icuSources/i18n/dtptngen.cpp

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 2007-2013, 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 U_CDECL_BEGIN
  70
  71 static int32_t U_CALLCONV
  72 ures_a_codepointSort(const void *context, const void *left, const void *right) {
  73     //CompareContext *cmp=(CompareContext *)context;
  74     return u_strcmp(((const UResAEntry *)left)->key,
  75                     ((const UResAEntry *)right)->key);
  76 }
  77
  78 U_CDECL_END
  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_U, UDATPG_YEAR_FIELD, DT_SHORT, 1, 3},
 148     {CAP_U, UDATPG_YEAR_FIELD, DT_LONG, 4, 0},
 149     {CAP_U, UDATPG_YEAR_FIELD, DT_NARROW, 5, 0},
 150     {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2},
 151     {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0},
 152     {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0},
 153     {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
 154     {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT + DT_DELTA, 3, 0},
 155     {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG + DT_DELTA, 4, 0},
 156     {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2},
 157     {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0},
 158     {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0},
 159     {CAP_M, UDATPG_MONTH_FIELD, DT_NARROW, 5, 0},
 160     {CAP_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
 161     {CAP_L, UDATPG_MONTH_FIELD, DT_SHORT - DT_DELTA, 3, 0},
 162     {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0},
 163     {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0},
 164     {LOW_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 1},
 165     {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2},
 166     {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 0},
 167     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3},
 168     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0},
 169     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0},
 170     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2},
 171     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0},
 172     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
 173     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0},
 174     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical
 175     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0},
 176     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0},
 177     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0},
 178     {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2},
 179     {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 3},
 180     {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 0},
 181     {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20}, // really internal use, so we don't care
 182     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 0},
 183     {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour
 184     {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2},
 185     {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour
 186     {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
 187     {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2},
 188     {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2},
 189     {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000},
 190     {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 1000},
 191     {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0},
 192     {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
 193     {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3},
 194     {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0},
 195     {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3},
 196     {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
 197     {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3},
 198     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
 199     {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[]
 200  };
 201
 202 static const char* const CLDR_FIELD_APPEND[] = {
 203     "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week", "Day", "*", "*", "*",
 204     "Hour", "Minute", "Second", "*", "Timezone"
 205 };
 206
 207 static const char* const CLDR_FIELD_NAME[] = {
 208     "era", "year", "quarter", "month", "week", "*", "weekday", "*", "*", "day", "dayperiod",
 209     "hour", "minute", "second", "*", "zone"
 210 };
 211
 212 static const char* const Resource_Fields[] = {
 213     "day", "dayperiod", "era", "hour", "minute", "month", "second", "week",
 214     "weekday", "year", "zone", "quarter" };
 215
 216 // For appendItems
 217 static const UChar UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A,
 218     0x20, 0x7B, 0x31, 0x7D, 0x2524, 0};  // {0} \u251C{2}: {1}\u2524
 219
 220 //static const UChar repeatedPatterns[6]={CAP_G, CAP_E, LOW_Z, LOW_V, CAP_Q, 0}; // "GEzvQ"
 221
 222 static const char DT_DateTimePatternsTag[]="DateTimePatterns";
 223 static const char DT_DateTimeCalendarTag[]="calendar";
 224 static const char DT_DateTimeGregorianTag[]="gregorian";
 225 static const char DT_DateTimeAppendItemsTag[]="appendItems";
 226 static const char DT_DateTimeFieldsTag[]="fields";
 227 static const char DT_DateTimeAvailableFormatsTag[]="availableFormats";
 228 //static const UnicodeString repeatedPattern=UnicodeString(repeatedPatterns);
 229
 230 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateTimePatternGenerator)
 231 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)
 232 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTRedundantEnumeration)
 233
 234 DateTimePatternGenerator*  U_EXPORT2
 235 DateTimePatternGenerator::createInstance(UErrorCode& status) {
 236     return createInstance(Locale::getDefault(), status);
 237 }
 238
 239 DateTimePatternGenerator* U_EXPORT2
 240 DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) {
 241     DateTimePatternGenerator *result = new DateTimePatternGenerator(locale, status);
 242     if (result == NULL) {
 243         status = U_MEMORY_ALLOCATION_ERROR;
 244     }
 245     if (U_FAILURE(status)) {
 246         delete result;
 247         result = NULL;
 248     }
 249     return result;
 250 }
 251
 252 DateTimePatternGenerator*  U_EXPORT2
 253 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) {
 254     DateTimePatternGenerator *result = new DateTimePatternGenerator(status);
 255     if (result == NULL) {
 256         status = U_MEMORY_ALLOCATION_ERROR;
 257     }
 258     if (U_FAILURE(status)) {
 259         delete result;
 260         result = NULL;
 261     }
 262     return result;
 263 }
 264
 265 DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) :
 266     skipMatcher(NULL),
 267     fAvailableFormatKeyHash(NULL)
 268 {
 269     fp = new FormatParser();
 270     dtMatcher = new DateTimeMatcher();
 271     distanceInfo = new DistanceInfo();
 272     patternMap = new PatternMap();
 273     if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) {
 274         status = U_MEMORY_ALLOCATION_ERROR;
 275     }
 276 }
 277
 278 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status) :
 279     skipMatcher(NULL),
 280     fAvailableFormatKeyHash(NULL)
 281 {
 282     fp = new FormatParser();
 283     dtMatcher = new DateTimeMatcher();
 284     distanceInfo = new DistanceInfo();
 285     patternMap = new PatternMap();
 286     if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) {
 287         status = U_MEMORY_ALLOCATION_ERROR;
 288     }
 289     else {
 290         initData(locale, status);
 291     }
 292 }
 293
 294 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) :
 295     UObject(),
 296     skipMatcher(NULL),
 297     fAvailableFormatKeyHash(NULL)
 298 {
 299     fp = new FormatParser();
 300     dtMatcher = new DateTimeMatcher();
 301     distanceInfo = new DistanceInfo();
 302     patternMap = new PatternMap();
 303     *this=other;
 304 }
 305
 306 DateTimePatternGenerator&
 307 DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) {
 308     pLocale = other.pLocale;
 309     fDefaultHourFormatChar = other.fDefaultHourFormatChar;
 310     *fp = *(other.fp);
 311     dtMatcher->copyFrom(other.dtMatcher->skeleton);
 312     *distanceInfo = *(other.distanceInfo);
 313     dateTimeFormat = other.dateTimeFormat;
 314     decimal = other.decimal;
 315     // NUL-terminate for the C API.
 316     dateTimeFormat.getTerminatedBuffer();
 317     decimal.getTerminatedBuffer();
 318     delete skipMatcher;
 319     if ( other.skipMatcher == NULL ) {
 320         skipMatcher = NULL;
 321     }
 322     else {
 323         skipMatcher = new DateTimeMatcher(*other.skipMatcher);
 324     }
 325     for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) {
 326         appendItemFormats[i] = other.appendItemFormats[i];
 327         appendItemNames[i] = other.appendItemNames[i];
 328         // NUL-terminate for the C API.
 329         appendItemFormats[i].getTerminatedBuffer();
 330         appendItemNames[i].getTerminatedBuffer();
 331     }
 332     UErrorCode status = U_ZERO_ERROR;
 333     patternMap->copyFrom(*other.patternMap, status);
 334     copyHashtable(other.fAvailableFormatKeyHash, status);
 335     return *this;
 336 }
 337
 338
 339 UBool
 340 DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) const {
 341     if (this == &other) {
 342         return TRUE;
 343     }
 344     if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) &&
 345         (dateTimeFormat==other.dateTimeFormat) && (decimal==other.decimal)) {
 346         for ( int32_t i=0 ; i<UDATPG_FIELD_COUNT; ++i ) {
 347            if ((appendItemFormats[i] != other.appendItemFormats[i]) ||
 348                (appendItemNames[i] != other.appendItemNames[i]) ) {
 349                return FALSE;
 350            }
 351         }
 352         return TRUE;
 353     }
 354     else {
 355         return FALSE;
 356     }
 357 }
 358
 359 UBool
 360 DateTimePatternGenerator::operator!=(const DateTimePatternGenerator& other) const {
 361     return  !operator==(other);
 362 }
 363
 364 DateTimePatternGenerator::~DateTimePatternGenerator() {
 365     if (fAvailableFormatKeyHash!=NULL) {
 366         delete fAvailableFormatKeyHash;
 367     }
 368
 369     if (fp != NULL) delete fp;
 370     if (dtMatcher != NULL) delete dtMatcher;
 371     if (distanceInfo != NULL) delete distanceInfo;
 372     if (patternMap != NULL) delete patternMap;
 373     if (skipMatcher != NULL) delete skipMatcher;
 374 }
 375
 376 void
 377 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status) {
 378     //const char *baseLangName = locale.getBaseName(); // unused
 379
 380     skipMatcher = NULL;
 381     fAvailableFormatKeyHash=NULL;
 382     addCanonicalItems();
 383     addICUPatterns(locale, status);
 384     if (U_FAILURE(status)) {
 385         return;
 386     }
 387     addCLDRData(locale, status);
 388     setDateTimeFromCalendar(locale, status);
 389     setDecimalSymbols(locale, status);
 390 } // DateTimePatternGenerator::initData
 391
 392 UnicodeString
 393 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode&
 394 /*status*/) {
 395     dtMatcher->set(pattern, fp);
 396     return dtMatcher->getSkeletonPtr()->getSkeleton();
 397 }
 398
 399 UnicodeString
 400 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) {
 401     dtMatcher->set(pattern, fp);
 402     return dtMatcher->getSkeletonPtr()->getBaseSkeleton();
 403 }
 404
 405 void
 406 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) {
 407     UnicodeString dfPattern;
 408     UnicodeString conflictingString;
 409     DateFormat* df;
 410
 411     if (U_FAILURE(status)) {
 412         return;
 413     }
 414
 415     // Load with ICU patterns
 416     for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) {
 417         DateFormat::EStyle style = (DateFormat::EStyle)i;
 418         df = DateFormat::createDateInstance(style, locale);
 419         SimpleDateFormat* sdf;
 420         if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
 421             addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
 422         }
 423         // TODO Maybe we should return an error when the date format isn't simple.
 424         delete df;
 425         if (U_FAILURE(status)) {
 426             return;
 427         }
 428
 429         df = DateFormat::createTimeInstance(style, locale);
 430         if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
 431             addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
 432             // HACK for hh:ss
 433             if ( i==DateFormat::kMedium ) {
 434                 hackPattern = dfPattern;
 435             }
 436         }
 437         // TODO Maybe we should return an error when the date format isn't simple.
 438         delete df;
 439         if (U_FAILURE(status)) {
 440             return;
 441         }
 442     }
 443 }
 444
 445 void
 446 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status)  {
 447     UnicodeString conflictingString;
 448
 449     fp->set(hackPattern);
 450     UnicodeString mmss;
 451     UBool gotMm=FALSE;
 452     for (int32_t i=0; i<fp->itemNumber; ++i) {
 453         UnicodeString field = fp->items[i];
 454         if ( fp->isQuoteLiteral(field) ) {
 455             if ( gotMm ) {
 456                UnicodeString quoteLiteral;
 457                fp->getQuoteLiteral(quoteLiteral, &i);
 458                mmss += quoteLiteral;
 459             }
 460         }
 461         else {
 462             if (fp->isPatternSeparator(field) && gotMm) {
 463                 mmss+=field;
 464             }
 465             else {
 466                 UChar ch=field.charAt(0);
 467                 if (ch==LOW_M) {
 468                     gotMm=TRUE;
 469                     mmss+=field;
 470                 }
 471                 else {
 472                     if (ch==LOW_S) {
 473                         if (!gotMm) {
 474                             break;
 475                         }
 476                         mmss+= field;
 477                         addPattern(mmss, FALSE, conflictingString, status);
 478                         break;
 479                     }
 480                     else {
 481                         if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) {
 482                             break;
 483                         }
 484                     }
 485                 }
 486             }
 487         }
 488     }
 489 }
 490
 491 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY)
 492
 493 static const UChar hourFormatChars[] = { CAP_H, LOW_H, CAP_K, LOW_K, 0 }; // HhKk, the hour format characters
 494
 495 void
 496 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& err) {
 497     UResourceBundle *rb, *calTypeBundle, *calBundle;
 498     UResourceBundle *patBundle, *fieldBundle, *fBundle;
 499     UnicodeString rbPattern, value, field;
 500     UnicodeString conflictingPattern;
 501     const char *key=NULL;
 502     int32_t i;
 503
 504     UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, LENGTHOF(UDATPG_ItemFormat)-1);  // Read-only alias.
 505
 506     err = U_ZERO_ERROR;
 507
 508     fDefaultHourFormatChar = 0;
 509     for (i=0; i<UDATPG_FIELD_COUNT; ++i ) {
 510         appendItemNames[i]=CAP_F;
 511         if (i<10) {
 512             appendItemNames[i]+=(UChar)(i+0x30);
 513         }
 514         else {
 515             appendItemNames[i]+=(UChar)0x31;
 516             appendItemNames[i]+=(UChar)(i-10 + 0x30);
 517         }
 518         // NUL-terminate for the C API.
 519         appendItemNames[i].getTerminatedBuffer();
 520     }
 521
 522     rb = ures_open(NULL, locale.getName(), &err);
 523     if (rb == NULL || U_FAILURE(err)) {
 524         return;
 525     }
 526     const char *curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
 527     const char * calendarTypeToUse = DT_DateTimeGregorianTag; // initial default
 528     char         calendarType[ULOC_KEYWORDS_CAPACITY]; // to be filled in with the type to use, if all goes well
 529     if ( U_SUCCESS(err) ) {
 530         char    localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
 531         // obtain a locale that always has the calendar key value that should be used
 532         (void)ures_getFunctionalEquivalent(localeWithCalendarKey, ULOC_LOCALE_IDENTIFIER_CAPACITY, NULL,
 533                                             "calendar", "calendar", locale.getName(), NULL, FALSE, &err);
 534         localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination
 535         // now get the calendar key value from that locale
 536         int32_t calendarTypeLen = uloc_getKeywordValue(localeWithCalendarKey, "calendar", calendarType, ULOC_KEYWORDS_CAPACITY, &err);
 537         if (U_SUCCESS(err) && calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
 538             calendarTypeToUse = calendarType;
 539         }
 540         err = U_ZERO_ERROR;
 541     }
 542     calBundle = ures_getByKeyWithFallback(rb, DT_DateTimeCalendarTag, NULL, &err);
 543     calTypeBundle = ures_getByKeyWithFallback(calBundle, calendarTypeToUse, NULL, &err);
 544
 545     key=NULL;
 546     int32_t dtCount=0;
 547     patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimePatternsTag, NULL, &err);
 548     while (U_SUCCESS(err)) {
 549         rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
 550         dtCount++;
 551         if (rbPattern.length()==0 ) {
 552             break;  // no more pattern
 553         }
 554         else {
 555             if (dtCount==9) {
 556                 setDateTimeFormat(rbPattern);
 557             } else if (dtCount==4) { // short time format
 558                 // set fDefaultHourFormatChar to the hour format character from this pattern
 559                 int32_t tfIdx, tfLen = rbPattern.length();
 560                 UBool ignoreChars = FALSE;
 561                 for (tfIdx = 0; tfIdx < tfLen; tfIdx++) {
 562                     UChar tfChar = rbPattern.charAt(tfIdx);
 563                     if ( tfChar == SINGLE_QUOTE ) {
 564                         ignoreChars = !ignoreChars; // toggle (handle quoted literals & '' for single quote)
 565                     } else if ( !ignoreChars && u_strchr(hourFormatChars, tfChar) != NULL ) {
 566                         fDefaultHourFormatChar = tfChar;
 567                         break;
 568                     }
 569                 }
 570             }
 571         }
 572     }
 573     ures_close(patBundle);
 574
 575     err = U_ZERO_ERROR;
 576     patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAppendItemsTag, NULL, &err);
 577     key=NULL;
 578     UnicodeString itemKey;
 579     while (U_SUCCESS(err)) {
 580         rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
 581         if (rbPattern.length()==0 ) {
 582             break;  // no more pattern
 583         }
 584         else {
 585             setAppendItemFormat(getAppendFormatNumber(key), rbPattern);
 586         }
 587     }
 588     ures_close(patBundle);
 589
 590     key=NULL;
 591     err = U_ZERO_ERROR;
 592     fBundle = ures_getByKeyWithFallback(rb, DT_DateTimeFieldsTag, NULL, &err);
 593     for (i=0; i<MAX_RESOURCE_FIELD; ++i) {
 594         err = U_ZERO_ERROR;
 595         patBundle = ures_getByKeyWithFallback(fBundle, Resource_Fields[i], NULL, &err);
 596         fieldBundle = ures_getByKeyWithFallback(patBundle, "dn", NULL, &err);
 597         rbPattern = ures_getNextUnicodeString(fieldBundle, &key, &err);
 598         ures_close(fieldBundle);
 599         ures_close(patBundle);
 600         if (rbPattern.length()==0 ) {
 601             continue;
 602         }
 603         else {
 604             setAppendItemName(getAppendNameNumber(Resource_Fields[i]), rbPattern);
 605         }
 606     }
 607     ures_close(fBundle);
 608
 609     // add available formats
 610     UBool firstTimeThrough = TRUE;
 611     err = U_ZERO_ERROR;
 612     initHashtable(err);
 613     UBool override = TRUE;
 614     while (TRUE) {
 615         // At the start of the loop:
 616         // - rb is the open resource bundle for the current locale being processed,
 617         //   whose actual name is in curLocaleName.
 618         // - if U_SUCCESS(err), then calBundle and calTypeBundle are open;
 619         //   process contents of calTypeBundle, then close calBundle and calTypeBundle.
 620         if (U_SUCCESS(err)) {
 621             // process contents of calTypeBundle
 622             patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAvailableFormatsTag, NULL, &err);
 623             if (U_SUCCESS(err)) {
 624                 int32_t numberKeys = ures_getSize(patBundle);
 625                 int32_t len;
 626                 const UChar *retPattern;
 627                 key=NULL;
 628 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 629                 UResourceBundleAIterator aiter;
 630                 ures_a_open(&aiter, patBundle, &err);
 631 #endif
 632                 for(i=0; i<numberKeys; ++i) {
 633 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 634                     retPattern=ures_a_getNextString(&aiter, &len, &key, &err);
 635 #else
 636                     retPattern=ures_getNextString(patBundle, &len, &key, &err);
 637 #endif
 638                     UnicodeString format=UnicodeString(retPattern);
 639                     UnicodeString retKey=UnicodeString(key, -1, US_INV);
 640                     if ( firstTimeThrough || !isAvailableFormatSet(retKey) ) {
 641                         setAvailableFormat(retKey, err);
 642                         // Add pattern with its associated skeleton. Override any duplicate derived from std patterns,
 643                         // but not a previous availableFormats entry:
 644                         addPatternWithSkeleton(format, &retKey, override, conflictingPattern, err);
 645                     }
 646                 }
 647 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
 648                 ures_a_close(&aiter);
 649 #endif
 650                 ures_close(patBundle);
 651             }
 652             firstTimeThrough = FALSE;
 653             // close calBundle and calTypeBundle
 654             ures_close(calTypeBundle);
 655             ures_close(calBundle);
 656         }
 657         if (uprv_strcmp(curLocaleName,"root")==0 || uprv_strlen(curLocaleName)==0) {
 658             // we just finished handling root, nothing more to check
 659             ures_close(rb);
 660             break;
 661         }
 662         // Find the name of the appropriate parent locale (from %%Parent if present, else
 663         // uloc_getParent on the actual locale name)
 664         // (It would be nice to have a ures function that did this...)
 665         err = U_ZERO_ERROR;
 666         char parentLocale[ULOC_FULLNAME_CAPACITY];
 667         int32_t locNameLen;
 668         const UChar * parentUName = ures_getStringByKey(rb, "%%Parent", &locNameLen, &err);
 669         if (U_SUCCESS(err) && err != U_USING_FALLBACK_WARNING && locNameLen < ULOC_FULLNAME_CAPACITY) {
 670             u_UCharsToChars(parentUName, parentLocale, locNameLen + 1);
 671         } else {
 672             err = U_ZERO_ERROR;
 673             uloc_getParent(curLocaleName, parentLocale, ULOC_FULLNAME_CAPACITY, &err);
 674             if (U_FAILURE(err) || err == U_STRING_NOT_TERMINATED_WARNING) {
 675                 // just fallback to root, since we are not already there
 676                 parentLocale[0] = 0;
 677                 err = U_ZERO_ERROR;
 678             }
 679         }
 680         // Close current locale bundle
 681         ures_close(rb);
 682         // And open its parent, which becomes the new current locale being processed
 683         rb = ures_open(NULL, parentLocale, &err);
 684         if ( U_FAILURE(err) ) {
 685             err = U_ZERO_ERROR;
 686             break;
 687         }
 688         // Get the name of the parent / new current locale
 689         curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
 690         if ( U_FAILURE(err) ) {
 691             curLocaleName = parentLocale;
 692             err = U_ZERO_ERROR;
 693         }
 694         if (uprv_strcmp(curLocaleName,"root")==0 || uprv_strlen(curLocaleName)==0) {
 695             override = FALSE;
 696         }
 697         // Open calBundle and calTypeBundle
 698         calBundle = ures_getByKeyWithFallback(rb, DT_DateTimeCalendarTag, NULL, &err);
 699         if (U_SUCCESS(err)) {
 700             calTypeBundle = ures_getByKeyWithFallback(calBundle, calendarTypeToUse, NULL, &err);
 701             if ( U_FAILURE(err) ) {
 702                 ures_close(calBundle);
 703             }
 704         }
 705         // Go to the top of the loop to process contents of calTypeBundle
 706     }
 707
 708     if (hackPattern.length()>0) {
 709         hackTimes(hackPattern, err);
 710     }
 711 }
 712
 713 void
 714 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
 715     if (fAvailableFormatKeyHash!=NULL) {
 716         return;
 717     }
 718     if ((fAvailableFormatKeyHash = new Hashtable(FALSE, err))==NULL) {
 719         err=U_MEMORY_ALLOCATION_ERROR;
 720         return;
 721     }
 722 }
 723
 724
 725 void
 726 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
 727     appendItemFormats[field] = value;
 728     // NUL-terminate for the C API.
 729     appendItemFormats[field].getTerminatedBuffer();
 730 }
 731
 732 const UnicodeString&
 733 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
 734     return appendItemFormats[field];
 735 }
 736
 737 void
 738 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
 739     appendItemNames[field] = value;
 740     // NUL-terminate for the C API.
 741     appendItemNames[field].getTerminatedBuffer();
 742 }
 743
 744 const UnicodeString&
 745 DateTimePatternGenerator:: getAppendItemName(UDateTimePatternField field) const {
 746     return appendItemNames[field];
 747 }
 748
 749 void
 750 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
 751     value = SINGLE_QUOTE;
 752     value += appendItemNames[field];
 753     value += SINGLE_QUOTE;
 754 }
 755
 756 UnicodeString
 757 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
 758     return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
 759 }
 760
 761 UnicodeString
 762 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
 763     const UnicodeString *bestPattern=NULL;
 764     UnicodeString dtFormat;
 765     UnicodeString resultPattern;
 766
 767     int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
 768     int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
 769
 770     UnicodeString patternFormCopy = UnicodeString(patternForm);
 771     patternFormCopy.findAndReplace(UnicodeString(LOW_J), UnicodeString(fDefaultHourFormatChar));
 772
 773     resultPattern.remove();
 774     dtMatcher->set(patternFormCopy, fp);
 775     const PtnSkeleton* specifiedSkeleton=NULL;
 776     bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, &specifiedSkeleton);
 777     if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
 778         resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, FALSE, options);
 779
 780         return resultPattern;
 781     }
 782     int32_t neededFields = dtMatcher->getFieldMask();
 783     UnicodeString datePattern=getBestAppending(neededFields & dateMask, options);
 784     UnicodeString timePattern=getBestAppending(neededFields & timeMask, options);
 785     if (datePattern.length()==0) {
 786         if (timePattern.length()==0) {
 787             resultPattern.remove();
 788         }
 789         else {
 790             return timePattern;
 791         }
 792     }
 793     if (timePattern.length()==0) {
 794         return datePattern;
 795     }
 796     resultPattern.remove();
 797     status = U_ZERO_ERROR;
 798     dtFormat=getDateTimeFormat();
 799     Formattable dateTimeObject[] = { timePattern, datePattern };
 800     resultPattern = MessageFormat::format(dtFormat, dateTimeObject, 2, resultPattern, status );
 801     return resultPattern;
 802 }
 803
 804 UnicodeString
 805 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
 806                                             const UnicodeString& skeleton,
 807                                             UErrorCode& status) {
 808     return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
 809 }
 810
 811 UnicodeString
 812 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
 813                                             const UnicodeString& skeleton,
 814                                             UDateTimePatternMatchOptions options,
 815                                             UErrorCode& /*status*/) {
 816     dtMatcher->set(skeleton, fp);
 817     UnicodeString result = adjustFieldTypes(pattern, NULL, FALSE, options);
 818     return result;
 819 }
 820
 821 void
 822 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
 823     this->decimal = newDecimal;
 824     // NUL-terminate for the C API.
 825     this->decimal.getTerminatedBuffer();
 826 }
 827
 828 const UnicodeString&
 829 DateTimePatternGenerator::getDecimal() const {
 830     return decimal;
 831 }
 832
 833 void
 834 DateTimePatternGenerator::addCanonicalItems() {
 835     UnicodeString  conflictingPattern;
 836     UErrorCode status = U_ZERO_ERROR;
 837
 838     for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
 839         addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
 840     }
 841 }
 842
 843 void
 844 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
 845     dateTimeFormat = dtFormat;
 846     // NUL-terminate for the C API.
 847     dateTimeFormat.getTerminatedBuffer();
 848 }
 849
 850 const UnicodeString&
 851 DateTimePatternGenerator::getDateTimeFormat() const {
 852     return dateTimeFormat;
 853 }
 854
 855 void
 856 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
 857     const UChar *resStr;
 858     int32_t resStrLen = 0;
 859
 860     Calendar* fCalendar = Calendar::createInstance(locale, status);
 861     CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
 862     UResourceBundle *dateTimePatterns = calData.getByKey(DT_DateTimePatternsTag, status);
 863     if (U_FAILURE(status)) return;
 864
 865     if (ures_getSize(dateTimePatterns) <= DateFormat::kDateTime)
 866     {
 867         status = U_INVALID_FORMAT_ERROR;
 868         return;
 869     }
 870     resStr = ures_getStringByIndex(dateTimePatterns, (int32_t)DateFormat::kDateTime, &resStrLen, &status);
 871     setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
 872
 873     delete fCalendar;
 874 }
 875
 876 void
 877 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
 878     DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
 879     if(U_SUCCESS(status)) {
 880         decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
 881         // NUL-terminate for the C API.
 882         decimal.getTerminatedBuffer();
 883     }
 884 }
 885
 886 UDateTimePatternConflict
 887 DateTimePatternGenerator::addPattern(
 888     const UnicodeString& pattern,
 889     UBool override,
 890     UnicodeString &conflictingPattern,
 891     UErrorCode& status)
 892 {
 893     return addPatternWithSkeleton(pattern, NULL, override, conflictingPattern, status);
 894 }
 895
 896 // For DateTimePatternGenerator::addPatternWithSkeleton -
 897 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
 898 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
 899 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
 900 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
 901 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
 902 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
 903 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
 904 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
 905 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
 906 UDateTimePatternConflict
 907 DateTimePatternGenerator::addPatternWithSkeleton(
 908     const UnicodeString& pattern,
 909     const UnicodeString* skeletonToUse,
 910     UBool override,
 911     UnicodeString& conflictingPattern,
 912     UErrorCode& status)
 913 {
 914
 915     UnicodeString basePattern;
 916     PtnSkeleton   skeleton;
 917     UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
 918
 919     DateTimeMatcher matcher;
 920     if ( skeletonToUse == NULL ) {
 921         matcher.set(pattern, fp, skeleton);
 922         matcher.getBasePattern(basePattern);
 923     } else {
 924         matcher.set(*skeletonToUse, fp, skeleton); // no longer trims skeleton fields to max len 3, per #7930
 925         matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
 926     }
 927     // We only care about base conflicts - and replacing the pattern associated with a base - if:
 928     // 1. the conflicting previous base pattern did *not* have an explicit skeleton; in that case the previous
 929     // base + pattern combination was derived from either (a) a canonical item, (b) a standard format, or
 930     // (c) a pattern specified programmatically with a previous call to addPattern (which would only happen
 931     // if we are getting here from a subsequent call to addPattern).
 932     // 2. a skeleton is specified for the current pattern, but override=false; in that case we are checking
 933     // availableFormats items from root, which should not override any previous entry with the same base.
 934     UBool entryHadSpecifiedSkeleton;
 935     const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
 936     if (duplicatePattern != NULL && (!entryHadSpecifiedSkeleton || (skeletonToUse != NULL && !override))) {
 937         conflictingStatus = UDATPG_BASE_CONFLICT;
 938         conflictingPattern = *duplicatePattern;
 939         if (!override) {
 940             return conflictingStatus;
 941         }
 942     }
 943     // The only time we get here with override=true and skeletonToUse!=null is when adding availableFormats
 944     // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with
 945     // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for
 946     // the previously-specified conflicting item.
 947     const PtnSkeleton* entrySpecifiedSkeleton = NULL;
 948     duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
 949     if (duplicatePattern != NULL ) {
 950         conflictingStatus = UDATPG_CONFLICT;
 951         conflictingPattern = *duplicatePattern;
 952         if (!override || (skeletonToUse != NULL && entrySpecifiedSkeleton != NULL)) {
 953             return conflictingStatus;
 954         }
 955     }
 956     patternMap->add(basePattern, skeleton, pattern, skeletonToUse != NULL, status);
 957     if(U_FAILURE(status)) {
 958         return conflictingStatus;
 959     }
 960
 961     return UDATPG_NO_CONFLICT;
 962 }
 963
 964
 965 UDateTimePatternField
 966 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
 967     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
 968         if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
 969             return (UDateTimePatternField)i;
 970         }
 971     }
 972     return UDATPG_FIELD_COUNT;
 973 }
 974
 975 UDateTimePatternField
 976 DateTimePatternGenerator::getAppendNameNumber(const char* field) const {
 977     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
 978         if (uprv_strcmp(CLDR_FIELD_NAME[i],field)==0) {
 979             return (UDateTimePatternField)i;
 980         }
 981     }
 982     return UDATPG_FIELD_COUNT;
 983 }
 984
 985 const UnicodeString*
 986 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
 987                                      int32_t includeMask,
 988                                      DistanceInfo* missingFields,
 989                                      const PtnSkeleton** specifiedSkeletonPtr) {
 990     int32_t bestDistance = 0x7fffffff;
 991     DistanceInfo tempInfo;
 992     const UnicodeString *bestPattern=NULL;
 993     const PtnSkeleton* specifiedSkeleton=NULL;
 994
 995     PatternMapIterator it;
 996     for (it.set(*patternMap); it.hasNext(); ) {
 997         DateTimeMatcher trial = it.next();
 998         if (trial.equals(skipMatcher)) {
 999             continue;
1000         }
1001         int32_t distance=source.getDistance(trial, includeMask, tempInfo);
1002         if (distance<bestDistance) {
1003             bestDistance=distance;
1004             bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
1005             missingFields->setTo(tempInfo);
1006             if (distance==0) {
1007                 break;
1008             }
1009         }
1010     }
1011
1012     // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
1013     // then return it too. This generally happens when the caller needs to pass that skeleton
1014     // through to adjustFieldTypes so the latter can do a better job.
1015     if (bestPattern && specifiedSkeletonPtr) {
1016         *specifiedSkeletonPtr = specifiedSkeleton;
1017     }
1018     return bestPattern;
1019 }
1020
1021 UnicodeString
1022 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
1023                                            const PtnSkeleton* specifiedSkeleton,
1024                                            UBool fixFractionalSeconds,
1025                                            UDateTimePatternMatchOptions options) {
1026     UnicodeString newPattern;
1027     fp->set(pattern);
1028     for (int32_t i=0; i < fp->itemNumber; i++) {
1029         UnicodeString field = fp->items[i];
1030         if ( fp->isQuoteLiteral(field) ) {
1031
1032             UnicodeString quoteLiteral;
1033             fp->getQuoteLiteral(quoteLiteral, &i);
1034             newPattern += quoteLiteral;
1035         }
1036         else {
1037             if (fp->isPatternSeparator(field)) {
1038                 newPattern+=field;
1039                 continue;
1040             }
1041             int32_t canonicalIndex = fp->getCanonicalIndex(field);
1042             if (canonicalIndex < 0) {
1043                 newPattern+=field;
1044                 continue;  // don't adjust
1045             }
1046             const dtTypeElem *row = &dtTypes[canonicalIndex];
1047             int32_t typeValue = row->field;
1048             if (fixFractionalSeconds && typeValue == UDATPG_SECOND_FIELD) {
1049                 UnicodeString newField=dtMatcher->skeleton.original[UDATPG_FRACTIONAL_SECOND_FIELD];
1050                 field = field + decimal + newField;
1051             } else if (dtMatcher->skeleton.type[typeValue]!=0) {
1052                     // Here:
1053                     // - "reqField" is the field from the originally requested skeleton, with length
1054                     // "reqFieldLen".
1055                     // - "field" is the field from the found pattern.
1056                     //
1057                     // The adjusted field should consist of characters from the originally requested
1058                     // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1059                     // UDATPG_WEEKDAY_FIELD or UDATPG_YEAR_FIELD, in which case it should consist
1060                     // of characters from the  found pattern.
1061                     //
1062                     // The length of the adjusted field (adjFieldLen) should match that in the originally
1063                     // requested skeleton, except that in the following cases the length of the adjusted field
1064                     // should match that in the found pattern (i.e. the length of this pattern field should
1065                     // not be adjusted):
1066                     // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1067                     //    not set (ticket #7180). Note, we may want to implement a similar change for other
1068                     //    numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1069                     //    field length, but options bits can be used to override this.
1070                     // 2. There is a specified skeleton for the found pattern and one of the following is true:
1071                     //    a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1072                     //    b) The pattern field is numeric and the skeleton field is not, or vice versa.
1073
1074                     UnicodeString reqField = dtMatcher->skeleton.original[typeValue];
1075                     int32_t reqFieldLen = reqField.length();
1076                     if (reqField.charAt(0) == CAP_E && reqFieldLen < 3)
1077                         reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1078                     int32_t adjFieldLen = reqFieldLen;
1079                     if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1080                          (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1081                          (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1082                          adjFieldLen = field.length();
1083                     } else if (specifiedSkeleton) {
1084                         UnicodeString skelField = specifiedSkeleton->original[typeValue];
1085                         int32_t skelFieldLen = skelField.length();
1086                         UBool patFieldIsNumeric = (row->type > 0);
1087                         UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1088                         if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1089                             // don't adjust the field length in the found pattern
1090                             adjFieldLen = field.length();
1091                         }
1092                     }
1093                     UChar c = (typeValue!= UDATPG_HOUR_FIELD && typeValue!= UDATPG_MONTH_FIELD &&
1094                                typeValue!= UDATPG_WEEKDAY_FIELD && (typeValue!= UDATPG_YEAR_FIELD || reqField.charAt(0)==CAP_Y))?
1095                         reqField.charAt(0): field.charAt(0);
1096                     field.remove();
1097                     for (int32_t i=adjFieldLen; i>0; --i) {
1098                         field+=c;
1099                     }
1100             }
1101             newPattern+=field;
1102         }
1103     }
1104     return newPattern;
1105 }
1106
1107 UnicodeString
1108 DateTimePatternGenerator::getBestAppending(int32_t missingFields, UDateTimePatternMatchOptions options) {
1109     UnicodeString  resultPattern, tempPattern;
1110     UErrorCode err=U_ZERO_ERROR;
1111     int32_t lastMissingFieldMask=0;
1112     if (missingFields!=0) {
1113         resultPattern=UnicodeString();
1114         const PtnSkeleton* specifiedSkeleton=NULL;
1115         tempPattern = *getBestRaw(*dtMatcher, missingFields, distanceInfo, &specifiedSkeleton);
1116         resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options);
1117         if ( distanceInfo->missingFieldMask==0 ) {
1118             return resultPattern;
1119         }
1120         while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1121             if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1122                 break;  // cannot find the proper missing field
1123             }
1124             if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1125                 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1126                 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, TRUE, options);
1127                 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1128                 continue;
1129             }
1130             int32_t startingMask = distanceInfo->missingFieldMask;
1131             tempPattern = *getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, &specifiedSkeleton);
1132             tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options);
1133             int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1134             int32_t topField=getTopBitNumber(foundMask);
1135             UnicodeString appendName;
1136             getAppendName((UDateTimePatternField)topField, appendName);
1137             const Formattable formatPattern[] = {
1138                 resultPattern,
1139                 tempPattern,
1140                 appendName
1141             };
1142             UnicodeString emptyStr;
1143             resultPattern = MessageFormat::format(appendItemFormats[topField], formatPattern, 3, emptyStr, err);
1144             lastMissingFieldMask = distanceInfo->missingFieldMask;
1145         }
1146     }
1147     return resultPattern;
1148 }
1149
1150 int32_t
1151 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) {
1152     if ( foundMask==0 ) {
1153         return 0;
1154     }
1155     int32_t i=0;
1156     while (foundMask!=0) {
1157         foundMask >>=1;
1158         ++i;
1159     }
1160     if (i-1 >UDATPG_ZONE_FIELD) {
1161         return UDATPG_ZONE_FIELD;
1162     }
1163     else
1164         return i-1;
1165 }
1166
1167 void
1168 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1169 {
1170     fAvailableFormatKeyHash->puti(key, 1, err);
1171 }
1172
1173 UBool
1174 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1175     return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1176 }
1177
1178 void
1179 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1180
1181     if (other == NULL) {
1182         return;
1183     }
1184     if (fAvailableFormatKeyHash != NULL) {
1185         delete fAvailableFormatKeyHash;
1186         fAvailableFormatKeyHash = NULL;
1187     }
1188     initHashtable(status);
1189     if(U_FAILURE(status)){
1190         return;
1191     }
1192     int32_t pos = -1;
1193     const UHashElement* elem = NULL;
1194     // walk through the hash table and create a deep clone
1195     while((elem = other->nextElement(pos))!= NULL){
1196         const UHashTok otherKeyTok = elem->key;
1197         UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1198         fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1199         if(U_FAILURE(status)){
1200             return;
1201         }
1202     }
1203 }
1204
1205 StringEnumeration*
1206 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1207     StringEnumeration* skeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status);
1208     return skeletonEnumerator;
1209 }
1210
1211 const UnicodeString&
1212 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1213     PtnElem *curElem;
1214
1215     if (skeleton.length() ==0) {
1216         return emptyString;
1217     }
1218     curElem = patternMap->getHeader(skeleton.charAt(0));
1219     while ( curElem != NULL ) {
1220         if ( curElem->skeleton->getSkeleton()==skeleton ) {
1221             return curElem->pattern;
1222         }
1223         curElem=curElem->next;
1224     }
1225     return emptyString;
1226 }
1227
1228 StringEnumeration*
1229 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1230     StringEnumeration* baseSkeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status);
1231     return baseSkeletonEnumerator;
1232 }
1233
1234 StringEnumeration*
1235 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1236     StringEnumeration* output = new DTRedundantEnumeration();
1237     const UnicodeString *pattern;
1238     PatternMapIterator it;
1239     for (it.set(*patternMap); it.hasNext(); ) {
1240         DateTimeMatcher current = it.next();
1241         pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1242         if ( isCanonicalItem(*pattern) ) {
1243             continue;
1244         }
1245         if ( skipMatcher == NULL ) {
1246             skipMatcher = new DateTimeMatcher(current);
1247         }
1248         else {
1249             *skipMatcher = current;
1250         }
1251         UnicodeString trial = getBestPattern(current.getPattern(), status);
1252         if (trial == *pattern) {
1253             ((DTRedundantEnumeration *)output)->add(*pattern, status);
1254         }
1255         if (current.equals(skipMatcher)) {
1256             continue;
1257         }
1258     }
1259     return output;
1260 }
1261
1262 UBool
1263 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1264     if ( item.length() != 1 ) {
1265         return FALSE;
1266     }
1267     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1268         if (item.charAt(0)==Canonical_Items[i]) {
1269             return TRUE;
1270         }
1271     }
1272     return FALSE;
1273 }
1274
1275
1276 DateTimePatternGenerator*
1277 DateTimePatternGenerator::clone() const {
1278     return new DateTimePatternGenerator(*this);
1279 }
1280
1281 PatternMap::PatternMap() {
1282    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1283       boot[i]=NULL;
1284    }
1285    isDupAllowed = TRUE;
1286 }
1287
1288 void
1289 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1290     this->isDupAllowed = other.isDupAllowed;
1291     for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1292         PtnElem *curElem, *otherElem, *prevElem=NULL;
1293         otherElem = other.boot[bootIndex];
1294         while (otherElem!=NULL) {
1295             if ((curElem = new PtnElem(otherElem->basePattern, otherElem->pattern))==NULL) {
1296                 // out of memory
1297                 status = U_MEMORY_ALLOCATION_ERROR;
1298                 return;
1299             }
1300             if ( this->boot[bootIndex]== NULL ) {
1301                 this->boot[bootIndex] = curElem;
1302             }
1303             if ((curElem->skeleton=new PtnSkeleton(*(otherElem->skeleton))) == NULL ) {
1304                 // out of memory
1305                 status = U_MEMORY_ALLOCATION_ERROR;
1306                 return;
1307             }
1308
1309             if (prevElem!=NULL) {
1310                 prevElem->next=curElem;
1311             }
1312             curElem->next=NULL;
1313             prevElem = curElem;
1314             otherElem = otherElem->next;
1315         }
1316
1317     }
1318 }
1319
1320 PtnElem*
1321 PatternMap::getHeader(UChar baseChar) {
1322     PtnElem* curElem;
1323
1324     if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1325          curElem = boot[baseChar-CAP_A];
1326     }
1327     else {
1328         if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1329             curElem = boot[26+baseChar-LOW_A];
1330         }
1331         else {
1332             return NULL;
1333         }
1334     }
1335     return curElem;
1336 }
1337
1338 PatternMap::~PatternMap() {
1339    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1340        if (boot[i]!=NULL ) {
1341            delete boot[i];
1342            boot[i]=NULL;
1343        }
1344    }
1345 }  // PatternMap destructor
1346
1347 void
1348 PatternMap::add(const UnicodeString& basePattern,
1349                 const PtnSkeleton& skeleton,
1350                 const UnicodeString& value,// mapped pattern value
1351                 UBool skeletonWasSpecified,
1352                 UErrorCode &status) {
1353     UChar baseChar = basePattern.charAt(0);
1354     PtnElem *curElem, *baseElem;
1355     status = U_ZERO_ERROR;
1356
1357     // the baseChar must be A-Z or a-z
1358     if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1359         baseElem = boot[baseChar-CAP_A];
1360     }
1361     else {
1362         if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1363             baseElem = boot[26+baseChar-LOW_A];
1364          }
1365          else {
1366              status = U_ILLEGAL_CHARACTER;
1367              return;
1368          }
1369     }
1370
1371     if (baseElem == NULL) {
1372         if ((curElem = new PtnElem(basePattern, value)) == NULL ) {
1373             // out of memory
1374             status = U_MEMORY_ALLOCATION_ERROR;
1375             return;
1376         }
1377         if (baseChar >= LOW_A) {
1378             boot[26 + (baseChar-LOW_A)] = curElem;
1379         }
1380         else {
1381             boot[baseChar-CAP_A] = curElem;
1382         }
1383         curElem->skeleton = new PtnSkeleton(skeleton);
1384         curElem->skeletonWasSpecified = skeletonWasSpecified;
1385     }
1386     if ( baseElem != NULL ) {
1387         curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1388
1389         if (curElem == NULL) {
1390             // add new element to the list.
1391             curElem = baseElem;
1392             while( curElem -> next != NULL )
1393             {
1394                 curElem = curElem->next;
1395             }
1396             if ((curElem->next = new PtnElem(basePattern, value)) == NULL ) {
1397                 // out of memory
1398                 status = U_MEMORY_ALLOCATION_ERROR;
1399                 return;
1400             }
1401             curElem=curElem->next;
1402             curElem->skeleton = new PtnSkeleton(skeleton);
1403             curElem->skeletonWasSpecified = skeletonWasSpecified;
1404         }
1405         else {
1406             // Pattern exists in the list already.
1407             if ( !isDupAllowed ) {
1408                 return;
1409             }
1410             // Overwrite the value.
1411             curElem->pattern = value;
1412             // It was a bug that we were not doing the following previously,
1413             // though that bug hid other problems by making things partly work.
1414             curElem->skeletonWasSpecified = skeletonWasSpecified;
1415         }
1416     }
1417 }  // PatternMap::add
1418
1419 // Find the pattern from the given basePattern string.
1420 const UnicodeString *
1421 PatternMap::getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeletonWasSpecified) { // key to search for
1422    PtnElem *curElem;
1423
1424    if ((curElem=getHeader(basePattern.charAt(0)))==NULL) {
1425        return NULL;  // no match
1426    }
1427
1428    do  {
1429        if ( basePattern.compare(curElem->basePattern)==0 ) {
1430           skeletonWasSpecified = curElem->skeletonWasSpecified;
1431           return &(curElem->pattern);
1432        }
1433        curElem=curElem->next;
1434    }while (curElem != NULL);
1435
1436    return NULL;
1437 }  // PatternMap::getFromBasePattern
1438
1439
1440 // Find the pattern from the given skeleton.
1441 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
1442 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
1443 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
1444 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
1445 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
1446 const UnicodeString *
1447 PatternMap::getPatternFromSkeleton(PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) { // key to search for
1448    PtnElem *curElem;
1449
1450    if (specifiedSkeletonPtr) {
1451        *specifiedSkeletonPtr = NULL;
1452    }
1453
1454    // find boot entry
1455    UChar baseChar='\0';
1456    for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1457        if (skeleton.baseOriginal[i].length() !=0 ) {
1458            baseChar = skeleton.baseOriginal[i].charAt(0);
1459            break;
1460        }
1461    }
1462
1463    if ((curElem=getHeader(baseChar))==NULL) {
1464        return NULL;  // no match
1465    }
1466
1467    do  {
1468        int32_t i=0;
1469        if (specifiedSkeletonPtr != NULL) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
1470            for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1471                if (curElem->skeleton->original[i].compare(skeleton.original[i]) != 0 )
1472                {
1473                    break;
1474                }
1475            }
1476        } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
1477            for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1478                if (curElem->skeleton->baseOriginal[i].compare(skeleton.baseOriginal[i]) != 0 )
1479                {
1480                    break;
1481                }
1482            }
1483        }
1484        if (i == UDATPG_FIELD_COUNT) {
1485            if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
1486                *specifiedSkeletonPtr = curElem->skeleton;
1487            }
1488            return &(curElem->pattern);
1489        }
1490        curElem=curElem->next;
1491    }while (curElem != NULL);
1492
1493    return NULL;
1494 }
1495
1496 UBool
1497 PatternMap::equals(const PatternMap& other) {
1498     if ( this==&other ) {
1499         return TRUE;
1500     }
1501     for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1502         if ( boot[bootIndex]==other.boot[bootIndex] ) {
1503             continue;
1504         }
1505         if ( (boot[bootIndex]==NULL)||(other.boot[bootIndex]==NULL) ) {
1506             return FALSE;
1507         }
1508         PtnElem *otherElem = other.boot[bootIndex];
1509         PtnElem *myElem = boot[bootIndex];
1510         while ((otherElem!=NULL) || (myElem!=NULL)) {
1511             if ( myElem == otherElem ) {
1512                 break;
1513             }
1514             if ((otherElem==NULL) || (myElem==NULL)) {
1515                 return FALSE;
1516             }
1517             if ( (myElem->basePattern != otherElem->basePattern) ||
1518                  (myElem->pattern != otherElem->pattern) ) {
1519                 return FALSE;
1520             }
1521             if ((myElem->skeleton!=otherElem->skeleton)&&
1522                 !myElem->skeleton->equals(*(otherElem->skeleton))) {
1523                 return FALSE;
1524             }
1525             myElem = myElem->next;
1526             otherElem=otherElem->next;
1527         }
1528     }
1529     return TRUE;
1530 }
1531
1532 // find any key existing in the mapping table already.
1533 // return TRUE if there is an existing key, otherwise return FALSE.
1534 PtnElem*
1535 PatternMap::getDuplicateElem(
1536             const UnicodeString &basePattern,
1537             const PtnSkeleton &skeleton,
1538             PtnElem *baseElem)  {
1539    PtnElem *curElem;
1540
1541    if ( baseElem == (PtnElem *)NULL )  {
1542          return (PtnElem*)NULL;
1543    }
1544    else {
1545          curElem = baseElem;
1546    }
1547    do {
1548      if ( basePattern.compare(curElem->basePattern)==0 ) {
1549         UBool isEqual=TRUE;
1550         for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1551             if (curElem->skeleton->type[i] != skeleton.type[i] ) {
1552                 isEqual=FALSE;
1553                 break;
1554             }
1555         }
1556         if (isEqual) {
1557             return curElem;
1558         }
1559      }
1560      curElem = curElem->next;
1561    } while( curElem != (PtnElem *)NULL );
1562
1563    // end of the list
1564    return (PtnElem*)NULL;
1565
1566 }  // PatternMap::getDuplicateElem
1567
1568 DateTimeMatcher::DateTimeMatcher(void) {
1569 }
1570
1571 DateTimeMatcher::~DateTimeMatcher() {}
1572
1573 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
1574     copyFrom(other.skeleton);
1575 }
1576
1577
1578 void
1579 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
1580     PtnSkeleton localSkeleton;
1581     return set(pattern, fp, localSkeleton);
1582 }
1583
1584 void
1585 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
1586     int32_t i;
1587     for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1588         skeletonResult.type[i]=NONE;
1589     }
1590     fp->set(pattern);
1591     for (i=0; i < fp->itemNumber; i++) {
1592         UnicodeString field = fp->items[i];
1593         if ( field.charAt(0) == LOW_A ) {
1594             continue;  // skip 'a'
1595         }
1596
1597         if ( fp->isQuoteLiteral(field) ) {
1598             UnicodeString quoteLiteral;
1599             fp->getQuoteLiteral(quoteLiteral, &i);
1600             continue;
1601         }
1602         int32_t canonicalIndex = fp->getCanonicalIndex(field);
1603         if (canonicalIndex < 0 ) {
1604             continue;
1605         }
1606         const dtTypeElem *row = &dtTypes[canonicalIndex];
1607         int32_t typeValue = row->field;
1608         skeletonResult.original[typeValue]=field;
1609         UChar repeatChar = row->patternChar;
1610         int32_t repeatCount = row->minLen; // #7930 removes cap at 3
1611         while (repeatCount-- > 0) {
1612             skeletonResult.baseOriginal[typeValue] += repeatChar;
1613         }
1614         int16_t subTypeValue = row->type;
1615         if ( row->type > 0) {
1616             subTypeValue += field.length();
1617         }
1618         skeletonResult.type[typeValue] = subTypeValue;
1619     }
1620     copyFrom(skeletonResult);
1621 }
1622
1623 void
1624 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
1625     result.remove(); // Reset the result first.
1626     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1627         if (skeleton.baseOriginal[i].length()!=0) {
1628             result += skeleton.baseOriginal[i];
1629         }
1630     }
1631 }
1632
1633 UnicodeString
1634 DateTimeMatcher::getPattern() {
1635     UnicodeString result;
1636
1637     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1638         if (skeleton.original[i].length()!=0) {
1639             result += skeleton.original[i];
1640         }
1641     }
1642     return result;
1643 }
1644
1645 int32_t
1646 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) {
1647     int32_t result=0;
1648     distanceInfo.clear();
1649     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1650         int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
1651         int32_t otherType = other.skeleton.type[i];
1652         if (myType==otherType) {
1653             continue;
1654         }
1655         if (myType==0) {// and other is not
1656             result += EXTRA_FIELD;
1657             distanceInfo.addExtra(i);
1658         }
1659         else {
1660             if (otherType==0) {
1661                 result += MISSING_FIELD;
1662                 distanceInfo.addMissing(i);
1663             }
1664             else {
1665                 result += abs(myType - otherType);
1666             }
1667         }
1668
1669     }
1670     return result;
1671 }
1672
1673 void
1674 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
1675     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1676         this->skeleton.type[i]=newSkeleton.type[i];
1677         this->skeleton.original[i]=newSkeleton.original[i];
1678         this->skeleton.baseOriginal[i]=newSkeleton.baseOriginal[i];
1679     }
1680 }
1681
1682 void
1683 DateTimeMatcher::copyFrom() {
1684     // same as clear
1685     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1686         this->skeleton.type[i]=0;
1687         this->skeleton.original[i].remove();
1688         this->skeleton.baseOriginal[i].remove();
1689     }
1690 }
1691
1692 UBool
1693 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
1694     if (other==NULL) {
1695         return FALSE;
1696     }
1697     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1698         if (this->skeleton.original[i]!=other->skeleton.original[i] ) {
1699             return FALSE;
1700         }
1701     }
1702     return TRUE;
1703 }
1704
1705 int32_t
1706 DateTimeMatcher::getFieldMask() {
1707     int32_t result=0;
1708
1709     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1710         if (skeleton.type[i]!=0) {
1711             result |= (1<<i);
1712         }
1713     }
1714     return result;
1715 }
1716
1717 PtnSkeleton*
1718 DateTimeMatcher::getSkeletonPtr() {
1719     return &skeleton;
1720 }
1721
1722 FormatParser::FormatParser () {
1723     status = START;
1724     itemNumber=0;
1725 }
1726
1727
1728 FormatParser::~FormatParser () {
1729 }
1730
1731
1732 // Find the next token with the starting position and length
1733 // Note: the startPos may
1734 FormatParser::TokenStatus
1735 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
1736     int32_t  curLoc = startPos;
1737     if ( curLoc >= pattern.length()) {
1738         return DONE;
1739     }
1740     // check the current char is between A-Z or a-z
1741     do {
1742         UChar c=pattern.charAt(curLoc);
1743         if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
1744            curLoc++;
1745         }
1746         else {
1747                startPos = curLoc;
1748                *len=1;
1749                return ADD_TOKEN;
1750         }
1751
1752         if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
1753             break;  // not the same token
1754         }
1755     } while(curLoc <= pattern.length());
1756     *len = curLoc-startPos;
1757     return ADD_TOKEN;
1758 }
1759
1760 void
1761 FormatParser::set(const UnicodeString& pattern) {
1762     int32_t startPos=0;
1763     TokenStatus result=START;
1764     int32_t len=0;
1765     itemNumber =0;
1766
1767     do {
1768         result = setTokens( pattern, startPos, &len );
1769         if ( result == ADD_TOKEN )
1770         {
1771             items[itemNumber++] = UnicodeString(pattern, startPos, len );
1772             startPos += len;
1773         }
1774         else {
1775             break;
1776         }
1777     } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
1778 }
1779
1780 int32_t
1781 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
1782     int32_t len = s.length();
1783     if (len == 0) {
1784         return -1;
1785     }
1786     UChar ch = s.charAt(0);
1787
1788     // Verify that all are the same character.
1789     for (int32_t l = 1; l < len; l++) {
1790         if (ch != s.charAt(l)) {
1791             return -1;
1792         }
1793     }
1794     int32_t i = 0;
1795     int32_t bestRow = -1;
1796     while (dtTypes[i].patternChar != '\0') {
1797         if ( dtTypes[i].patternChar != ch ) {
1798             ++i;
1799             continue;
1800         }
1801         bestRow = i;
1802         if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
1803             return i;
1804         }
1805         if (dtTypes[i+1].minLen <= len) {
1806             ++i;
1807             continue;
1808         }
1809         return i;
1810     }
1811     return strict ? -1 : bestRow;
1812 }
1813
1814 UBool
1815 FormatParser::isQuoteLiteral(const UnicodeString& s) const {
1816     return (UBool)(s.charAt(0)==SINGLE_QUOTE);
1817 }
1818
1819 // This function aussumes the current itemIndex points to the quote literal.
1820 // Please call isQuoteLiteral prior to this function.
1821 void
1822 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
1823     int32_t i=*itemIndex;
1824
1825     quote.remove();
1826     if (items[i].charAt(0)==SINGLE_QUOTE) {
1827         quote += items[i];
1828         ++i;
1829     }
1830     while ( i < itemNumber ) {
1831         if ( items[i].charAt(0)==SINGLE_QUOTE ) {
1832             if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
1833                 // two single quotes e.g. 'o''clock'
1834                 quote += items[i++];
1835                 quote += items[i++];
1836                 continue;
1837             }
1838             else {
1839                 quote += items[i];
1840                 break;
1841             }
1842         }
1843         else {
1844             quote += items[i];
1845         }
1846         ++i;
1847     }
1848     *itemIndex=i;
1849 }
1850
1851 UBool
1852 FormatParser::isPatternSeparator(UnicodeString& field) {
1853     for (int32_t i=0; i<field.length(); ++i ) {
1854         UChar c= field.charAt(i);
1855         if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
1856              (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
1857             continue;
1858         }
1859         else {
1860             return FALSE;
1861         }
1862     }
1863     return TRUE;
1864 }
1865
1866 DistanceInfo::~DistanceInfo() {}
1867
1868 void
1869 DistanceInfo::setTo(DistanceInfo &other) {
1870     missingFieldMask = other.missingFieldMask;
1871     extraFieldMask= other.extraFieldMask;
1872 }
1873
1874 PatternMapIterator::PatternMapIterator() {
1875     bootIndex = 0;
1876     nodePtr = NULL;
1877     patternMap=NULL;
1878     matcher= new DateTimeMatcher();
1879 }
1880
1881
1882 PatternMapIterator::~PatternMapIterator() {
1883     delete matcher;
1884 }
1885
1886 void
1887 PatternMapIterator::set(PatternMap& newPatternMap) {
1888     this->patternMap=&newPatternMap;
1889 }
1890
1891 PtnSkeleton*
1892 PatternMapIterator::getSkeleton() {
1893     if ( nodePtr == NULL ) {
1894         return NULL;
1895     }
1896     else {
1897         return nodePtr->skeleton;
1898     }
1899 }
1900
1901 UBool
1902 PatternMapIterator::hasNext() {
1903     int32_t headIndex=bootIndex;
1904     PtnElem *curPtr=nodePtr;
1905
1906     if (patternMap==NULL) {
1907         return FALSE;
1908     }
1909     while ( headIndex < MAX_PATTERN_ENTRIES ) {
1910         if ( curPtr != NULL ) {
1911             if ( curPtr->next != NULL ) {
1912                 return TRUE;
1913             }
1914             else {
1915                 headIndex++;
1916                 curPtr=NULL;
1917                 continue;
1918             }
1919         }
1920         else {
1921             if ( patternMap->boot[headIndex] != NULL ) {
1922                 return TRUE;
1923             }
1924             else {
1925                 headIndex++;
1926                 continue;
1927             }
1928         }
1929
1930     }
1931     return FALSE;
1932 }
1933
1934 DateTimeMatcher&
1935 PatternMapIterator::next() {
1936     while ( bootIndex < MAX_PATTERN_ENTRIES ) {
1937         if ( nodePtr != NULL ) {
1938             if ( nodePtr->next != NULL ) {
1939                 nodePtr = nodePtr->next;
1940                 break;
1941             }
1942             else {
1943                 bootIndex++;
1944                 nodePtr=NULL;
1945                 continue;
1946             }
1947         }
1948         else {
1949             if ( patternMap->boot[bootIndex] != NULL ) {
1950                 nodePtr = patternMap->boot[bootIndex];
1951                 break;
1952             }
1953             else {
1954                 bootIndex++;
1955                 continue;
1956             }
1957         }
1958     }
1959     if (nodePtr!=NULL) {
1960         matcher->copyFrom(*nodePtr->skeleton);
1961     }
1962     else {
1963         matcher->copyFrom();
1964     }
1965     return *matcher;
1966 }
1967
1968 PtnSkeleton::PtnSkeleton() {
1969 }
1970
1971
1972 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
1973     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1974         this->type[i]=other.type[i];
1975         this->original[i]=other.original[i];
1976         this->baseOriginal[i]=other.baseOriginal[i];
1977     }
1978 }
1979
1980 UBool
1981 PtnSkeleton::equals(const PtnSkeleton& other)  {
1982     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1983         if ( (type[i]!= other.type[i]) ||
1984              (original[i]!=other.original[i]) ||
1985              (baseOriginal[i]!=other.baseOriginal[i]) ) {
1986             return FALSE;
1987         }
1988     }
1989     return TRUE;
1990 }
1991
1992 UnicodeString
1993 PtnSkeleton::getSkeleton() {
1994     UnicodeString result;
1995
1996     for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
1997         if (original[i].length()!=0) {
1998             result += original[i];
1999         }
2000     }
2001     return result;
2002 }
2003
2004 UnicodeString
2005 PtnSkeleton::getBaseSkeleton() {
2006     UnicodeString result;
2007
2008     for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
2009         if (baseOriginal[i].length()!=0) {
2010             result += baseOriginal[i];
2011         }
2012     }
2013     return result;
2014 }
2015
2016 PtnSkeleton::~PtnSkeleton() {
2017 }
2018
2019 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
2020 basePattern(basePat),
2021 skeleton(NULL),
2022 pattern(pat),
2023 next(NULL)
2024 {
2025 }
2026
2027 PtnElem::~PtnElem() {
2028
2029     if (next!=NULL) {
2030         delete next;
2031     }
2032     delete skeleton;
2033 }
2034
2035 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum type, UErrorCode& status) {
2036     PtnElem  *curElem;
2037     PtnSkeleton *curSkeleton;
2038     UnicodeString s;
2039     int32_t bootIndex;
2040
2041     pos=0;
2042     fSkeletons = new UVector(status);
2043     if (U_FAILURE(status)) {
2044         delete fSkeletons;
2045         return;
2046     }
2047     for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2048         curElem = patternMap.boot[bootIndex];
2049         while (curElem!=NULL) {
2050             switch(type) {
2051                 case DT_BASESKELETON:
2052                     s=curElem->basePattern;
2053                     break;
2054                 case DT_PATTERN:
2055                     s=curElem->pattern;
2056                     break;
2057                 case DT_SKELETON:
2058                     curSkeleton=curElem->skeleton;
2059                     s=curSkeleton->getSkeleton();
2060                     break;
2061             }
2062             if ( !isCanonicalItem(s) ) {
2063                 fSkeletons->addElement(new UnicodeString(s), status);
2064                 if (U_FAILURE(status)) {
2065                     delete fSkeletons;
2066                     fSkeletons = NULL;
2067                     return;
2068                 }
2069             }
2070             curElem = curElem->next;
2071         }
2072     }
2073     if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=NULL) ) {
2074         status = U_BUFFER_OVERFLOW_ERROR;
2075     }
2076 }
2077
2078 const UnicodeString*
2079 DTSkeletonEnumeration::snext(UErrorCode& status) {
2080     if (U_SUCCESS(status) && pos < fSkeletons->size()) {
2081         return (const UnicodeString*)fSkeletons->elementAt(pos++);
2082     }
2083     return NULL;
2084 }
2085
2086 void
2087 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2088     pos=0;
2089 }
2090
2091 int32_t
2092 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2093    return (fSkeletons==NULL) ? 0 : fSkeletons->size();
2094 }
2095
2096 UBool
2097 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2098     if ( item.length() != 1 ) {
2099         return FALSE;
2100     }
2101     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2102         if (item.charAt(0)==Canonical_Items[i]) {
2103             return TRUE;
2104         }
2105     }
2106     return FALSE;
2107 }
2108
2109 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2110     UnicodeString *s;
2111     for (int32_t i=0; i<fSkeletons->size(); ++i) {
2112         if ((s=(UnicodeString *)fSkeletons->elementAt(i))!=NULL) {
2113             delete s;
2114         }
2115     }
2116     delete fSkeletons;
2117 }
2118
2119 DTRedundantEnumeration::DTRedundantEnumeration() {
2120     pos=0;
2121     fPatterns = NULL;
2122 }
2123
2124 void
2125 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2126     if (U_FAILURE(status)) return;
2127     if (fPatterns == NULL)  {
2128         fPatterns = new UVector(status);
2129         if (U_FAILURE(status)) {
2130             delete fPatterns;
2131             fPatterns = NULL;
2132             return;
2133        }
2134     }
2135     fPatterns->addElement(new UnicodeString(pattern), status);
2136     if (U_FAILURE(status)) {
2137         delete fPatterns;
2138         fPatterns = NULL;
2139         return;
2140     }
2141 }
2142
2143 const UnicodeString*
2144 DTRedundantEnumeration::snext(UErrorCode& status) {
2145     if (U_SUCCESS(status) && pos < fPatterns->size()) {
2146         return (const UnicodeString*)fPatterns->elementAt(pos++);
2147     }
2148     return NULL;
2149 }
2150
2151 void
2152 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2153     pos=0;
2154 }
2155
2156 int32_t
2157 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2158        return (fPatterns==NULL) ? 0 : fPatterns->size();
2159 }
2160
2161 UBool
2162 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) {
2163     if ( item.length() != 1 ) {
2164         return FALSE;
2165     }
2166     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2167         if (item.charAt(0)==Canonical_Items[i]) {
2168             return TRUE;
2169         }
2170     }
2171     return FALSE;
2172 }
2173
2174 DTRedundantEnumeration::~DTRedundantEnumeration() {
2175     UnicodeString *s;
2176     for (int32_t i=0; i<fPatterns->size(); ++i) {
2177         if ((s=(UnicodeString *)fPatterns->elementAt(i))!=NULL) {
2178             delete s;
2179         }
2180     }
2181     delete fPatterns;
2182 }
2183
2184 U_NAMESPACE_END
2185
2186
2187 #endif /* #if !UCONFIG_NO_FORMATTING */
2188
2189 //eof