcher = new DateTimeMatcher(current);
1220         }
1221         else {
1222             *skipMatcher = current;
1223         }
1224         UnicodeString trial = getBestPattern(current.getPattern(), status);
1225         if (trial == *pattern) {
1226             ((DTRedundantEnumeration *)output)->add(*pattern, status);
1227         }
1228         if (current.equals(skipMatcher)) {
1229             continue;
1230         }
1231     }
1232     return output;
1233 }
1234
1235 UBool
1236 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1237     if ( item.length() != 1 ) {
1238         return FALSE;
1239     }
1240     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1241         if (item.charAt(0)==Canonical_Items[i]) {
1242             return TRUE;
1243         }
1244     }
1245     return FALSE;
1246 }
1247
1248
1249 DateTimePatternGenerator*
1250 DateTimePatternGenerator::clone() const {
1251     return new DateTimePatternGenerator(*this);
1252 }
1253
1254 PatternMap::PatternMap() {
1255    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1256       boot[i]=NULL;
1257    }
1258    isDupAllowed = TRUE;
1259 }
1260
1261 void
1262 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1263     this->isDupAllowed = other.isDupAllowed;
1264     for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1265         PtnElem *curElem, *otherElem, *prevElem=NULL;
1266         otherElem = other.boot[bootIndex];
1267         while (otherElem!=NULL) {
1268             if ((curElem = new PtnElem(otherElem->basePattern, otherElem->pattern))==NULL) {
1269                 // out of memory
1270                 status = U_MEMORY_ALLOCATION_ERROR;
1271                 return;
1272             }
1273             if ( this->boot[bootIndex]== NULL ) {
1274                 this->boot[bootIndex] = curElem;
1275             }
1276             if ((curElem->skeleton=new PtnSkeleton(*(otherElem->skeleton))) == NULL ) {
1277                 // out of memory
1278                 status = U_MEMORY_ALLOCATION_ERROR;
1279                 return;
1280             }
1281
1282             if (prevElem!=NULL) {
1283                 prevElem->next=curElem;
1284             }
1285             curElem->next=NULL;
1286             prevElem = curElem;
1287             otherElem = otherElem->next;
1288         }
1289
1290     }
1291 }
1292
1293 PtnElem*
1294 PatternMap::getHeader(UChar baseChar) {
1295     PtnElem* curElem;
1296
1297     if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1298          curElem = boot[baseChar-CAP_A];
1299     }
1300     else {
1301         if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1302             curElem = boot[26+baseChar-LOW_A];
1303         }
1304         else {
1305             return NULL;
1306         }
1307     }
1308     return curElem;
1309 }
1310
1311 PatternMap::~PatternMap() {
1312    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1313        if (boot[i]!=NULL ) {
1314            delete boot[i];
1315            boot[i]=NULL;
1316        }
1317    }
1318 }  // PatternMap destructor
1319
1320 void
1321 PatternMap::add(const UnicodeString& basePattern,
1322                 const PtnSkeleton& skeleton,
1323                 const UnicodeString& value,// mapped pattern value
1324                 UBool skeletonWasSpecified,
1325                 UErrorCode &status) {
1326     UChar baseChar = basePattern.charAt(0);
1327     PtnElem *curElem, *baseElem;
1328     status = U_ZERO_ERROR;
1329
1330     // the baseChar must be A-Z or a-z
1331     if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1332         baseElem = boot[baseChar-CAP_A];
1333     }
1334     else {
1335         if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1336             baseElem = boot[26+baseChar-LOW_A];
1337          }
1338          else {
1339              status = U_ILLEGAL_CHARACTER;
1340              return;
1341          }
1342     }
1343
1344     if (baseElem == NULL) {
1345         if ((curElem = new PtnElem(basePattern, value)) == NULL ) {
1346             // out of memory
1347             status = U_MEMORY_ALLOCATION_ERROR;
1348             return;
1349         }
1350         if (baseChar >= LOW_A) {
1351             boot[26 + (baseChar-LOW_A)] = curElem;
1352         }
1353         else {
1354             boot[baseChar-CAP_A] = curElem;
1355         }
1356         curElem->skeleton = new PtnSkeleton(skeleton);
1357         curElem->skeletonWasSpecified = skeletonWasSpecified;
1358     }
1359     if ( baseElem != NULL ) {
1360         curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1361
1362         if (curElem == NULL) {
1363             // add new element to the list.
1364             curElem = baseElem;
1365             while( curElem -> next != NULL )
1366             {
1367                 curElem = curElem->next;
1368             }
1369             if ((curElem->next = new PtnElem(basePattern, value)) == NULL ) {
1370                 // out of memory
1371                 status = U_MEMORY_ALLOCATION_ERROR;
1372                 return;
1373             }
1374             curElem=curElem->next;
1375             curElem->skeleton = new PtnSkeleton(skeleton);
1376             curElem->skeletonWasSpecified = skeletonWasSpecified;
1377         }
1378         else {
1379             // Pattern exists in the list already.
1380             if ( !isDupAllowed ) {
1381                 return;
1382             }
1383             // Overwrite the value.
1384             curElem->pattern = value;
1385         }
1386     }
1387 }  // PatternMap::add
1388
1389 // Find the pattern from the given basePattern string.
1390 const UnicodeString *
1391 PatternMap::getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeletonWasSpecified) { // key to search for
1392    PtnElem *curElem;
1393
1394    if ((curElem=getHeader(basePattern.charAt(0)))==NULL) {
1395        return NULL;  // no match
1396    }
1397
1398    do  {
1399        if ( basePattern.compare(curElem->basePattern)==0 ) {
1400           skeletonWasSpecified = curElem->skeletonWasSpecified;
1401           return &(curElem->pattern);
1402        }
1403        curElem=curElem->next;
1404    }while (curElem != NULL);
1405
1406    return NULL;
1407 }  // PatternMap::getFromBasePattern
1408
1409
1410 // Find the pattern from the given skeleton.
1411 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
1412 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
1413 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
1414 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
1415 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
1416 const UnicodeString *
1417 PatternMap::getPatternFromSkeleton(PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) { // key to search for
1418    PtnElem *curElem;
1419
1420    if (specifiedSkeletonPtr) {
1421        *specifiedSkeletonPtr = NULL;
1422    }
1423
1424    // find boot entry
1425    UChar baseChar='\0';
1426    for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1427        if (skeleton.baseOriginal[i].length() !=0 ) {
1428            baseChar = skeleton.baseOriginal[i].charAt(0);
1429            break;
1430        }
1431    }
1432
1433    if ((curElem=getHeader(baseChar))==NULL) {
1434        return NULL;  // no match
1435    }
1436
1437    do  {
1438        int32_t i=0;
1439        if (specifiedSkeletonPtr != NULL) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
1440            for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1441                if (curElem->skeleton->original[i].compare(skeleton.original[i]) != 0 )
1442                {
1443                    break;
1444                }
1445            }
1446        } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
1447            for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1448                if (curElem->skeleton->baseOriginal[i].compare(skeleton.baseOriginal[i]) != 0 )
1449                {
1450                    break;
1451                }
1452            }
1453        }
1454        if (i == UDATPG_FIELD_COUNT) {
1455            if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
1456                *specifiedSkeletonPtr = curElem->skeleton;
1457            }
1458            return &(curElem->pattern);
1459        }
1460        curElem=curElem->next;
1461    }while (curElem != NULL);
1462
1463    return NULL;
1464 }
1465
1466 UBool
1467 PatternMap::equals(const PatternMap& other) {
1468     if ( this==&other ) {
1469         return TRUE;
1470     }
1471     for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1472         if ( boot[bootIndex]==other.boot[bootIndex] ) {
1473             continue;
1474         }
1475         if ( (boot[bootIndex]==NULL)||(other.boot[bootIndex]==NULL) ) {
1476             return FALSE;
1477         }
1478         PtnElem *otherElem = other.boot[bootIndex];
1479         PtnElem *myElem = boot[bootIndex];
1480         while ((otherElem!=NULL) || (myElem!=NULL)) {
1481             if ( myElem == otherElem ) {
1482                 break;
1483             }
1484             if ((otherElem==NULL) || (myElem==NULL)) {
1485                 return FALSE;
1486             }
1487             if ( (myElem->basePattern != otherElem->basePattern) ||
1488                  (myElem->pattern != otherElem->pattern) ) {
1489                 return FALSE;
1490             }
1491             if ((myElem->skeleton!=otherElem->skeleton)&&
1492                 !myElem->skeleton->equals(*(otherElem->skeleton))) {
1493                 return FALSE;
1494             }
1495             myElem = myElem->next;
1496             otherElem=otherElem->next;
1497         }
1498     }
1499     return TRUE;
1500 }
1501
1502 // find any key existing in the mapping table already.
1503 // return TRUE if there is an existing key, otherwise return FALSE.
1504 PtnElem*
1505 PatternMap::getDuplicateElem(
1506             const UnicodeString &basePattern,
1507             const PtnSkeleton &skeleton,
1508             PtnElem *baseElem)  {
1509    PtnElem *curElem;
1510
1511    if ( baseElem == (PtnElem *)NULL )  {
1512          return (PtnElem*)NULL;
1513    }
1514    else {
1515          curElem = baseElem;
1516    }
1517    do {
1518      if ( basePattern.compare(curElem->basePattern)==0 ) {
1519         UBool isEqual=TRUE;
1520         for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1521             if (curElem->skeleton->type[i] != skeleton.type[i] ) {
1522                 isEqual=FALSE;
1523                 break;
1524             }
1525         }
1526         if (isEqual) {
1527             return curElem;
1528         }
1529      }
1530      curElem = curElem->next;
1531    } while( curElem != (PtnElem *)NULL );
1532
1533    // end of the list
1534    return (PtnElem*)NULL;
1535
1536 }  // PatternMap::getDuplicateElem
1537
1538 DateTimeMatcher::DateTimeMatcher(void) {
1539 }
1540
1541 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
1542     copyFrom(other.skeleton);
1543 }
1544
1545
1546 void
1547 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
1548     PtnSkeleton localSkeleton;
1549     return set(pattern, fp, localSkeleton);
1550 }
1551
1552 void
1553 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
1554     int32_t i;
1555     for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1556         skeletonResult.type[i]=NONE;
1557     }
1558     fp->set(pattern);
1559     for (i=0; i < fp->itemNumber; i++) {
1560         UnicodeString field = fp->items[i];
1561         if ( field.charAt(0) == LOW_A ) {
1562             continue;  // skip 'a'
1563         }
1564
1565         if ( fp->isQuoteLiteral(field) ) {
1566             UnicodeString quoteLiteral;
1567             fp->getQuoteLiteral(quoteLiteral, &i);
1568             continue;
1569         }
1570         int32_t canonicalIndex = fp->getCanonicalIndex(field);
1571         if (canonicalIndex < 0 ) {
1572             continue;
1573         }
1574         const dtTypeElem *row = &dtTypes[canonicalIndex];
1575         int32_t typeValue = row->field;
1576         skeletonResult.original[typeValue]=field;
1577         UChar repeatChar = row->patternChar;
1578         int32_t repeatCount = row->minLen > 3 ? 3: row->minLen;
1579         while (repeatCount-- > 0) {
1580             skeletonResult.baseOriginal[typeValue] += repeatChar;
1581         }
1582         int16_t subTypeValue = row->type;
1583         if ( row->type > 0) {
1584             subTypeValue += field.length();
1585         }
1586         skeletonResult.type[typeValue] = subTypeValue;
1587     }
1588     copyFrom(skeletonResult);
1589 }
1590
1591 void
1592 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
1593     result.remove(); // Reset the result first.
1594     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1595         if (skeleton.baseOriginal[i].length()!=0) {
1596             result += skeleton.baseOriginal[i];
1597         }
1598     }
1599 }
1600
1601 UnicodeString
1602 DateTimeMatcher::getPattern() {
1603     UnicodeString result;
1604
1605     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1606         if (skeleton.original[i].length()!=0) {
1607             result += skeleton.original[i];
1608         }
1609     }
1610     return result;
1611 }
1612
1613 int32_t
1614 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) {
1615     int32_t result=0;
1616     distanceInfo.clear();
1617     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1618         int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
1619         int32_t otherType = other.skeleton.type[i];
1620         if (myType==otherType) {
1621             continue;
1622         }
1623         if (myType==0) {// and other is not
1624             result += EXTRA_FIELD;
1625             distanceInfo.addExtra(i);
1626         }
1627         else {
1628             if (otherType==0) {
1629                 result += MISSING_FIELD;
1630                 distanceInfo.addMissing(i);
1631             }
1632             else {
1633                 result += abs(myType - otherType);
1634             }
1635         }
1636
1637     }
1638     return result;
1639 }
1640
1641 void
1642 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
1643     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1644         this->skeleton.type[i]=newSkeleton.type[i];
1645         this->skeleton.original[i]=newSkeleton.original[i];
1646         this->skeleton.baseOriginal[i]=newSkeleton.baseOriginal[i];
1647     }
1648 }
1649
1650 void
1651 DateTimeMatcher::copyFrom() {
1652     // same as clear
1653     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1654         this->skeleton.type[i]=0;
1655         this->skeleton.original[i].remove();
1656         this->skeleton.baseOriginal[i].remove();
1657     }
1658 }
1659
1660 UBool
1661 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
1662     if (other==NULL) {
1663         return FALSE;
1664     }
1665     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1666         if (this->skeleton.original[i]!=other->skeleton.original[i] ) {
1667             return FALSE;
1668         }
1669     }
1670     return TRUE;
1671 }
1672
1673 int32_t
1674 DateTimeMatcher::getFieldMask() {
1675     int32_t result=0;
1676
1677     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1678         if (skeleton.type[i]!=0) {
1679             result |= (1<<i);
1680         }
1681     }
1682     return result;
1683 }
1684
1685 PtnSkeleton*
1686 DateTimeMatcher::getSkeletonPtr() {
1687     return &skeleton;
1688 }
1689
1690 FormatParser::FormatParser () {
1691     status = START;
1692     itemNumber=0;
1693 }
1694
1695
1696 FormatParser::~FormatParser () {
1697 }
1698
1699
1700 // Find the next token with the starting position and length
1701 // Note: the startPos may
1702 FormatParser::TokenStatus
1703 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
1704     int32_t  curLoc = startPos;
1705     if ( curLoc >= pattern.length()) {
1706         return DONE;
1707     }
1708     // check the current char is between A-Z or a-z
1709     do {
1710         UChar c=pattern.charAt(curLoc);
1711         if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
1712            curLoc++;
1713         }
1714         else {
1715                startPos = curLoc;
1716                *len=1;
1717                return ADD_TOKEN;
1718         }
1719
1720         if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
1721             break;  // not the same token
1722         }
1723     } while(curLoc <= pattern.length());
1724     *len = curLoc-startPos;
1725     return ADD_TOKEN;
1726 }
1727
1728 void
1729 FormatParser::set(const UnicodeString& pattern) {
1730     int32_t startPos=0;
1731     TokenStatus result=START;
1732     int32_t len=0;
1733     itemNumber =0;
1734
1735     do {
1736         result = setTokens( pattern, startPos, &len );
1737         if ( result == ADD_TOKEN )
1738         {
1739             items[itemNumber++] = UnicodeString(pattern, startPos, len );
1740             startPos += len;
1741         }
1742         else {
1743             break;
1744         }
1745     } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
1746 }
1747
1748 int32_t
1749 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
1750     int32_t len = s.length();
1751     if (len == 0) {
1752         return -1;
1753     }
1754     UChar ch = s.charAt(0);
1755
1756     // Verify that all are the same character.
1757     for (int32_t l = 1; l < len; l++) {
1758         if (ch != s.charAt(l)) {
1759             return -1;
1760         }
1761     }
1762     int32_t i = 0;
1763     int32_t bestRow = -1;
1764     while (dtTypes[i].patternChar != '\0') {
1765         if ( dtTypes[i].patternChar != ch ) {
1766             ++i;
1767             continue;
1768         }
1769         bestRow = i;
1770         if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
1771             return i;
1772         }
1773         if (dtTypes[i+1].minLen <= len) {
1774             ++i;
1775             continue;
1776         }
1777         return i;
1778     }
1779     return strict ? -1 : bestRow;
1780 }
1781
1782 UBool
1783 FormatParser::isQuoteLiteral(const UnicodeString& s) const {
1784     return (UBool)(s.charAt(0)==SINGLE_QUOTE);
1785 }
1786
1787 // This function aussumes the current itemIndex points to the quote literal.
1788 // Please call isQuoteLiteral prior to this function.
1789 void
1790 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
1791     int32_t i=*itemIndex;
1792
1793     quote.remove();
1794     if (items[i].charAt(0)==SINGLE_QUOTE) {
1795         quote += items[i];
1796         ++i;
1797     }
1798     while ( i < itemNumber ) {
1799         if ( items[i].charAt(0)==SINGLE_QUOTE ) {
1800             if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
1801                 // two single quotes e.g. 'o''clock'
1802                 quote += items[i++];
1803                 quote += items[i++];
1804                 continue;
1805             }
1806             else {
1807                 quote += items[i];
1808                 break;
1809             }
1810         }
1811         else {
1812             quote += items[i];
1813         }
1814         ++i;
1815     }
1816     *itemIndex=i;
1817 }
1818
1819 UBool
1820 FormatParser::isPatternSeparator(UnicodeString& field) {
1821     for (int32_t i=0; i<field.length(); ++i ) {
1822         UChar c= field.charAt(i);
1823         if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
1824              (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
1825             continue;
1826         }
1827         else {
1828             return FALSE;
1829         }
1830     }
1831     return TRUE;
1832 }
1833
1834 void
1835 DistanceInfo::setTo(DistanceInfo &other) {
1836     missingFieldMask = other.missingFieldMask;
1837     extraFieldMask= other.extraFieldMask;
1838 }
1839
1840 PatternMapIterator::PatternMapIterator() {
1841     bootIndex = 0;
1842     nodePtr = NULL;
1843     patternMap=NULL;
1844     matcher= new DateTimeMatcher();
1845 }
1846
1847
1848 PatternMapIterator::~PatternMapIterator() {
1849     delete matcher;
1850 }
1851
1852 void
1853 PatternMapIterator::set(PatternMap& newPatternMap) {
1854     this->patternMap=&newPatternMap;
1855 }
1856
1857 PtnSkeleton*
1858 PatternMapIterator::getSkeleton() {
1859     if ( nodePtr == NULL ) {
1860         return NULL;
1861     }
1862     else {
1863         return nodePtr->skeleton;
1864     }
1865 }
1866
1867 UBool
1868 PatternMapIterator::hasNext() {
1869     int32_t headIndex=bootIndex;
1870     PtnElem *curPtr=nodePtr;
1871
1872     if (patternMap==NULL) {
1873         return FALSE;
1874     }
1875     while ( headIndex < MAX_PATTERN_ENTRIES ) {
1876         if ( curPtr != NULL ) {
1877             if ( curPtr->next != NULL ) {
1878                 return TRUE;
1879             }
1880             else {
1881                 headIndex++;
1882                 curPtr=NULL;
1883                 continue;
1884             }
1885         }
1886         else {
1887             if ( patternMap->boot[headIndex] != NULL ) {
1888                 return TRUE;
1889             }
1890             else {
1891                 headIndex++;
1892                 continue;
1893             }
1894         }
1895
1896     }
1897     return FALSE;
1898 }
1899
1900 DateTimeMatcher&
1901 PatternMapIterator::next() {
1902     while ( bootIndex < MAX_PATTERN_ENTRIES ) {
1903         if ( nodePtr != NULL ) {
1904             if ( nodePtr->next != NULL ) {
1905                 nodePtr = nodePtr->next;
1906                 break;
1907             }
1908             else {
1909                 bootIndex++;
1910                 nodePtr=NULL;
1911                 continue;
1912             }
1913         }
1914         else {
1915             if ( patternMap->boot[bootIndex] != NULL ) {
1916                 nodePtr = patternMap->boot[bootIndex];
1917                 break;
1918             }
1919             else {
1920                 bootIndex++;
1921                 continue;
1922             }
1923         }
1924     }
1925     if (nodePtr!=NULL) {
1926         matcher->copyFrom(*nodePtr->skeleton);
1927     }
1928     else {
1929         matcher->copyFrom();
1930     }
1931     return *matcher;
1932 }
1933
1934 PtnSkeleton::PtnSkeleton() {
1935 }
1936
1937
1938 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
1939     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1940         this->type[i]=other.type[i];
1941         this->original[i]=other.original[i];
1942         this->baseOriginal[i]=other.baseOriginal[i];
1943     }
1944 }
1945
1946 UBool
1947 PtnSkeleton::equals(const PtnSkeleton& other)  {
1948     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1949         if ( (type[i]!= other.type[i]) ||
1950              (original[i]!=other.original[i]) ||
1951              (baseOriginal[i]!=other.baseOriginal[i]) ) {
1952             return FALSE;
1953         }
1954     }
1955     return TRUE;
1956 }
1957
1958 UnicodeString
1959 PtnSkeleton::getSkeleton() {
1960     UnicodeString result;
1961
1962     for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
1963         if (original[i].length()!=0) {
1964             result += original[i];
1965         }
1966     }
1967     return result;
1968 }
1969
1970 UnicodeString
1971 PtnSkeleton::getBaseSkeleton() {
1972     UnicodeString result;
1973
1974     for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
1975         if (baseOriginal[i].length()!=0) {
1976             result += baseOriginal[i];
1977         }
1978     }
1979     return result;
1980 }
1981
1982 PtnSkeleton::~PtnSkeleton() {
1983 }
1984
1985 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
1986 basePattern(basePat),
1987 skeleton(NULL),
1988 pattern(pat),
1989 next(NULL)
1990 {
1991 }
1992
1993 PtnElem::~PtnElem() {
1994
1995     if (next!=NULL) {
1996         delete next;
1997     }
1998     delete skeleton;
1999 }
2000
2001 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum type, UErrorCode& status) {
2002     PtnElem  *curElem;
2003     PtnSkeleton *curSkeleton;
2004     UnicodeString s;
2005     int32_t bootIndex;
2006
2007     pos=0;
2008     fSkeletons = new UVector(status);
2009     if (U_FAILURE(status)) {
2010         delete fSkeletons;
2011         return;
2012     }
2013     for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2014         curElem = patternMap.boot[bootIndex];
2015         while (curElem!=NULL) {
2016             switch(type) {
2017                 case DT_BASESKELETON:
2018                     s=curElem->basePattern;
2019                     break;
2020                 case DT_PATTERN:
2021                     s=curElem->pattern;
2022                     break;
2023                 case DT_SKELETON:
2024                     curSkeleton=curElem->skeleton;
2025                     s=curSkeleton->getSkeleton();
2026                     break;
2027             }
2028             if ( !isCanonicalItem(s) ) {
2029                 fSkeletons->addElement(new UnicodeString(s), status);
2030                 if (U_FAILURE(status)) {
2031                     delete fSkeletons;
2032                     fSkeletons = NULL;
2033                     return;
2034                 }
2035             }
2036             curElem = curElem->next;
2037         }
2038     }
2039     if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=NULL) ) {
2040         status = U_BUFFER_OVERFLOW_ERROR;
2041     }
2042 }
2043
2044 const UnicodeString*
2045 DTSkeletonEnumeration::snext(UErrorCode& status) {
2046     if (U_SUCCESS(status) && pos < fSkeletons->size()) {
2047         return (const UnicodeString*)fSkeletons->elementAt(pos++);
2048     }
2049     return NULL;
2050 }
2051
2052 void
2053 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2054     pos=0;
2055 }
2056
2057 int32_t
2058 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2059    return (fSkeletons==NULL) ? 0 : fSkeletons->size();
2060 }
2061
2062 UBool
2063 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2064     if ( item.length() != 1 ) {
2065         return FALSE;
2066     }
2067     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2068         if (item.charAt(0)==Canonical_Items[i]) {
2069             return TRUE;
2070         }
2071     }
2072     return FALSE;
2073 }
2074
2075 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2076     UnicodeString *s;
2077     for (int32_t i=0; i<fSkeletons->size(); ++i) {
2078         if ((s=(UnicodeString *)fSkeletons->elementAt(i))!=NULL) {
2079             delete s;
2080         }
2081     }
2082     delete fSkeletons;
2083 }
2084
2085 DTRedundantEnumeration::DTRedundantEnumeration() {
2086     pos=0;
2087     fPatterns = NULL;
2088 }
2089
2090 void
2091 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2092     if (U_FAILURE(status)) return;
2093     if (fPatterns == NULL)  {
2094         fPatterns = new UVector(status);
2095         if (U_FAILURE(status)) {
2096             delete fPatterns;
2097             fPatterns = NULL;
2098             return;
2099        }
2100     }
2101     fPatterns->addElement(new UnicodeString(pattern), status);
2102     if (U_FAILURE(status)) {
2103         delete fPatterns;
2104         fPatterns = NULL;
2105         return;
2106     }
2107 }
2108
2109 const UnicodeString*
2110 DTRedundantEnumeration::snext(UErrorCode& status) {
2111     if (U_SUCCESS(status) && pos < fPatterns->size()) {
2112         return (const UnicodeString*)fPatterns->elementAt(pos++);
2113     }
2114     return NULL;
2115 }
2116
2117 void
2118 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2119     pos=0;
2120 }
2121
2122 int32_t
2123 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2124        return (fPatterns==NULL) ? 0 : fPatterns->size();
2125 }
2126
2127 UBool
2128 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) {
2129     if ( item.length() != 1 ) {
2130         return FALSE;
2131     }
2132     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2133         if (item.charAt(0)==Canonical_Items[i]) {
2134             return TRUE;
2135         }
2136     }
2137     return FALSE;
2138 }
2139
2140 DTRedundantEnumeration::~DTRedundantEnumeration() {
2141     UnicodeString *s;
2142     for (int32_t i=0; i<fPatterns->size(); ++i) {
2143         if ((s=(UnicodeString *)fPatterns->elementAt(i))!=NULL) {
2144             delete s;
2145         }
2146     }
2147     delete fPatterns;
2148 }
2149
2150 U_NAMESPACE_END
2151
2152
2153 #endif /* #if !UCONFIG_NO_FORMATTING */
2154
2155 //eof