• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2 *******************************************************************************
3 * Copyright (C) 1997-2012, International Business Machines Corporation and    *
4 * others. All Rights Reserved.                                                *
5 *******************************************************************************
6 *
7 * File SMPDTFMT.CPP
8 *
9 * Modification History:
10 *
11 *   Date        Name        Description
12 *   02/19/97    aliu        Converted from java.
13 *   03/31/97    aliu        Modified extensively to work with 50 locales.
14 *   04/01/97    aliu        Added support for centuries.
15 *   07/09/97    helena      Made ParsePosition into a class.
16 *   07/21/98    stephen     Added initializeDefaultCentury.
17 *                             Removed getZoneIndex (added in DateFormatSymbols)
18 *                             Removed subParseLong
19 *                             Removed chk
20 *  02/22/99     stephen     Removed character literals for EBCDIC safety
21 *   10/14/99    aliu        Updated 2-digit year parsing so that only "00" thru
22 *                           "99" are recognized. {j28 4182066}
23 *   11/15/99    weiv        Added support for week of year/day of week format
24 ********************************************************************************
25 */
26 
27 #define ZID_KEY_MAX 128
28 
29 #include "unicode/utypes.h"
30 
31 #if !UCONFIG_NO_FORMATTING
32 
33 #include "unicode/smpdtfmt.h"
34 #include "unicode/dtfmtsym.h"
35 #include "unicode/ures.h"
36 #include "unicode/msgfmt.h"
37 #include "unicode/calendar.h"
38 #include "unicode/gregocal.h"
39 #include "unicode/timezone.h"
40 #include "unicode/decimfmt.h"
41 #include "unicode/dcfmtsym.h"
42 #include "unicode/uchar.h"
43 #include "unicode/uniset.h"
44 #include "unicode/ustring.h"
45 #include "unicode/basictz.h"
46 #include "unicode/simpletz.h"
47 #include "unicode/rbtz.h"
48 #include "unicode/tzfmt.h"
49 #include "unicode/utf16.h"
50 #include "unicode/vtzone.h"
51 #include "unicode/udisplaycontext.h"
52 #include "olsontz.h"
53 #include "patternprops.h"
54 #include "fphdlimp.h"
55 #include "gregoimp.h"
56 #include "hebrwcal.h"
57 #include "cstring.h"
58 #include "uassert.h"
59 #include "cmemory.h"
60 #include "umutex.h"
61 #include <float.h>
62 #include "smpdtfst.h"
63 
64 #if defined( U_DEBUG_CALSVC ) || defined (U_DEBUG_CAL)
65 #include <stdio.h>
66 #endif
67 
68 // *****************************************************************************
69 // class SimpleDateFormat
70 // *****************************************************************************
71 
72 U_NAMESPACE_BEGIN
73 
74 static const UChar PATTERN_CHAR_BASE = 0x40;
75 
76 /**
77  * Last-resort string to use for "GMT" when constructing time zone strings.
78  */
79 // For time zones that have no names, use strings GMT+minutes and
80 // GMT-minutes. For instance, in France the time zone is GMT+60.
81 // Also accepted are GMT+H:MM or GMT-H:MM.
82 // Currently not being used
83 //static const UChar gGmt[]      = {0x0047, 0x004D, 0x0054, 0x0000};         // "GMT"
84 //static const UChar gGmtPlus[]  = {0x0047, 0x004D, 0x0054, 0x002B, 0x0000}; // "GMT+"
85 //static const UChar gGmtMinus[] = {0x0047, 0x004D, 0x0054, 0x002D, 0x0000}; // "GMT-"
86 //static const UChar gDefGmtPat[]       = {0x0047, 0x004D, 0x0054, 0x007B, 0x0030, 0x007D, 0x0000}; /* GMT{0} */
87 //static const UChar gDefGmtNegHmsPat[] = {0x002D, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x003A, 0x0073, 0x0073, 0x0000}; /* -HH:mm:ss */
88 //static const UChar gDefGmtNegHmPat[]  = {0x002D, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x0000}; /* -HH:mm */
89 //static const UChar gDefGmtPosHmsPat[] = {0x002B, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x003A, 0x0073, 0x0073, 0x0000}; /* +HH:mm:ss */
90 //static const UChar gDefGmtPosHmPat[]  = {0x002B, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x0000}; /* +HH:mm */
91 //static const UChar gUt[]       = {0x0055, 0x0054, 0x0000};  // "UT"
92 //static const UChar gUtc[]      = {0x0055, 0x0054, 0x0043, 0x0000};  // "UT"
93 
94 typedef enum GmtPatSize {
95     kGmtLen = 3,
96     kGmtPatLen = 6,
97     kNegHmsLen = 9,
98     kNegHmLen = 6,
99     kPosHmsLen = 9,
100     kPosHmLen = 6,
101     kUtLen = 2,
102     kUtcLen = 3
103 } GmtPatSize;
104 
105 // Stuff needed for numbering system overrides
106 
107 typedef enum OvrStrType {
108     kOvrStrDate = 0,
109     kOvrStrTime = 1,
110     kOvrStrBoth = 2
111 } OvrStrType;
112 
113 static const UDateFormatField kDateFields[] = {
114     UDAT_YEAR_FIELD,
115     UDAT_MONTH_FIELD,
116     UDAT_DATE_FIELD,
117     UDAT_DAY_OF_YEAR_FIELD,
118     UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
119     UDAT_WEEK_OF_YEAR_FIELD,
120     UDAT_WEEK_OF_MONTH_FIELD,
121     UDAT_YEAR_WOY_FIELD,
122     UDAT_EXTENDED_YEAR_FIELD,
123     UDAT_JULIAN_DAY_FIELD,
124     UDAT_STANDALONE_DAY_FIELD,
125     UDAT_STANDALONE_MONTH_FIELD,
126     UDAT_QUARTER_FIELD,
127     UDAT_STANDALONE_QUARTER_FIELD,
128     UDAT_YEAR_NAME_FIELD };
129 static const int8_t kDateFieldsCount = 15;
130 
131 static const UDateFormatField kTimeFields[] = {
132     UDAT_HOUR_OF_DAY1_FIELD,
133     UDAT_HOUR_OF_DAY0_FIELD,
134     UDAT_MINUTE_FIELD,
135     UDAT_SECOND_FIELD,
136     UDAT_FRACTIONAL_SECOND_FIELD,
137     UDAT_HOUR1_FIELD,
138     UDAT_HOUR0_FIELD,
139     UDAT_MILLISECONDS_IN_DAY_FIELD,
140     UDAT_TIMEZONE_RFC_FIELD };
141 static const int8_t kTimeFieldsCount = 9;
142 
143 
144 // This is a pattern-of-last-resort used when we can't load a usable pattern out
145 // of a resource.
146 static const UChar gDefaultPattern[] =
147 {
148     0x79, 0x79, 0x79, 0x79, 0x4D, 0x4D, 0x64, 0x64, 0x20, 0x68, 0x68, 0x3A, 0x6D, 0x6D, 0x20, 0x61, 0
149 };  /* "yyyyMMdd hh:mm a" */
150 
151 // This prefix is designed to NEVER MATCH real text, in order to
152 // suppress the parsing of negative numbers.  Adjust as needed (if
153 // this becomes valid Unicode).
154 static const UChar SUPPRESS_NEGATIVE_PREFIX[] = {0xAB00, 0};
155 
156 /**
157  * These are the tags we expect to see in normal resource bundle files associated
158  * with a locale.
159  */
160 static const char gDateTimePatternsTag[]="DateTimePatterns";
161 
162 //static const UChar gEtcUTC[] = {0x45, 0x74, 0x63, 0x2F, 0x55, 0x54, 0x43, 0x00}; // "Etc/UTC"
163 static const UChar QUOTE = 0x27; // Single quote
164 
165 /*
166  * The field range check bias for each UDateFormatField.
167  * The bias is added to the minimum and maximum values
168  * before they are compared to the parsed number.
169  * For example, the calendar stores zero-based month numbers
170  * but the parsed month numbers start at 1, so the bias is 1.
171  *
172  * A value of -1 means that the value is not checked.
173  */
174 static const int32_t gFieldRangeBias[] = {
175     -1,  // 'G' - UDAT_ERA_FIELD
176     -1,  // 'y' - UDAT_YEAR_FIELD
177      1,  // 'M' - UDAT_MONTH_FIELD
178      0,  // 'd' - UDAT_DATE_FIELD
179     -1,  // 'k' - UDAT_HOUR_OF_DAY1_FIELD
180     -1,  // 'H' - UDAT_HOUR_OF_DAY0_FIELD
181      0,  // 'm' - UDAT_MINUTE_FIELD
182      0,  // 's' - UDAT_SEOND_FIELD
183     -1,  // 'S' - UDAT_FRACTIONAL_SECOND_FIELD (0-999?)
184     -1,  // 'E' - UDAT_DAY_OF_WEEK_FIELD (1-7?)
185     -1,  // 'D' - UDAT_DAY_OF_YEAR_FIELD (1 - 366?)
186     -1,  // 'F' - UDAT_DAY_OF_WEEK_IN_MONTH_FIELD (1-5?)
187     -1,  // 'w' - UDAT_WEEK_OF_YEAR_FIELD (1-52?)
188     -1,  // 'W' - UDAT_WEEK_OF_MONTH_FIELD (1-5?)
189     -1,  // 'a' - UDAT_AM_PM_FIELD
190     -1,  // 'h' - UDAT_HOUR1_FIELD
191     -1,  // 'K' - UDAT_HOUR0_FIELD
192     -1,  // 'z' - UDAT_TIMEZONE_FIELD
193     -1,  // 'Y' - UDAT_YEAR_WOY_FIELD
194     -1,  // 'e' - UDAT_DOW_LOCAL_FIELD
195     -1,  // 'u' - UDAT_EXTENDED_YEAR_FIELD
196     -1,  // 'g' - UDAT_JULIAN_DAY_FIELD
197     -1,  // 'A' - UDAT_MILLISECONDS_IN_DAY_FIELD
198     -1,  // 'Z' - UDAT_TIMEZONE_RFC_FIELD
199     -1,  // 'v' - UDAT_TIMEZONE_GENERIC_FIELD
200      0,  // 'c' - UDAT_STANDALONE_DAY_FIELD
201      1,  // 'L' - UDAT_STANDALONE_MONTH_FIELD
202     -1,  // 'Q' - UDAT_QUARTER_FIELD (1-4?)
203     -1,  // 'q' - UDAT_STANDALONE_QUARTER_FIELD
204     -1   // 'V' - UDAT_TIMEZONE_SPECIAL_FIELD
205     -1,  // 'U' - UDAT_YEAR_NAME_FIELD
206 };
207 
208 // When calendar uses hebr numbering (i.e. he@calendar=hebrew),
209 // offset the years within the current millenium down to 1-999
210 static const int32_t HEBREW_CAL_CUR_MILLENIUM_START_YEAR = 5000;
211 static const int32_t HEBREW_CAL_CUR_MILLENIUM_END_YEAR = 6000;
212 
213 static UMutex LOCK = U_MUTEX_INITIALIZER;
214 
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)215 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)
216 
217 //----------------------------------------------------------------------
218 
219 SimpleDateFormat::~SimpleDateFormat()
220 {
221     delete fSymbols;
222     if (fNumberFormatters) {
223         uprv_free(fNumberFormatters);
224     }
225     if (fTimeZoneFormat) {
226         delete fTimeZoneFormat;
227     }
228 
229     while (fOverrideList) {
230         NSOverride *cur = fOverrideList;
231         fOverrideList = cur->next;
232         delete cur->nf;
233         uprv_free(cur);
234     }
235 }
236 
237 //----------------------------------------------------------------------
238 
SimpleDateFormat(UErrorCode & status)239 SimpleDateFormat::SimpleDateFormat(UErrorCode& status)
240   :   fLocale(Locale::getDefault()),
241       fSymbols(NULL),
242       fTimeZoneFormat(NULL),
243       fNumberFormatters(NULL),
244       fOverrideList(NULL),
245       fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
246 {
247     construct(kShort, (EStyle) (kShort + kDateOffset), fLocale, status);
248     initializeDefaultCentury();
249 }
250 
251 //----------------------------------------------------------------------
252 
SimpleDateFormat(const UnicodeString & pattern,UErrorCode & status)253 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
254                                    UErrorCode &status)
255 :   fPattern(pattern),
256     fLocale(Locale::getDefault()),
257     fSymbols(NULL),
258     fTimeZoneFormat(NULL),
259     fNumberFormatters(NULL),
260     fOverrideList(NULL),
261     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
262 {
263     fDateOverride.setToBogus();
264     fTimeOverride.setToBogus();
265     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
266     initialize(fLocale, status);
267     initializeDefaultCentury();
268 
269 }
270 //----------------------------------------------------------------------
271 
SimpleDateFormat(const UnicodeString & pattern,const UnicodeString & override,UErrorCode & status)272 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
273                                    const UnicodeString& override,
274                                    UErrorCode &status)
275 :   fPattern(pattern),
276     fLocale(Locale::getDefault()),
277     fSymbols(NULL),
278     fTimeZoneFormat(NULL),
279     fNumberFormatters(NULL),
280     fOverrideList(NULL),
281     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
282 {
283     fDateOverride.setTo(override);
284     fTimeOverride.setToBogus();
285     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
286     initialize(fLocale, status);
287     initializeDefaultCentury();
288 
289     processOverrideString(fLocale,override,kOvrStrBoth,status);
290 
291 }
292 
293 //----------------------------------------------------------------------
294 
SimpleDateFormat(const UnicodeString & pattern,const Locale & locale,UErrorCode & status)295 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
296                                    const Locale& locale,
297                                    UErrorCode& status)
298 :   fPattern(pattern),
299     fLocale(locale),
300     fTimeZoneFormat(NULL),
301     fNumberFormatters(NULL),
302     fOverrideList(NULL),
303     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
304 {
305 
306     fDateOverride.setToBogus();
307     fTimeOverride.setToBogus();
308 
309     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
310     initialize(fLocale, status);
311     initializeDefaultCentury();
312 }
313 
314 //----------------------------------------------------------------------
315 
SimpleDateFormat(const UnicodeString & pattern,const UnicodeString & override,const Locale & locale,UErrorCode & status)316 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
317                                    const UnicodeString& override,
318                                    const Locale& locale,
319                                    UErrorCode& status)
320 :   fPattern(pattern),
321     fLocale(locale),
322     fTimeZoneFormat(NULL),
323     fNumberFormatters(NULL),
324     fOverrideList(NULL),
325     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
326 {
327 
328     fDateOverride.setTo(override);
329     fTimeOverride.setToBogus();
330 
331     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
332     initialize(fLocale, status);
333     initializeDefaultCentury();
334 
335     processOverrideString(locale,override,kOvrStrBoth,status);
336 
337 }
338 
339 //----------------------------------------------------------------------
340 
SimpleDateFormat(const UnicodeString & pattern,DateFormatSymbols * symbolsToAdopt,UErrorCode & status)341 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
342                                    DateFormatSymbols* symbolsToAdopt,
343                                    UErrorCode& status)
344 :   fPattern(pattern),
345     fLocale(Locale::getDefault()),
346     fSymbols(symbolsToAdopt),
347     fTimeZoneFormat(NULL),
348     fNumberFormatters(NULL),
349     fOverrideList(NULL),
350     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
351 {
352 
353     fDateOverride.setToBogus();
354     fTimeOverride.setToBogus();
355 
356     initializeCalendar(NULL,fLocale,status);
357     initialize(fLocale, status);
358     initializeDefaultCentury();
359 }
360 
361 //----------------------------------------------------------------------
362 
SimpleDateFormat(const UnicodeString & pattern,const DateFormatSymbols & symbols,UErrorCode & status)363 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
364                                    const DateFormatSymbols& symbols,
365                                    UErrorCode& status)
366 :   fPattern(pattern),
367     fLocale(Locale::getDefault()),
368     fSymbols(new DateFormatSymbols(symbols)),
369     fTimeZoneFormat(NULL),
370     fNumberFormatters(NULL),
371     fOverrideList(NULL),
372     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
373 {
374 
375     fDateOverride.setToBogus();
376     fTimeOverride.setToBogus();
377 
378     initializeCalendar(NULL, fLocale, status);
379     initialize(fLocale, status);
380     initializeDefaultCentury();
381 }
382 
383 //----------------------------------------------------------------------
384 
385 // Not for public consumption; used by DateFormat
SimpleDateFormat(EStyle timeStyle,EStyle dateStyle,const Locale & locale,UErrorCode & status)386 SimpleDateFormat::SimpleDateFormat(EStyle timeStyle,
387                                    EStyle dateStyle,
388                                    const Locale& locale,
389                                    UErrorCode& status)
390 :   fLocale(locale),
391     fSymbols(NULL),
392     fTimeZoneFormat(NULL),
393     fNumberFormatters(NULL),
394     fOverrideList(NULL),
395     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
396 {
397     construct(timeStyle, dateStyle, fLocale, status);
398     if(U_SUCCESS(status)) {
399       initializeDefaultCentury();
400     }
401 }
402 
403 //----------------------------------------------------------------------
404 
405 /**
406  * Not for public consumption; used by DateFormat.  This constructor
407  * never fails.  If the resource data is not available, it uses the
408  * the last resort symbols.
409  */
SimpleDateFormat(const Locale & locale,UErrorCode & status)410 SimpleDateFormat::SimpleDateFormat(const Locale& locale,
411                                    UErrorCode& status)
412 :   fPattern(gDefaultPattern),
413     fLocale(locale),
414     fSymbols(NULL),
415     fTimeZoneFormat(NULL),
416     fNumberFormatters(NULL),
417     fOverrideList(NULL),
418     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
419 {
420     if (U_FAILURE(status)) return;
421     initializeSymbols(fLocale, initializeCalendar(NULL, fLocale, status),status);
422     if (U_FAILURE(status))
423     {
424         status = U_ZERO_ERROR;
425         delete fSymbols;
426         // This constructor doesn't fail; it uses last resort data
427         fSymbols = new DateFormatSymbols(status);
428         /* test for NULL */
429         if (fSymbols == 0) {
430             status = U_MEMORY_ALLOCATION_ERROR;
431             return;
432         }
433     }
434 
435     fDateOverride.setToBogus();
436     fTimeOverride.setToBogus();
437 
438     initialize(fLocale, status);
439     if(U_SUCCESS(status)) {
440       initializeDefaultCentury();
441     }
442 }
443 
444 //----------------------------------------------------------------------
445 
SimpleDateFormat(const SimpleDateFormat & other)446 SimpleDateFormat::SimpleDateFormat(const SimpleDateFormat& other)
447 :   DateFormat(other),
448     fLocale(other.fLocale),
449     fSymbols(NULL),
450     fTimeZoneFormat(NULL),
451     fNumberFormatters(NULL),
452     fOverrideList(NULL),
453     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
454 {
455     *this = other;
456 }
457 
458 //----------------------------------------------------------------------
459 
operator =(const SimpleDateFormat & other)460 SimpleDateFormat& SimpleDateFormat::operator=(const SimpleDateFormat& other)
461 {
462     if (this == &other) {
463         return *this;
464     }
465     DateFormat::operator=(other);
466 
467     delete fSymbols;
468     fSymbols = NULL;
469 
470     if (other.fSymbols)
471         fSymbols = new DateFormatSymbols(*other.fSymbols);
472 
473     fDefaultCenturyStart         = other.fDefaultCenturyStart;
474     fDefaultCenturyStartYear     = other.fDefaultCenturyStartYear;
475     fHaveDefaultCentury          = other.fHaveDefaultCentury;
476 
477     fPattern = other.fPattern;
478 
479     // TimeZoneFormat in ICU4C only depends on a locale for now
480     if (fLocale != other.fLocale) {
481         delete fTimeZoneFormat;
482         fTimeZoneFormat = NULL; // forces lazy instantiation with the other locale
483         fLocale = other.fLocale;
484     }
485 
486     fCapitalizationContext = other.fCapitalizationContext;
487 
488     return *this;
489 }
490 
491 //----------------------------------------------------------------------
492 
493 Format*
clone() const494 SimpleDateFormat::clone() const
495 {
496     return new SimpleDateFormat(*this);
497 }
498 
499 //----------------------------------------------------------------------
500 
501 UBool
operator ==(const Format & other) const502 SimpleDateFormat::operator==(const Format& other) const
503 {
504     if (DateFormat::operator==(other)) {
505         // DateFormat::operator== guarantees following cast is safe
506         SimpleDateFormat* that = (SimpleDateFormat*)&other;
507         return (fPattern             == that->fPattern &&
508                 fSymbols             != NULL && // Check for pathological object
509                 that->fSymbols       != NULL && // Check for pathological object
510                 *fSymbols            == *that->fSymbols &&
511                 fHaveDefaultCentury  == that->fHaveDefaultCentury &&
512                 fDefaultCenturyStart == that->fDefaultCenturyStart &&
513                 fCapitalizationContext == that->fCapitalizationContext);
514     }
515     return FALSE;
516 }
517 
518 //----------------------------------------------------------------------
519 
construct(EStyle timeStyle,EStyle dateStyle,const Locale & locale,UErrorCode & status)520 void SimpleDateFormat::construct(EStyle timeStyle,
521                                  EStyle dateStyle,
522                                  const Locale& locale,
523                                  UErrorCode& status)
524 {
525     // called by several constructors to load pattern data from the resources
526     if (U_FAILURE(status)) return;
527 
528     // We will need the calendar to know what type of symbols to load.
529     initializeCalendar(NULL, locale, status);
530     if (U_FAILURE(status)) return;
531 
532     CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
533     UResourceBundle *dateTimePatterns = calData.getByKey(gDateTimePatternsTag, status);
534     UResourceBundle *currentBundle;
535 
536     if (U_FAILURE(status)) return;
537 
538     if (ures_getSize(dateTimePatterns) <= kDateTime)
539     {
540         status = U_INVALID_FORMAT_ERROR;
541         return;
542     }
543 
544     setLocaleIDs(ures_getLocaleByType(dateTimePatterns, ULOC_VALID_LOCALE, &status),
545                  ures_getLocaleByType(dateTimePatterns, ULOC_ACTUAL_LOCALE, &status));
546 
547     // create a symbols object from the locale
548     initializeSymbols(locale,fCalendar, status);
549     if (U_FAILURE(status)) return;
550     /* test for NULL */
551     if (fSymbols == 0) {
552         status = U_MEMORY_ALLOCATION_ERROR;
553         return;
554     }
555 
556     const UChar *resStr,*ovrStr;
557     int32_t resStrLen,ovrStrLen = 0;
558     fDateOverride.setToBogus();
559     fTimeOverride.setToBogus();
560 
561     // if the pattern should include both date and time information, use the date/time
562     // pattern string as a guide to tell use how to glue together the appropriate date
563     // and time pattern strings.  The actual gluing-together is handled by a convenience
564     // method on MessageFormat.
565     if ((timeStyle != kNone) && (dateStyle != kNone))
566     {
567         Formattable timeDateArray[2];
568 
569         // use Formattable::adoptString() so that we can use fastCopyFrom()
570         // instead of Formattable::setString()'s unaware, safe, deep string clone
571         // see Jitterbug 2296
572 
573         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
574         if (U_FAILURE(status)) {
575            status = U_INVALID_FORMAT_ERROR;
576            return;
577         }
578         switch (ures_getType(currentBundle)) {
579             case URES_STRING: {
580                resStr = ures_getString(currentBundle, &resStrLen, &status);
581                break;
582             }
583             case URES_ARRAY: {
584                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
585                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
586                fTimeOverride.setTo(TRUE, ovrStr, ovrStrLen);
587                break;
588             }
589             default: {
590                status = U_INVALID_FORMAT_ERROR;
591                ures_close(currentBundle);
592                return;
593             }
594         }
595         ures_close(currentBundle);
596 
597         UnicodeString *tempus1 = new UnicodeString(TRUE, resStr, resStrLen);
598         // NULL pointer check
599         if (tempus1 == NULL) {
600             status = U_MEMORY_ALLOCATION_ERROR;
601             return;
602         }
603         timeDateArray[0].adoptString(tempus1);
604 
605         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)dateStyle, NULL, &status);
606         if (U_FAILURE(status)) {
607            status = U_INVALID_FORMAT_ERROR;
608            return;
609         }
610         switch (ures_getType(currentBundle)) {
611             case URES_STRING: {
612                resStr = ures_getString(currentBundle, &resStrLen, &status);
613                break;
614             }
615             case URES_ARRAY: {
616                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
617                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
618                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
619                break;
620             }
621             default: {
622                status = U_INVALID_FORMAT_ERROR;
623                ures_close(currentBundle);
624                return;
625             }
626         }
627         ures_close(currentBundle);
628 
629         UnicodeString *tempus2 = new UnicodeString(TRUE, resStr, resStrLen);
630         // Null pointer check
631         if (tempus2 == NULL) {
632             status = U_MEMORY_ALLOCATION_ERROR;
633             return;
634         }
635         timeDateArray[1].adoptString(tempus2);
636 
637         int32_t glueIndex = kDateTime;
638         int32_t patternsSize = ures_getSize(dateTimePatterns);
639         if (patternsSize >= (kDateTimeOffset + kShort + 1)) {
640             // Get proper date time format
641             glueIndex = (int32_t)(kDateTimeOffset + (dateStyle - kDateOffset));
642         }
643 
644         resStr = ures_getStringByIndex(dateTimePatterns, glueIndex, &resStrLen, &status);
645         MessageFormat::format(UnicodeString(TRUE, resStr, resStrLen), timeDateArray, 2, fPattern, status);
646     }
647     // if the pattern includes just time data or just date date, load the appropriate
648     // pattern string from the resources
649     // setTo() - see DateFormatSymbols::assignArray comments
650     else if (timeStyle != kNone) {
651         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
652         if (U_FAILURE(status)) {
653            status = U_INVALID_FORMAT_ERROR;
654            return;
655         }
656         switch (ures_getType(currentBundle)) {
657             case URES_STRING: {
658                resStr = ures_getString(currentBundle, &resStrLen, &status);
659                break;
660             }
661             case URES_ARRAY: {
662                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
663                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
664                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
665                break;
666             }
667             default: {
668                status = U_INVALID_FORMAT_ERROR;
669                 ures_close(currentBundle);
670                return;
671             }
672         }
673         fPattern.setTo(TRUE, resStr, resStrLen);
674         ures_close(currentBundle);
675     }
676     else if (dateStyle != kNone) {
677         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)dateStyle, NULL, &status);
678         if (U_FAILURE(status)) {
679            status = U_INVALID_FORMAT_ERROR;
680            return;
681         }
682         switch (ures_getType(currentBundle)) {
683             case URES_STRING: {
684                resStr = ures_getString(currentBundle, &resStrLen, &status);
685                break;
686             }
687             case URES_ARRAY: {
688                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
689                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
690                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
691                break;
692             }
693             default: {
694                status = U_INVALID_FORMAT_ERROR;
695                ures_close(currentBundle);
696                return;
697             }
698         }
699         fPattern.setTo(TRUE, resStr, resStrLen);
700         ures_close(currentBundle);
701     }
702 
703     // and if it includes _neither_, that's an error
704     else
705         status = U_INVALID_FORMAT_ERROR;
706 
707     // finally, finish initializing by creating a Calendar and a NumberFormat
708     initialize(locale, status);
709 }
710 
711 //----------------------------------------------------------------------
712 
713 Calendar*
initializeCalendar(TimeZone * adoptZone,const Locale & locale,UErrorCode & status)714 SimpleDateFormat::initializeCalendar(TimeZone* adoptZone, const Locale& locale, UErrorCode& status)
715 {
716     if(!U_FAILURE(status)) {
717         fCalendar = Calendar::createInstance(adoptZone?adoptZone:TimeZone::createDefault(), locale, status);
718     }
719     if (U_SUCCESS(status) && fCalendar == NULL) {
720         status = U_MEMORY_ALLOCATION_ERROR;
721     }
722     return fCalendar;
723 }
724 
725 void
initializeSymbols(const Locale & locale,Calendar * calendar,UErrorCode & status)726 SimpleDateFormat::initializeSymbols(const Locale& locale, Calendar* calendar, UErrorCode& status)
727 {
728   if(U_FAILURE(status)) {
729     fSymbols = NULL;
730   } else {
731     // pass in calendar type - use NULL (default) if no calendar set (or err).
732     fSymbols = new DateFormatSymbols(locale, calendar?calendar->getType() :NULL , status);
733     // Null pointer check
734     if (fSymbols == NULL) {
735         status = U_MEMORY_ALLOCATION_ERROR;
736         return;
737     }
738   }
739 }
740 
741 void
initialize(const Locale & locale,UErrorCode & status)742 SimpleDateFormat::initialize(const Locale& locale,
743                              UErrorCode& status)
744 {
745     if (U_FAILURE(status)) return;
746 
747     // We don't need to check that the row count is >= 1, since all 2d arrays have at
748     // least one row
749     fNumberFormat = NumberFormat::createInstance(locale, status);
750     if (fNumberFormat != NULL && U_SUCCESS(status))
751     {
752         // no matter what the locale's default number format looked like, we want
753         // to modify it so that it doesn't use thousands separators, doesn't always
754         // show the decimal point, and recognizes integers only when parsing
755 
756         fNumberFormat->setGroupingUsed(FALSE);
757         DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(fNumberFormat);
758         if (decfmt != NULL) {
759             decfmt->setDecimalSeparatorAlwaysShown(FALSE);
760         }
761         fNumberFormat->setParseIntegerOnly(TRUE);
762         fNumberFormat->setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
763 
764         //fNumberFormat->setLenient(TRUE); // Java uses a custom DateNumberFormat to format/parse
765 
766         initNumberFormatters(locale,status);
767 
768     }
769     else if (U_SUCCESS(status))
770     {
771         status = U_MISSING_RESOURCE_ERROR;
772     }
773 }
774 
775 /* Initialize the fields we use to disambiguate ambiguous years. Separate
776  * so we can call it from readObject().
777  */
initializeDefaultCentury()778 void SimpleDateFormat::initializeDefaultCentury()
779 {
780   if(fCalendar) {
781     fHaveDefaultCentury = fCalendar->haveDefaultCentury();
782     if(fHaveDefaultCentury) {
783       fDefaultCenturyStart = fCalendar->defaultCenturyStart();
784       fDefaultCenturyStartYear = fCalendar->defaultCenturyStartYear();
785     } else {
786       fDefaultCenturyStart = DBL_MIN;
787       fDefaultCenturyStartYear = -1;
788     }
789   }
790 }
791 
792 /* Define one-century window into which to disambiguate dates using
793  * two-digit years. Make public in JDK 1.2.
794  */
parseAmbiguousDatesAsAfter(UDate startDate,UErrorCode & status)795 void SimpleDateFormat::parseAmbiguousDatesAsAfter(UDate startDate, UErrorCode& status)
796 {
797     if(U_FAILURE(status)) {
798         return;
799     }
800     if(!fCalendar) {
801       status = U_ILLEGAL_ARGUMENT_ERROR;
802       return;
803     }
804 
805     fCalendar->setTime(startDate, status);
806     if(U_SUCCESS(status)) {
807         fHaveDefaultCentury = TRUE;
808         fDefaultCenturyStart = startDate;
809         fDefaultCenturyStartYear = fCalendar->get(UCAL_YEAR, status);
810     }
811 }
812 
813 //----------------------------------------------------------------------
814 
815 UnicodeString&
format(Calendar & cal,UnicodeString & appendTo,FieldPosition & pos) const816 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo, FieldPosition& pos) const
817 {
818   UErrorCode status = U_ZERO_ERROR;
819   FieldPositionOnlyHandler handler(pos);
820   return _format(cal, appendTo, handler, status);
821 }
822 
823 //----------------------------------------------------------------------
824 
825 UnicodeString&
format(Calendar & cal,UnicodeString & appendTo,FieldPositionIterator * posIter,UErrorCode & status) const826 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo,
827                          FieldPositionIterator* posIter, UErrorCode& status) const
828 {
829   FieldPositionIteratorHandler handler(posIter, status);
830   return _format(cal, appendTo, handler, status);
831 }
832 
833 //----------------------------------------------------------------------
834 
835 UnicodeString&
_format(Calendar & cal,UnicodeString & appendTo,FieldPositionHandler & handler,UErrorCode & status) const836 SimpleDateFormat::_format(Calendar& cal, UnicodeString& appendTo,
837                             FieldPositionHandler& handler, UErrorCode& status) const
838 {
839     if ( U_FAILURE(status) ) {
840        return appendTo;
841     }
842     Calendar* workCal = &cal;
843     Calendar* calClone = NULL;
844     if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
845         // Different calendar type
846         // We use the time and time zone from the input calendar, but
847         // do not use the input calendar for field calculation.
848         calClone = fCalendar->clone();
849         if (calClone != NULL) {
850             UDate t = cal.getTime(status);
851             calClone->setTime(t, status);
852             calClone->setTimeZone(cal.getTimeZone());
853             workCal = calClone;
854         } else {
855             status = U_MEMORY_ALLOCATION_ERROR;
856             return appendTo;
857         }
858     }
859 
860     UBool inQuote = FALSE;
861     UChar prevCh = 0;
862     int32_t count = 0;
863     int32_t fieldNum = 0;
864 
865     // loop through the pattern string character by character
866     for (int32_t i = 0; i < fPattern.length() && U_SUCCESS(status); ++i) {
867         UChar ch = fPattern[i];
868 
869         // Use subFormat() to format a repeated pattern character
870         // when a different pattern or non-pattern character is seen
871         if (ch != prevCh && count > 0) {
872             subFormat(appendTo, prevCh, count, fCapitalizationContext, fieldNum++, handler, *workCal, status);
873             count = 0;
874         }
875         if (ch == QUOTE) {
876             // Consecutive single quotes are a single quote literal,
877             // either outside of quotes or between quotes
878             if ((i+1) < fPattern.length() && fPattern[i+1] == QUOTE) {
879                 appendTo += (UChar)QUOTE;
880                 ++i;
881             } else {
882                 inQuote = ! inQuote;
883             }
884         }
885         else if ( ! inQuote && ((ch >= 0x0061 /*'a'*/ && ch <= 0x007A /*'z'*/)
886                     || (ch >= 0x0041 /*'A'*/ && ch <= 0x005A /*'Z'*/))) {
887             // ch is a date-time pattern character to be interpreted
888             // by subFormat(); count the number of times it is repeated
889             prevCh = ch;
890             ++count;
891         }
892         else {
893             // Append quoted characters and unquoted non-pattern characters
894             appendTo += ch;
895         }
896     }
897 
898     // Format the last item in the pattern, if any
899     if (count > 0) {
900         subFormat(appendTo, prevCh, count, fCapitalizationContext, fieldNum++, handler, *workCal, status);
901     }
902 
903     if (calClone != NULL) {
904         delete calClone;
905     }
906 
907     return appendTo;
908 }
909 
910 //----------------------------------------------------------------------
911 
912 /* Map calendar field into calendar field level.
913  * the larger the level, the smaller the field unit.
914  * For example, UCAL_ERA level is 0, UCAL_YEAR level is 10,
915  * UCAL_MONTH level is 20.
916  * NOTE: if new fields adds in, the table needs to update.
917  */
918 const int32_t
919 SimpleDateFormat::fgCalendarFieldToLevel[] =
920 {
921     /*GyM*/ 0, 10, 20,
922     /*wW*/ 20, 30,
923     /*dDEF*/ 30, 20, 30, 30,
924     /*ahHm*/ 40, 50, 50, 60,
925     /*sS..*/ 70, 80,
926     /*z?Y*/ 0, 0, 10,
927     /*eug*/ 30, 10, 0,
928     /*A*/ 40
929 };
930 
931 
932 /* Map calendar field LETTER into calendar field level.
933  * the larger the level, the smaller the field unit.
934  * NOTE: if new fields adds in, the table needs to update.
935  */
936 const int32_t
937 SimpleDateFormat::fgPatternCharToLevel[] = {
938     //       A   B   C   D   E   F   G   H   I   J   K   L   M   N   O
939         -1, 40, -1, -1, 20, 30, 30,  0, 50, -1, -1, 50, 20, 20, -1, -1,
940     //   P   Q   R   S   T   U   V   W   X   Y   Z
941         -1, 20, -1, 80, -1, 10,  0, 30, -1, 10,  0, -1, -1, -1, -1, -1,
942     //       a   b   c   d   e   f   g   h   i   j   k   l   m   n   o
943         -1, 40, -1, 30, 30, 30, -1,  0, 50, -1, -1, 50, -1, 60, -1, -1,
944     //   p   q   r   s   t   u   v   w   x   y   z
945         -1, 20, -1, 70, -1, 10,  0, 20, -1, 10,  0, -1, -1, -1, -1, -1
946 };
947 
948 
949 // Map index into pattern character string to Calendar field number.
950 const UCalendarDateFields
951 SimpleDateFormat::fgPatternIndexToCalendarField[] =
952 {
953     /*GyM*/ UCAL_ERA, UCAL_YEAR, UCAL_MONTH,
954     /*dkH*/ UCAL_DATE, UCAL_HOUR_OF_DAY, UCAL_HOUR_OF_DAY,
955     /*msS*/ UCAL_MINUTE, UCAL_SECOND, UCAL_MILLISECOND,
956     /*EDF*/ UCAL_DAY_OF_WEEK, UCAL_DAY_OF_YEAR, UCAL_DAY_OF_WEEK_IN_MONTH,
957     /*wWa*/ UCAL_WEEK_OF_YEAR, UCAL_WEEK_OF_MONTH, UCAL_AM_PM,
958     /*hKz*/ UCAL_HOUR, UCAL_HOUR, UCAL_ZONE_OFFSET,
959     /*Yeu*/ UCAL_YEAR_WOY, UCAL_DOW_LOCAL, UCAL_EXTENDED_YEAR,
960     /*gAZ*/ UCAL_JULIAN_DAY, UCAL_MILLISECONDS_IN_DAY, UCAL_ZONE_OFFSET,
961     /*v*/   UCAL_ZONE_OFFSET,
962     /*c*/   UCAL_DOW_LOCAL,
963     /*L*/   UCAL_MONTH,
964     /*Q*/   UCAL_MONTH,
965     /*q*/   UCAL_MONTH,
966     /*V*/   UCAL_ZONE_OFFSET,
967     /*U*/   UCAL_YEAR,
968 };
969 
970 // Map index into pattern character string to DateFormat field number
971 const UDateFormatField
972 SimpleDateFormat::fgPatternIndexToDateFormatField[] = {
973     /*GyM*/ UDAT_ERA_FIELD, UDAT_YEAR_FIELD, UDAT_MONTH_FIELD,
974     /*dkH*/ UDAT_DATE_FIELD, UDAT_HOUR_OF_DAY1_FIELD, UDAT_HOUR_OF_DAY0_FIELD,
975     /*msS*/ UDAT_MINUTE_FIELD, UDAT_SECOND_FIELD, UDAT_FRACTIONAL_SECOND_FIELD,
976     /*EDF*/ UDAT_DAY_OF_WEEK_FIELD, UDAT_DAY_OF_YEAR_FIELD, UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
977     /*wWa*/ UDAT_WEEK_OF_YEAR_FIELD, UDAT_WEEK_OF_MONTH_FIELD, UDAT_AM_PM_FIELD,
978     /*hKz*/ UDAT_HOUR1_FIELD, UDAT_HOUR0_FIELD, UDAT_TIMEZONE_FIELD,
979     /*Yeu*/ UDAT_YEAR_WOY_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_EXTENDED_YEAR_FIELD,
980     /*gAZ*/ UDAT_JULIAN_DAY_FIELD, UDAT_MILLISECONDS_IN_DAY_FIELD, UDAT_TIMEZONE_RFC_FIELD,
981     /*v*/   UDAT_TIMEZONE_GENERIC_FIELD,
982     /*c*/   UDAT_STANDALONE_DAY_FIELD,
983     /*L*/   UDAT_STANDALONE_MONTH_FIELD,
984     /*Q*/   UDAT_QUARTER_FIELD,
985     /*q*/   UDAT_STANDALONE_QUARTER_FIELD,
986     /*V*/   UDAT_TIMEZONE_SPECIAL_FIELD,
987     /*U*/   UDAT_YEAR_NAME_FIELD,
988 };
989 
990 //----------------------------------------------------------------------
991 
992 /**
993  * Append symbols[value] to dst.  Make sure the array index is not out
994  * of bounds.
995  */
996 static inline void
_appendSymbol(UnicodeString & dst,int32_t value,const UnicodeString * symbols,int32_t symbolsCount)997 _appendSymbol(UnicodeString& dst,
998               int32_t value,
999               const UnicodeString* symbols,
1000               int32_t symbolsCount) {
1001     U_ASSERT(0 <= value && value < symbolsCount);
1002     if (0 <= value && value < symbolsCount) {
1003         dst += symbols[value];
1004     }
1005 }
1006 
1007 static inline void
_appendSymbolWithMonthPattern(UnicodeString & dst,int32_t value,const UnicodeString * symbols,int32_t symbolsCount,const UnicodeString * monthPattern,UErrorCode & status)1008 _appendSymbolWithMonthPattern(UnicodeString& dst, int32_t value, const UnicodeString* symbols, int32_t symbolsCount,
1009               const UnicodeString* monthPattern, UErrorCode& status) {
1010     U_ASSERT(0 <= value && value < symbolsCount);
1011     if (0 <= value && value < symbolsCount) {
1012         if (monthPattern == NULL) {
1013             dst += symbols[value];
1014         } else {
1015             Formattable monthName((const UnicodeString&)(symbols[value]));
1016             MessageFormat::format(*monthPattern, &monthName, 1, dst, status);
1017         }
1018     }
1019 }
1020 
1021 //----------------------------------------------------------------------
1022 void
initNumberFormatters(const Locale & locale,UErrorCode & status)1023 SimpleDateFormat::initNumberFormatters(const Locale &locale,UErrorCode &status) {
1024     if (U_FAILURE(status)) {
1025         return;
1026     }
1027     if ( fDateOverride.isBogus() && fTimeOverride.isBogus() ) {
1028         return;
1029     }
1030     umtx_lock(&LOCK);
1031     if (fNumberFormatters == NULL) {
1032         fNumberFormatters = (NumberFormat**)uprv_malloc(UDAT_FIELD_COUNT * sizeof(NumberFormat*));
1033         if (fNumberFormatters) {
1034             for (int32_t i = 0; i < UDAT_FIELD_COUNT; i++) {
1035                 fNumberFormatters[i] = fNumberFormat;
1036             }
1037         } else {
1038             status = U_MEMORY_ALLOCATION_ERROR;
1039         }
1040     }
1041     umtx_unlock(&LOCK);
1042 
1043     processOverrideString(locale,fDateOverride,kOvrStrDate,status);
1044     processOverrideString(locale,fTimeOverride,kOvrStrTime,status);
1045 
1046 }
1047 
1048 void
processOverrideString(const Locale & locale,const UnicodeString & str,int8_t type,UErrorCode & status)1049 SimpleDateFormat::processOverrideString(const Locale &locale, const UnicodeString &str, int8_t type, UErrorCode &status) {
1050     if (str.isBogus()) {
1051         return;
1052     }
1053     int32_t start = 0;
1054     int32_t len;
1055     UnicodeString nsName;
1056     UnicodeString ovrField;
1057     UBool moreToProcess = TRUE;
1058 
1059     while (moreToProcess) {
1060         int32_t delimiterPosition = str.indexOf((UChar)ULOC_KEYWORD_ITEM_SEPARATOR_UNICODE,start);
1061         if (delimiterPosition == -1) {
1062             moreToProcess = FALSE;
1063             len = str.length() - start;
1064         } else {
1065             len = delimiterPosition - start;
1066         }
1067         UnicodeString currentString(str,start,len);
1068         int32_t equalSignPosition = currentString.indexOf((UChar)ULOC_KEYWORD_ASSIGN_UNICODE,0);
1069         if (equalSignPosition == -1) { // Simple override string such as "hebrew"
1070             nsName.setTo(currentString);
1071             ovrField.setToBogus();
1072         } else { // Field specific override string such as "y=hebrew"
1073             nsName.setTo(currentString,equalSignPosition+1);
1074             ovrField.setTo(currentString,0,1); // We just need the first character.
1075         }
1076 
1077         int32_t nsNameHash = nsName.hashCode();
1078         // See if the numbering system is in the override list, if not, then add it.
1079         NSOverride *cur = fOverrideList;
1080         NumberFormat *nf = NULL;
1081         UBool found = FALSE;
1082         while ( cur && !found ) {
1083             if ( cur->hash == nsNameHash ) {
1084                 nf = cur->nf;
1085                 found = TRUE;
1086             }
1087             cur = cur->next;
1088         }
1089 
1090         if (!found) {
1091            cur = (NSOverride *)uprv_malloc(sizeof(NSOverride));
1092            if (cur) {
1093                char kw[ULOC_KEYWORD_AND_VALUES_CAPACITY];
1094                uprv_strcpy(kw,"numbers=");
1095                nsName.extract(0,len,kw+8,ULOC_KEYWORD_AND_VALUES_CAPACITY-8,US_INV);
1096 
1097                Locale ovrLoc(locale.getLanguage(),locale.getCountry(),locale.getVariant(),kw);
1098                nf = NumberFormat::createInstance(ovrLoc,status);
1099 
1100                // no matter what the locale's default number format looked like, we want
1101                // to modify it so that it doesn't use thousands separators, doesn't always
1102                // show the decimal point, and recognizes integers only when parsing
1103 
1104                if (U_SUCCESS(status)) {
1105                    nf->setGroupingUsed(FALSE);
1106                    DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(nf);
1107                    if (decfmt != NULL) {
1108                        decfmt->setDecimalSeparatorAlwaysShown(FALSE);
1109                    }
1110                    nf->setParseIntegerOnly(TRUE);
1111                    nf->setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
1112 
1113                    cur->nf = nf;
1114                    cur->hash = nsNameHash;
1115                    cur->next = fOverrideList;
1116                    fOverrideList = cur;
1117                }
1118                else {
1119                    // clean up before returning
1120                    if (cur != NULL) {
1121                        uprv_free(cur);
1122                    }
1123                   return;
1124                }
1125 
1126            } else {
1127                status = U_MEMORY_ALLOCATION_ERROR;
1128                return;
1129            }
1130         }
1131 
1132         // Now that we have an appropriate number formatter, fill in the appropriate spaces in the
1133         // number formatters table.
1134 
1135         if (ovrField.isBogus()) {
1136             switch (type) {
1137                 case kOvrStrDate:
1138                 case kOvrStrBoth: {
1139                     for ( int8_t i=0 ; i<kDateFieldsCount; i++ ) {
1140                         fNumberFormatters[kDateFields[i]] = nf;
1141                     }
1142                     if (type==kOvrStrDate) {
1143                         break;
1144                     }
1145                 }
1146                 case kOvrStrTime : {
1147                     for ( int8_t i=0 ; i<kTimeFieldsCount; i++ ) {
1148                         fNumberFormatters[kTimeFields[i]] = nf;
1149                     }
1150                     break;
1151                 }
1152             }
1153         } else {
1154            // if the pattern character is unrecognized, signal an error and bail out
1155            UDateFormatField patternCharIndex =
1156               DateFormatSymbols::getPatternCharIndex(ovrField.charAt(0));
1157            if (patternCharIndex == UDAT_FIELD_COUNT) {
1158                status = U_INVALID_FORMAT_ERROR;
1159                return;
1160            }
1161 
1162            // Set the number formatter in the table
1163            fNumberFormatters[patternCharIndex] = nf;
1164         }
1165 
1166         start = delimiterPosition + 1;
1167     }
1168 }
1169 
1170 //---------------------------------------------------------------------
1171 void
subFormat(UnicodeString & appendTo,UChar ch,int32_t count,UDisplayContext capitalizationContext,int32_t fieldNum,FieldPositionHandler & handler,Calendar & cal,UErrorCode & status) const1172 SimpleDateFormat::subFormat(UnicodeString &appendTo,
1173                             UChar ch,
1174                             int32_t count,
1175                             UDisplayContext capitalizationContext,
1176                             int32_t fieldNum,
1177                             FieldPositionHandler& handler,
1178                             Calendar& cal,
1179                             UErrorCode& status) const
1180 {
1181     if (U_FAILURE(status)) {
1182         return;
1183     }
1184 
1185     // this function gets called by format() to produce the appropriate substitution
1186     // text for an individual pattern symbol (e.g., "HH" or "yyyy")
1187 
1188     UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
1189     const int32_t maxIntCount = 10;
1190     int32_t beginOffset = appendTo.length();
1191     NumberFormat *currentNumberFormat;
1192     DateFormatSymbols::ECapitalizationContextUsageType capContextUsageType = DateFormatSymbols::kCapContextUsageOther;
1193 
1194     UBool isHebrewCalendar = (uprv_strcmp(cal.getType(),"hebrew") == 0);
1195     UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0);
1196 
1197     // if the pattern character is unrecognized, signal an error and dump out
1198     if (patternCharIndex == UDAT_FIELD_COUNT)
1199     {
1200         if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
1201             status = U_INVALID_FORMAT_ERROR;
1202         }
1203         return;
1204     }
1205 
1206     UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex];
1207     int32_t value = cal.get(field, status);
1208     if (U_FAILURE(status)) {
1209         return;
1210     }
1211 
1212     currentNumberFormat = getNumberFormatByIndex(patternCharIndex);
1213     UnicodeString hebr("hebr", 4, US_INV);
1214 
1215     switch (patternCharIndex) {
1216 
1217     // for any "G" symbol, write out the appropriate era string
1218     // "GGGG" is wide era name, "GGGGG" is narrow era name, anything else is abbreviated name
1219     case UDAT_ERA_FIELD:
1220         if (isChineseCalendar) {
1221             zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, 9); // as in ICU4J
1222         } else {
1223             if (count == 5) {
1224                 _appendSymbol(appendTo, value, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount);
1225                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraNarrow;
1226             } else if (count == 4) {
1227                 _appendSymbol(appendTo, value, fSymbols->fEraNames, fSymbols->fEraNamesCount);
1228                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraWide;
1229             } else {
1230                 _appendSymbol(appendTo, value, fSymbols->fEras, fSymbols->fErasCount);
1231                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraAbbrev;
1232             }
1233         }
1234         break;
1235 
1236      case UDAT_YEAR_NAME_FIELD:
1237         if (fSymbols->fShortYearNames != NULL && value <= fSymbols->fShortYearNamesCount) {
1238             // the Calendar YEAR field runs 1 through 60 for cyclic years
1239             _appendSymbol(appendTo, value - 1, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount);
1240             break;
1241         }
1242         // else fall through to numeric year handling, do not break here
1243 
1244    // OLD: for "yyyy", write out the whole year; for "yy", write out the last 2 digits
1245     // NEW: UTS#35:
1246 //Year         y     yy     yyy     yyyy     yyyyy
1247 //AD 1         1     01     001     0001     00001
1248 //AD 12       12     12     012     0012     00012
1249 //AD 123     123     23     123     0123     00123
1250 //AD 1234   1234     34    1234     1234     01234
1251 //AD 12345 12345     45   12345    12345     12345
1252     case UDAT_YEAR_FIELD:
1253     case UDAT_YEAR_WOY_FIELD:
1254         if (fDateOverride.compare(hebr)==0 && value>HEBREW_CAL_CUR_MILLENIUM_START_YEAR && value<HEBREW_CAL_CUR_MILLENIUM_END_YEAR) {
1255             value-=HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
1256         }
1257         if(count == 2)
1258             zeroPaddingNumber(currentNumberFormat, appendTo, value, 2, 2);
1259         else
1260             zeroPaddingNumber(currentNumberFormat, appendTo, value, count, maxIntCount);
1261         break;
1262 
1263     // for "MMMM"/"LLLL", write out the whole month name, for "MMM"/"LLL", write out the month
1264     // abbreviation, for "M"/"L" or "MM"/"LL", write out the month as a number with the
1265     // appropriate number of digits
1266     // for "MMMMM"/"LLLLL", use the narrow form
1267     case UDAT_MONTH_FIELD:
1268     case UDAT_STANDALONE_MONTH_FIELD:
1269         if ( isHebrewCalendar ) {
1270            HebrewCalendar *hc = (HebrewCalendar*)&cal;
1271            if (hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value == 6 && count >= 3 )
1272                value = 13; // Show alternate form for Adar II in leap years in Hebrew calendar.
1273            if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6 && count < 3 )
1274                value--; // Adjust the month number down 1 in Hebrew non-leap years, i.e. Adar is 6, not 7.
1275         }
1276         {
1277             int32_t isLeapMonth = (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount)?
1278                         cal.get(UCAL_IS_LEAP_MONTH, status): 0;
1279             // should consolidate the next section by using arrays of pointers & counts for the right symbols...
1280             if (count == 5) {
1281                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1282                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fNarrowMonths, fSymbols->fNarrowMonthsCount,
1283                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatNarrow]): NULL, status);
1284                 } else {
1285                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneNarrowMonths, fSymbols->fStandaloneNarrowMonthsCount,
1286                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneNarrow]): NULL, status);
1287                 }
1288                 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthNarrow;
1289             } else if (count == 4) {
1290                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1291                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fMonths, fSymbols->fMonthsCount,
1292                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide]): NULL, status);
1293                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1294                 } else {
1295                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount,
1296                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide]): NULL, status);
1297                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1298                 }
1299             } else if (count == 3) {
1300                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1301                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fShortMonths, fSymbols->fShortMonthsCount,
1302                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev]): NULL, status);
1303                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1304                 } else {
1305                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount,
1306                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev]): NULL, status);
1307                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1308                 }
1309             } else {
1310                 UnicodeString monthNumber;
1311                 zeroPaddingNumber(currentNumberFormat,monthNumber, value + 1, count, maxIntCount);
1312                 _appendSymbolWithMonthPattern(appendTo, 0, &monthNumber, 1,
1313                         (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric]): NULL, status);
1314             }
1315         }
1316         break;
1317 
1318     // for "k" and "kk", write out the hour, adjusting midnight to appear as "24"
1319     case UDAT_HOUR_OF_DAY1_FIELD:
1320         if (value == 0)
1321             zeroPaddingNumber(currentNumberFormat,appendTo, cal.getMaximum(UCAL_HOUR_OF_DAY) + 1, count, maxIntCount);
1322         else
1323             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1324         break;
1325 
1326     case UDAT_FRACTIONAL_SECOND_FIELD:
1327         // Fractional seconds left-justify
1328         {
1329             currentNumberFormat->setMinimumIntegerDigits((count > 3) ? 3 : count);
1330             currentNumberFormat->setMaximumIntegerDigits(maxIntCount);
1331             if (count == 1) {
1332                 value /= 100;
1333             } else if (count == 2) {
1334                 value /= 10;
1335             }
1336             FieldPosition p(0);
1337             currentNumberFormat->format(value, appendTo, p);
1338             if (count > 3) {
1339                 currentNumberFormat->setMinimumIntegerDigits(count - 3);
1340                 currentNumberFormat->format((int32_t)0, appendTo, p);
1341             }
1342         }
1343         break;
1344 
1345     // for "ee" or "e", use local numeric day-of-the-week
1346     // for "EEEEE" or "eeeee", write out the narrow day-of-the-week name
1347     // for "EEEE" or "eeee", write out the wide day-of-the-week name
1348     // for "EEE" or "EE" or "E" or "eee", write out the abbreviated day-of-the-week name
1349     case UDAT_DOW_LOCAL_FIELD:
1350         if ( count < 3 ) {
1351             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1352             break;
1353         }
1354         // fall through to EEEEE-EEE handling, but for that we don't want local day-of-week,
1355         // we want standard day-of-week, so first fix value to work for EEEEE-EEE.
1356         value = cal.get(UCAL_DAY_OF_WEEK, status);
1357         if (U_FAILURE(status)) {
1358             return;
1359         }
1360         // fall through, do not break here
1361     case UDAT_DAY_OF_WEEK_FIELD:
1362         if (count == 5) {
1363             _appendSymbol(appendTo, value, fSymbols->fNarrowWeekdays,
1364                           fSymbols->fNarrowWeekdaysCount);
1365             capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1366         } else if (count == 4) {
1367             _appendSymbol(appendTo, value, fSymbols->fWeekdays,
1368                           fSymbols->fWeekdaysCount);
1369             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1370         } else {
1371             _appendSymbol(appendTo, value, fSymbols->fShortWeekdays,
1372                           fSymbols->fShortWeekdaysCount);
1373             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1374         }
1375         break;
1376 
1377     // for "ccc", write out the abbreviated day-of-the-week name
1378     // for "cccc", write out the wide day-of-the-week name
1379     // for "ccccc", use the narrow day-of-the-week name
1380     case UDAT_STANDALONE_DAY_FIELD:
1381         if ( count < 3 ) {
1382             zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, maxIntCount);
1383             break;
1384         }
1385         // fall through to alpha DOW handling, but for that we don't want local day-of-week,
1386         // we want standard day-of-week, so first fix value.
1387         value = cal.get(UCAL_DAY_OF_WEEK, status);
1388         if (U_FAILURE(status)) {
1389             return;
1390         }
1391         if (count == 5) {
1392             _appendSymbol(appendTo, value, fSymbols->fStandaloneNarrowWeekdays,
1393                           fSymbols->fStandaloneNarrowWeekdaysCount);
1394             capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1395         } else if (count == 4) {
1396             _appendSymbol(appendTo, value, fSymbols->fStandaloneWeekdays,
1397                           fSymbols->fStandaloneWeekdaysCount);
1398             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1399         } else { // count == 3
1400             _appendSymbol(appendTo, value, fSymbols->fStandaloneShortWeekdays,
1401                           fSymbols->fStandaloneShortWeekdaysCount);
1402             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1403         }
1404         break;
1405 
1406     // for and "a" symbol, write out the whole AM/PM string
1407     case UDAT_AM_PM_FIELD:
1408         _appendSymbol(appendTo, value, fSymbols->fAmPms,
1409                       fSymbols->fAmPmsCount);
1410         break;
1411 
1412     // for "h" and "hh", write out the hour, adjusting noon and midnight to show up
1413     // as "12"
1414     case UDAT_HOUR1_FIELD:
1415         if (value == 0)
1416             zeroPaddingNumber(currentNumberFormat,appendTo, cal.getLeastMaximum(UCAL_HOUR) + 1, count, maxIntCount);
1417         else
1418             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1419         break;
1420 
1421     // for the "z" symbols, we have to check our time zone data first.  If we have a
1422     // localized name for the time zone, then "zzzz" / "zzz" indicate whether
1423     // daylight time is in effect (long/short) and "zz" / "z" do not (long/short).
1424     // If we don't have a localized time zone name,
1425     // then the time zone shows up as "GMT+hh:mm" or "GMT-hh:mm" (where "hh:mm" is the
1426     // offset from GMT) regardless of how many z's were in the pattern symbol
1427     case UDAT_TIMEZONE_FIELD:
1428     case UDAT_TIMEZONE_GENERIC_FIELD:
1429     case UDAT_TIMEZONE_SPECIAL_FIELD:
1430     case UDAT_TIMEZONE_RFC_FIELD: // 'Z' - TIMEZONE_RFC
1431         {
1432             UnicodeString zoneString;
1433             const TimeZone& tz = cal.getTimeZone();
1434             UDate date = cal.getTime(status);
1435             if (U_SUCCESS(status)) {
1436                 if (patternCharIndex == UDAT_TIMEZONE_RFC_FIELD) {
1437                     if (count < 4) {
1438                         // "Z"
1439                         tzFormat()->format(UTZFMT_STYLE_RFC822, tz, date, zoneString);
1440                     } else if (count == 5) {
1441                         // "ZZZZZ"
1442                         tzFormat()->format(UTZFMT_STYLE_ISO8601, tz, date, zoneString);
1443                     } else {
1444                         // "ZZ", "ZZZ", "ZZZZ"
1445                         tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
1446                     }
1447                 } else if (patternCharIndex == UDAT_TIMEZONE_FIELD) {
1448                     if (count < 4) {
1449                         // "z", "zz", "zzz"
1450                         tzFormat()->format(UTZFMT_STYLE_SPECIFIC_SHORT, tz, date, zoneString);
1451                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1452                     } else {
1453                         // "zzzz"
1454                         tzFormat()->format(UTZFMT_STYLE_SPECIFIC_LONG, tz, date, zoneString);
1455                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1456                     }
1457                 } else if (patternCharIndex == UDAT_TIMEZONE_GENERIC_FIELD) {
1458                     if (count == 1) {
1459                         // "v"
1460                         tzFormat()->format(UTZFMT_STYLE_GENERIC_SHORT, tz, date, zoneString);
1461                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1462                     } else if (count == 4) {
1463                         // "vvvv"
1464                         tzFormat()->format(UTZFMT_STYLE_GENERIC_LONG, tz, date, zoneString);
1465                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1466                     }
1467                 } else { // patternCharIndex == UDAT_TIMEZONE_SPECIAL_FIELD
1468                     if (count == 1) {
1469                         // "V"
1470                         tzFormat()->format(UTZFMT_STYLE_SPECIFIC_SHORT, tz, date, zoneString);
1471                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1472                     } else if (count == 4) {
1473                         // "VVVV"
1474                         tzFormat()->format(UTZFMT_STYLE_GENERIC_LOCATION, tz, date, zoneString);
1475                         capContextUsageType = DateFormatSymbols::kCapContextUsageZoneLong;
1476                     }
1477                 }
1478             }
1479             appendTo += zoneString;
1480         }
1481         break;
1482 
1483     case UDAT_QUARTER_FIELD:
1484         if (count >= 4)
1485             _appendSymbol(appendTo, value/3, fSymbols->fQuarters,
1486                           fSymbols->fQuartersCount);
1487         else if (count == 3)
1488             _appendSymbol(appendTo, value/3, fSymbols->fShortQuarters,
1489                           fSymbols->fShortQuartersCount);
1490         else
1491             zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1492         break;
1493 
1494     case UDAT_STANDALONE_QUARTER_FIELD:
1495         if (count >= 4)
1496             _appendSymbol(appendTo, value/3, fSymbols->fStandaloneQuarters,
1497                           fSymbols->fStandaloneQuartersCount);
1498         else if (count == 3)
1499             _appendSymbol(appendTo, value/3, fSymbols->fStandaloneShortQuarters,
1500                           fSymbols->fStandaloneShortQuartersCount);
1501         else
1502             zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1503         break;
1504 
1505 
1506     // all of the other pattern symbols can be formatted as simple numbers with
1507     // appropriate zero padding
1508     default:
1509         zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1510         break;
1511     }
1512 #if !UCONFIG_NO_BREAK_ITERATION
1513     if (fieldNum == 0) {
1514         // first field, check to see whether we need to titlecase it
1515         UBool titlecase = FALSE;
1516         switch (capitalizationContext) {
1517             case UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE:
1518                 titlecase = TRUE;
1519                 break;
1520             case UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU:
1521                 titlecase = fSymbols->fCapitalization[capContextUsageType][0];
1522                 break;
1523             case UDISPCTX_CAPITALIZATION_FOR_STANDALONE:
1524                 titlecase = fSymbols->fCapitalization[capContextUsageType][1];
1525                 break;
1526             default:
1527                 // titlecase = FALSE;
1528                 break;
1529         }
1530         if (titlecase) {
1531             UnicodeString firstField(appendTo, beginOffset);
1532             firstField.toTitle(NULL, fLocale, U_TITLECASE_NO_LOWERCASE | U_TITLECASE_NO_BREAK_ADJUSTMENT);
1533             appendTo.replaceBetween(beginOffset, appendTo.length(), firstField);
1534         }
1535     }
1536 #endif
1537 
1538     handler.addAttribute(fgPatternIndexToDateFormatField[patternCharIndex], beginOffset, appendTo.length());
1539 }
1540 
1541 //----------------------------------------------------------------------
1542 
1543 NumberFormat *
getNumberFormatByIndex(UDateFormatField index) const1544 SimpleDateFormat::getNumberFormatByIndex(UDateFormatField index) const {
1545     if (fNumberFormatters != NULL) {
1546         return fNumberFormatters[index];
1547     } else {
1548         return fNumberFormat;
1549     }
1550 }
1551 
1552 //----------------------------------------------------------------------
1553 void
zeroPaddingNumber(NumberFormat * currentNumberFormat,UnicodeString & appendTo,int32_t value,int32_t minDigits,int32_t maxDigits) const1554 SimpleDateFormat::zeroPaddingNumber(NumberFormat *currentNumberFormat,UnicodeString &appendTo,
1555                                     int32_t value, int32_t minDigits, int32_t maxDigits) const
1556 {
1557     if (currentNumberFormat!=NULL) {
1558         FieldPosition pos(0);
1559 
1560         currentNumberFormat->setMinimumIntegerDigits(minDigits);
1561         currentNumberFormat->setMaximumIntegerDigits(maxDigits);
1562         currentNumberFormat->format(value, appendTo, pos);  // 3rd arg is there to speed up processing
1563     }
1564 }
1565 
1566 //----------------------------------------------------------------------
1567 
1568 /**
1569  * Return true if the given format character, occuring count
1570  * times, represents a numeric field.
1571  */
isNumeric(UChar formatChar,int32_t count)1572 UBool SimpleDateFormat::isNumeric(UChar formatChar, int32_t count) {
1573     return DateFormatSymbols::isNumericPatternChar(formatChar, count);
1574 }
1575 
1576 UBool
isAtNumericField(const UnicodeString & pattern,int32_t patternOffset)1577 SimpleDateFormat::isAtNumericField(const UnicodeString &pattern, int32_t patternOffset) {
1578     if (patternOffset >= pattern.length()) {
1579         // not at any field
1580         return FALSE;
1581     }
1582     UChar ch = pattern.charAt(patternOffset);
1583     UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
1584     if (f == UDAT_FIELD_COUNT) {
1585         // not at any field
1586         return FALSE;
1587     }
1588     int32_t i = patternOffset;
1589     while (pattern.charAt(++i) == ch) {}
1590     return DateFormatSymbols::isNumericField(f, i - patternOffset);
1591 }
1592 
1593 UBool
isAfterNonNumericField(const UnicodeString & pattern,int32_t patternOffset)1594 SimpleDateFormat::isAfterNonNumericField(const UnicodeString &pattern, int32_t patternOffset) {
1595     if (patternOffset <= 0) {
1596         // not after any field
1597         return FALSE;
1598     }
1599     UChar ch = pattern.charAt(--patternOffset);
1600     UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
1601     if (f == UDAT_FIELD_COUNT) {
1602         // not after any field
1603         return FALSE;
1604     }
1605     int32_t i = patternOffset;
1606     while (pattern.charAt(--i) == ch) {}
1607     return !DateFormatSymbols::isNumericField(f, patternOffset - i);
1608 }
1609 
1610 void
parse(const UnicodeString & text,Calendar & cal,ParsePosition & parsePos) const1611 SimpleDateFormat::parse(const UnicodeString& text, Calendar& cal, ParsePosition& parsePos) const
1612 {
1613     UErrorCode status = U_ZERO_ERROR;
1614     int32_t pos = parsePos.getIndex();
1615     int32_t start = pos;
1616 
1617     UBool ambiguousYear[] = { FALSE };
1618     int32_t saveHebrewMonth = -1;
1619     int32_t count = 0;
1620 
1621     UBool lenient = isLenient();
1622 
1623     // hack, reset tztype, cast away const
1624     ((SimpleDateFormat*)this)->tztype = UTZFMT_TIME_TYPE_UNKNOWN;
1625 
1626     // For parsing abutting numeric fields. 'abutPat' is the
1627     // offset into 'pattern' of the first of 2 or more abutting
1628     // numeric fields.  'abutStart' is the offset into 'text'
1629     // where parsing the fields begins. 'abutPass' starts off as 0
1630     // and increments each time we try to parse the fields.
1631     int32_t abutPat = -1; // If >=0, we are in a run of abutting numeric fields
1632     int32_t abutStart = 0;
1633     int32_t abutPass = 0;
1634     UBool inQuote = FALSE;
1635 
1636     MessageFormat * numericLeapMonthFormatter = NULL;
1637 
1638     Calendar* calClone = NULL;
1639     Calendar *workCal = &cal;
1640     if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
1641         // Different calendar type
1642         // We use the time/zone from the input calendar, but
1643         // do not use the input calendar for field calculation.
1644         calClone = fCalendar->clone();
1645         if (calClone != NULL) {
1646             calClone->setTime(cal.getTime(status),status);
1647             if (U_FAILURE(status)) {
1648                 goto ExitParse;
1649             }
1650             calClone->setTimeZone(cal.getTimeZone());
1651             workCal = calClone;
1652         } else {
1653             status = U_MEMORY_ALLOCATION_ERROR;
1654             goto ExitParse;
1655         }
1656     }
1657 
1658     if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
1659         numericLeapMonthFormatter = new MessageFormat(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric], fLocale, status);
1660         if (numericLeapMonthFormatter == NULL) {
1661              status = U_MEMORY_ALLOCATION_ERROR;
1662              goto ExitParse;
1663         } else if (U_FAILURE(status)) {
1664              goto ExitParse; // this will delete numericLeapMonthFormatter
1665         }
1666     }
1667 
1668     for (int32_t i=0; i<fPattern.length(); ++i) {
1669         UChar ch = fPattern.charAt(i);
1670 
1671         // Handle alphabetic field characters.
1672         if (!inQuote && ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { // [A-Za-z]
1673             int32_t fieldPat = i;
1674 
1675             // Count the length of this field specifier
1676             count = 1;
1677             while ((i+1)<fPattern.length() &&
1678                    fPattern.charAt(i+1) == ch) {
1679                 ++count;
1680                 ++i;
1681             }
1682 
1683             if (isNumeric(ch, count)) {
1684                 if (abutPat < 0) {
1685                     // Determine if there is an abutting numeric field.
1686                     // Record the start of a set of abutting numeric fields.
1687                     if (isAtNumericField(fPattern, i + 1)) {
1688                         abutPat = fieldPat;
1689                         abutStart = pos;
1690                         abutPass = 0;
1691                     }
1692                 }
1693             } else {
1694                 abutPat = -1; // End of any abutting fields
1695             }
1696 
1697             // Handle fields within a run of abutting numeric fields.  Take
1698             // the pattern "HHmmss" as an example. We will try to parse
1699             // 2/2/2 characters of the input text, then if that fails,
1700             // 1/2/2.  We only adjust the width of the leftmost field; the
1701             // others remain fixed.  This allows "123456" => 12:34:56, but
1702             // "12345" => 1:23:45.  Likewise, for the pattern "yyyyMMdd" we
1703             // try 4/2/2, 3/2/2, 2/2/2, and finally 1/2/2.
1704             if (abutPat >= 0) {
1705                 // If we are at the start of a run of abutting fields, then
1706                 // shorten this field in each pass.  If we can't shorten
1707                 // this field any more, then the parse of this set of
1708                 // abutting numeric fields has failed.
1709                 if (fieldPat == abutPat) {
1710                     count -= abutPass++;
1711                     if (count == 0) {
1712                         status = U_PARSE_ERROR;
1713                         goto ExitParse;
1714                     }
1715                 }
1716 
1717                 pos = subParse(text, pos, ch, count,
1718                                TRUE, FALSE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter);
1719 
1720                 // If the parse fails anywhere in the run, back up to the
1721                 // start of the run and retry.
1722                 if (pos < 0) {
1723                     i = abutPat - 1;
1724                     pos = abutStart;
1725                     continue;
1726                 }
1727             }
1728 
1729             // Handle non-numeric fields and non-abutting numeric
1730             // fields.
1731             else if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
1732                 int32_t s = subParse(text, pos, ch, count,
1733                                FALSE, TRUE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter);
1734 
1735                 if (s == -pos-1) {
1736                     // era not present, in special cases allow this to continue
1737                     // from the position where the era was expected
1738                     s = pos;
1739 
1740                     if (i+1 < fPattern.length()) {
1741                         // move to next pattern character
1742                         UChar ch = fPattern.charAt(i+1);
1743 
1744                         // check for whitespace
1745                         if (PatternProps::isWhiteSpace(ch)) {
1746                             i++;
1747                             // Advance over run in pattern
1748                             while ((i+1)<fPattern.length() &&
1749                                    PatternProps::isWhiteSpace(fPattern.charAt(i+1))) {
1750                                 ++i;
1751                             }
1752                         }
1753                     }
1754                 }
1755                 else if (s <= 0) {
1756                     status = U_PARSE_ERROR;
1757                     goto ExitParse;
1758                 }
1759                 pos = s;
1760             }
1761         }
1762 
1763         // Handle literal pattern characters.  These are any
1764         // quoted characters and non-alphabetic unquoted
1765         // characters.
1766         else {
1767 
1768             abutPat = -1; // End of any abutting fields
1769 
1770             if (! matchLiterals(fPattern, i, text, pos, lenient)) {
1771                 status = U_PARSE_ERROR;
1772                 goto ExitParse;
1773             }
1774         }
1775     }
1776 
1777     // Special hack for trailing "." after non-numeric field.
1778     if (text.charAt(pos) == 0x2e && lenient) {
1779         // only do if the last field is not numeric
1780         if (isAfterNonNumericField(fPattern, fPattern.length())) {
1781             pos++; // skip the extra "."
1782         }
1783     }
1784 
1785     // At this point the fields of Calendar have been set.  Calendar
1786     // will fill in default values for missing fields when the time
1787     // is computed.
1788 
1789     parsePos.setIndex(pos);
1790 
1791     // This part is a problem:  When we call parsedDate.after, we compute the time.
1792     // Take the date April 3 2004 at 2:30 am.  When this is first set up, the year
1793     // will be wrong if we're parsing a 2-digit year pattern.  It will be 1904.
1794     // April 3 1904 is a Sunday (unlike 2004) so it is the DST onset day.  2:30 am
1795     // is therefore an "impossible" time, since the time goes from 1:59 to 3:00 am
1796     // on that day.  It is therefore parsed out to fields as 3:30 am.  Then we
1797     // add 100 years, and get April 3 2004 at 3:30 am.  Note that April 3 2004 is
1798     // a Saturday, so it can have a 2:30 am -- and it should. [LIU]
1799     /*
1800         UDate parsedDate = calendar.getTime();
1801         if( ambiguousYear[0] && !parsedDate.after(fDefaultCenturyStart) ) {
1802             calendar.add(Calendar.YEAR, 100);
1803             parsedDate = calendar.getTime();
1804         }
1805     */
1806     // Because of the above condition, save off the fields in case we need to readjust.
1807     // The procedure we use here is not particularly efficient, but there is no other
1808     // way to do this given the API restrictions present in Calendar.  We minimize
1809     // inefficiency by only performing this computation when it might apply, that is,
1810     // when the two-digit year is equal to the start year, and thus might fall at the
1811     // front or the back of the default century.  This only works because we adjust
1812     // the year correctly to start with in other cases -- see subParse().
1813     if (ambiguousYear[0] || tztype != UTZFMT_TIME_TYPE_UNKNOWN) // If this is true then the two-digit year == the default start year
1814     {
1815         // We need a copy of the fields, and we need to avoid triggering a call to
1816         // complete(), which will recalculate the fields.  Since we can't access
1817         // the fields[] array in Calendar, we clone the entire object.  This will
1818         // stop working if Calendar.clone() is ever rewritten to call complete().
1819         Calendar *copy;
1820         if (ambiguousYear[0]) {
1821             copy = cal.clone();
1822             // Check for failed cloning.
1823             if (copy == NULL) {
1824                 status = U_MEMORY_ALLOCATION_ERROR;
1825                 goto ExitParse;
1826             }
1827             UDate parsedDate = copy->getTime(status);
1828             // {sfb} check internalGetDefaultCenturyStart
1829             if (fHaveDefaultCentury && (parsedDate < fDefaultCenturyStart)) {
1830                 // We can't use add here because that does a complete() first.
1831                 cal.set(UCAL_YEAR, fDefaultCenturyStartYear + 100);
1832             }
1833             delete copy;
1834         }
1835 
1836         if (tztype != UTZFMT_TIME_TYPE_UNKNOWN) {
1837             copy = cal.clone();
1838             // Check for failed cloning.
1839             if (copy == NULL) {
1840                 status = U_MEMORY_ALLOCATION_ERROR;
1841                 goto ExitParse;
1842             }
1843             const TimeZone & tz = cal.getTimeZone();
1844             BasicTimeZone *btz = NULL;
1845 
1846             if (dynamic_cast<const OlsonTimeZone *>(&tz) != NULL
1847                 || dynamic_cast<const SimpleTimeZone *>(&tz) != NULL
1848                 || dynamic_cast<const RuleBasedTimeZone *>(&tz) != NULL
1849                 || dynamic_cast<const VTimeZone *>(&tz) != NULL) {
1850                 btz = (BasicTimeZone*)&tz;
1851             }
1852 
1853             // Get local millis
1854             copy->set(UCAL_ZONE_OFFSET, 0);
1855             copy->set(UCAL_DST_OFFSET, 0);
1856             UDate localMillis = copy->getTime(status);
1857 
1858             // Make sure parsed time zone type (Standard or Daylight)
1859             // matches the rule used by the parsed time zone.
1860             int32_t raw, dst;
1861             if (btz != NULL) {
1862                 if (tztype == UTZFMT_TIME_TYPE_STANDARD) {
1863                     btz->getOffsetFromLocal(localMillis,
1864                         BasicTimeZone::kStandard, BasicTimeZone::kStandard, raw, dst, status);
1865                 } else {
1866                     btz->getOffsetFromLocal(localMillis,
1867                         BasicTimeZone::kDaylight, BasicTimeZone::kDaylight, raw, dst, status);
1868                 }
1869             } else {
1870                 // No good way to resolve ambiguous time at transition,
1871                 // but following code work in most case.
1872                 tz.getOffset(localMillis, TRUE, raw, dst, status);
1873             }
1874 
1875             // Now, compare the results with parsed type, either standard or daylight saving time
1876             int32_t resolvedSavings = dst;
1877             if (tztype == UTZFMT_TIME_TYPE_STANDARD) {
1878                 if (dst != 0) {
1879                     // Override DST_OFFSET = 0 in the result calendar
1880                     resolvedSavings = 0;
1881                 }
1882             } else { // tztype == TZTYPE_DST
1883                 if (dst == 0) {
1884                     if (btz != NULL) {
1885                         UDate time = localMillis + raw;
1886                         // We use the nearest daylight saving time rule.
1887                         TimeZoneTransition beforeTrs, afterTrs;
1888                         UDate beforeT = time, afterT = time;
1889                         int32_t beforeSav = 0, afterSav = 0;
1890                         UBool beforeTrsAvail, afterTrsAvail;
1891 
1892                         // Search for DST rule before or on the time
1893                         while (TRUE) {
1894                             beforeTrsAvail = btz->getPreviousTransition(beforeT, TRUE, beforeTrs);
1895                             if (!beforeTrsAvail) {
1896                                 break;
1897                             }
1898                             beforeT = beforeTrs.getTime() - 1;
1899                             beforeSav = beforeTrs.getFrom()->getDSTSavings();
1900                             if (beforeSav != 0) {
1901                                 break;
1902                             }
1903                         }
1904 
1905                         // Search for DST rule after the time
1906                         while (TRUE) {
1907                             afterTrsAvail = btz->getNextTransition(afterT, FALSE, afterTrs);
1908                             if (!afterTrsAvail) {
1909                                 break;
1910                             }
1911                             afterT = afterTrs.getTime();
1912                             afterSav = afterTrs.getTo()->getDSTSavings();
1913                             if (afterSav != 0) {
1914                                 break;
1915                             }
1916                         }
1917 
1918                         if (beforeTrsAvail && afterTrsAvail) {
1919                             if (time - beforeT > afterT - time) {
1920                                 resolvedSavings = afterSav;
1921                             } else {
1922                                 resolvedSavings = beforeSav;
1923                             }
1924                         } else if (beforeTrsAvail && beforeSav != 0) {
1925                             resolvedSavings = beforeSav;
1926                         } else if (afterTrsAvail && afterSav != 0) {
1927                             resolvedSavings = afterSav;
1928                         } else {
1929                             resolvedSavings = btz->getDSTSavings();
1930                         }
1931                     } else {
1932                         resolvedSavings = tz.getDSTSavings();
1933                     }
1934                     if (resolvedSavings == 0) {
1935                         // final fallback
1936                         resolvedSavings = U_MILLIS_PER_HOUR;
1937                     }
1938                 }
1939             }
1940             cal.set(UCAL_ZONE_OFFSET, raw);
1941             cal.set(UCAL_DST_OFFSET, resolvedSavings);
1942             delete copy;
1943         }
1944     }
1945 ExitParse:
1946     // Set the parsed result if local calendar is used
1947     // instead of the input calendar
1948     if (U_SUCCESS(status) && workCal != &cal) {
1949         cal.setTimeZone(workCal->getTimeZone());
1950         cal.setTime(workCal->getTime(status), status);
1951     }
1952 
1953     if (numericLeapMonthFormatter != NULL) {
1954         delete numericLeapMonthFormatter;
1955     }
1956     if (calClone != NULL) {
1957         delete calClone;
1958     }
1959 
1960     // If any Calendar calls failed, we pretend that we
1961     // couldn't parse the string, when in reality this isn't quite accurate--
1962     // we did parse it; the Calendar calls just failed.
1963     if (U_FAILURE(status)) {
1964         parsePos.setErrorIndex(pos);
1965         parsePos.setIndex(start);
1966     }
1967 }
1968 
1969 UDate
parse(const UnicodeString & text,ParsePosition & pos) const1970 SimpleDateFormat::parse( const UnicodeString& text,
1971                          ParsePosition& pos) const {
1972     // redefined here because the other parse() function hides this function's
1973     // cunterpart on DateFormat
1974     return DateFormat::parse(text, pos);
1975 }
1976 
1977 UDate
parse(const UnicodeString & text,UErrorCode & status) const1978 SimpleDateFormat::parse(const UnicodeString& text, UErrorCode& status) const
1979 {
1980     // redefined here because the other parse() function hides this function's
1981     // counterpart on DateFormat
1982     return DateFormat::parse(text, status);
1983 }
1984 //----------------------------------------------------------------------
1985 
1986 static UBool
1987 newBestMatchWithOptionalDot(const UnicodeString &lcaseText,
1988                             const UnicodeString &data,
1989                             UnicodeString &bestMatchName,
1990                             int32_t &bestMatchLength);
1991 
matchQuarterString(const UnicodeString & text,int32_t start,UCalendarDateFields field,const UnicodeString * data,int32_t dataCount,Calendar & cal) const1992 int32_t SimpleDateFormat::matchQuarterString(const UnicodeString& text,
1993                               int32_t start,
1994                               UCalendarDateFields field,
1995                               const UnicodeString* data,
1996                               int32_t dataCount,
1997                               Calendar& cal) const
1998 {
1999     int32_t i = 0;
2000     int32_t count = dataCount;
2001 
2002     // There may be multiple strings in the data[] array which begin with
2003     // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2004     // We keep track of the longest match, and return that.  Note that this
2005     // unfortunately requires us to test all array elements.
2006     int32_t bestMatchLength = 0, bestMatch = -1;
2007     UnicodeString bestMatchName;
2008 
2009     // {sfb} kludge to support case-insensitive comparison
2010     // {markus 2002oct11} do not just use caseCompareBetween because we do not know
2011     // the length of the match after case folding
2012     // {alan 20040607} don't case change the whole string, since the length
2013     // can change
2014     // TODO we need a case-insensitive startsWith function
2015     UnicodeString lcaseText;
2016     text.extract(start, INT32_MAX, lcaseText);
2017     lcaseText.foldCase();
2018 
2019     for (; i < count; ++i)
2020     {
2021         // Always compare if we have no match yet; otherwise only compare
2022         // against potentially better matches (longer strings).
2023 
2024         if (newBestMatchWithOptionalDot(lcaseText, data[i], bestMatchName, bestMatchLength)) {
2025             bestMatch = i;
2026         }
2027     }
2028     if (bestMatch >= 0)
2029     {
2030         cal.set(field, bestMatch * 3);
2031 
2032         // Once we have a match, we have to determine the length of the
2033         // original source string.  This will usually be == the length of
2034         // the case folded string, but it may differ (e.g. sharp s).
2035 
2036         // Most of the time, the length will be the same as the length
2037         // of the string from the locale data.  Sometimes it will be
2038         // different, in which case we will have to figure it out by
2039         // adding a character at a time, until we have a match.  We do
2040         // this all in one loop, where we try 'len' first (at index
2041         // i==0).
2042         int32_t len = bestMatchName.length(); // 99+% of the time
2043         int32_t n = text.length() - start;
2044         for (i=0; i<=n; ++i) {
2045             int32_t j=i;
2046             if (i == 0) {
2047                 j = len;
2048             } else if (i == len) {
2049                 continue; // already tried this when i was 0
2050             }
2051             text.extract(start, j, lcaseText);
2052             lcaseText.foldCase();
2053             if (bestMatchName == lcaseText) {
2054                 return start + j;
2055             }
2056         }
2057     }
2058 
2059     return -start;
2060 }
2061 
2062 //----------------------------------------------------------------------
matchLiterals(const UnicodeString & pattern,int32_t & patternOffset,const UnicodeString & text,int32_t & textOffset,UBool lenient)2063 UBool SimpleDateFormat::matchLiterals(const UnicodeString &pattern,
2064                                       int32_t &patternOffset,
2065                                       const UnicodeString &text,
2066                                       int32_t &textOffset,
2067                                       UBool lenient)
2068 {
2069     UBool inQuote = FALSE;
2070     UnicodeString literal;
2071     int32_t i = patternOffset;
2072 
2073     // scan pattern looking for contiguous literal characters
2074     for ( ; i < pattern.length(); i += 1) {
2075         UChar ch = pattern.charAt(i);
2076 
2077         if (!inQuote && ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { // unquoted [A-Za-z]
2078             break;
2079         }
2080 
2081         if (ch == QUOTE) {
2082             // Match a quote literal ('') inside OR outside of quotes
2083             if ((i + 1) < pattern.length() && pattern.charAt(i + 1) == QUOTE) {
2084                 i += 1;
2085             } else {
2086                 inQuote = !inQuote;
2087                 continue;
2088             }
2089         }
2090 
2091         literal += ch;
2092     }
2093 
2094     // at this point, literal contains the literal text
2095     // and i is the index of the next non-literal pattern character.
2096     int32_t p;
2097     int32_t t = textOffset;
2098 
2099     if (lenient) {
2100         // trim leading, trailing whitespace from
2101         // the literal text
2102         literal.trim();
2103 
2104         // ignore any leading whitespace in the text
2105         while (t < text.length() && u_isWhitespace(text.charAt(t))) {
2106             t += 1;
2107         }
2108     }
2109 
2110     for (p = 0; p < literal.length() && t < text.length();) {
2111         UBool needWhitespace = FALSE;
2112 
2113         while (p < literal.length() && PatternProps::isWhiteSpace(literal.charAt(p))) {
2114             needWhitespace = TRUE;
2115             p += 1;
2116         }
2117 
2118         if (needWhitespace) {
2119             int32_t tStart = t;
2120 
2121             while (t < text.length()) {
2122                 UChar tch = text.charAt(t);
2123 
2124                 if (!u_isUWhiteSpace(tch) && !PatternProps::isWhiteSpace(tch)) {
2125                     break;
2126                 }
2127 
2128                 t += 1;
2129             }
2130 
2131             // TODO: should we require internal spaces
2132             // in lenient mode? (There won't be any
2133             // leading or trailing spaces)
2134             if (!lenient && t == tStart) {
2135                 // didn't find matching whitespace:
2136                 // an error in strict mode
2137                 return FALSE;
2138             }
2139 
2140             // In strict mode, this run of whitespace
2141             // may have been at the end.
2142             if (p >= literal.length()) {
2143                 break;
2144             }
2145         }
2146 
2147         if (t >= text.length() || literal.charAt(p) != text.charAt(t)) {
2148             // Ran out of text, or found a non-matching character:
2149             // OK in lenient mode, an error in strict mode.
2150             if (lenient) {
2151                 if (t == textOffset && text.charAt(t) == 0x2e &&
2152                         isAfterNonNumericField(pattern, patternOffset)) {
2153                     // Lenient mode and the literal input text begins with a "." and
2154                     // we are after a non-numeric field: We skip the "."
2155                     ++t;
2156                     continue;  // Do not update p.
2157                 }
2158                 break;
2159             }
2160 
2161             return FALSE;
2162         }
2163         ++p;
2164         ++t;
2165     }
2166 
2167     // At this point if we're in strict mode we have a complete match.
2168     // If we're in lenient mode we may have a partial match, or no
2169     // match at all.
2170     if (p <= 0) {
2171         // no match. Pretend it matched a run of whitespace
2172         // and ignorables in the text.
2173         const  UnicodeSet *ignorables = NULL;
2174         UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(pattern.charAt(i));
2175         if (patternCharIndex != UDAT_FIELD_COUNT) {
2176             ignorables = SimpleDateFormatStaticSets::getIgnorables(patternCharIndex);
2177         }
2178 
2179         for (t = textOffset; t < text.length(); t += 1) {
2180             UChar ch = text.charAt(t);
2181 
2182             if (ignorables == NULL || !ignorables->contains(ch)) {
2183                 break;
2184             }
2185         }
2186     }
2187 
2188     // if we get here, we've got a complete match.
2189     patternOffset = i - 1;
2190     textOffset = t;
2191 
2192     return TRUE;
2193 }
2194 
2195 //----------------------------------------------------------------------
2196 
matchString(const UnicodeString & text,int32_t start,UCalendarDateFields field,const UnicodeString * data,int32_t dataCount,const UnicodeString * monthPattern,Calendar & cal) const2197 int32_t SimpleDateFormat::matchString(const UnicodeString& text,
2198                               int32_t start,
2199                               UCalendarDateFields field,
2200                               const UnicodeString* data,
2201                               int32_t dataCount,
2202                               const UnicodeString* monthPattern,
2203                               Calendar& cal) const
2204 {
2205     int32_t i = 0;
2206     int32_t count = dataCount;
2207 
2208     if (field == UCAL_DAY_OF_WEEK) i = 1;
2209 
2210     // There may be multiple strings in the data[] array which begin with
2211     // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2212     // We keep track of the longest match, and return that.  Note that this
2213     // unfortunately requires us to test all array elements.
2214     int32_t bestMatchLength = 0, bestMatch = -1;
2215     UnicodeString bestMatchName;
2216     int32_t isLeapMonth = 0;
2217 
2218     // {sfb} kludge to support case-insensitive comparison
2219     // {markus 2002oct11} do not just use caseCompareBetween because we do not know
2220     // the length of the match after case folding
2221     // {alan 20040607} don't case change the whole string, since the length
2222     // can change
2223     // TODO we need a case-insensitive startsWith function
2224     UnicodeString lcaseText;
2225     text.extract(start, INT32_MAX, lcaseText);
2226     lcaseText.foldCase();
2227 
2228     for (; i < count; ++i)
2229     {
2230         // Always compare if we have no match yet; otherwise only compare
2231         // against potentially better matches (longer strings).
2232 
2233         if (newBestMatchWithOptionalDot(lcaseText, data[i], bestMatchName, bestMatchLength)) {
2234             bestMatch = i;
2235             isLeapMonth = 0;
2236         }
2237 
2238         if (monthPattern != NULL) {
2239             UErrorCode status = U_ZERO_ERROR;
2240             UnicodeString leapMonthName;
2241             Formattable monthName((const UnicodeString&)(data[i]));
2242             MessageFormat::format(*monthPattern, &monthName, 1, leapMonthName, status);
2243             if (U_SUCCESS(status)) {
2244                 if (newBestMatchWithOptionalDot(lcaseText, leapMonthName, bestMatchName, bestMatchLength)) {
2245                     bestMatch = i;
2246                     isLeapMonth = 1;
2247                 }
2248             }
2249         }
2250     }
2251     if (bestMatch >= 0)
2252     {
2253         // Adjustment for Hebrew Calendar month Adar II
2254         if (!strcmp(cal.getType(),"hebrew") && field==UCAL_MONTH && bestMatch==13) {
2255             cal.set(field,6);
2256         }
2257         else {
2258             if (field == UCAL_YEAR) {
2259                 bestMatch++; // only get here for cyclic year names, which match 1-based years 1-60
2260             }
2261             cal.set(field, bestMatch);
2262         }
2263         if (monthPattern != NULL) {
2264             cal.set(UCAL_IS_LEAP_MONTH, isLeapMonth);
2265         }
2266 
2267         // Once we have a match, we have to determine the length of the
2268         // original source string.  This will usually be == the length of
2269         // the case folded string, but it may differ (e.g. sharp s).
2270 
2271         // Most of the time, the length will be the same as the length
2272         // of the string from the locale data.  Sometimes it will be
2273         // different, in which case we will have to figure it out by
2274         // adding a character at a time, until we have a match.  We do
2275         // this all in one loop, where we try 'len' first (at index
2276         // i==0).
2277         int32_t len = bestMatchName.length(); // 99+% of the time
2278         int32_t n = text.length() - start;
2279         for (i=0; i<=n; ++i) {
2280             int32_t j=i;
2281             if (i == 0) {
2282                 j = len;
2283             } else if (i == len) {
2284                 continue; // already tried this when i was 0
2285             }
2286             text.extract(start, j, lcaseText);
2287             lcaseText.foldCase();
2288             if (bestMatchName == lcaseText) {
2289                 return start + j;
2290             }
2291         }
2292     }
2293 
2294     return -start;
2295 }
2296 
2297 static UBool
newBestMatchWithOptionalDot(const UnicodeString & lcaseText,const UnicodeString & data,UnicodeString & bestMatchName,int32_t & bestMatchLength)2298 newBestMatchWithOptionalDot(const UnicodeString &lcaseText,
2299                             const UnicodeString &data,
2300                             UnicodeString &bestMatchName,
2301                             int32_t &bestMatchLength) {
2302     UnicodeString lcase;
2303     lcase.fastCopyFrom(data).foldCase();
2304     int32_t length = lcase.length();
2305     if (length <= bestMatchLength) {
2306         // data cannot provide a better match.
2307         return FALSE;
2308     }
2309 
2310     if (lcaseText.compareBetween(0, length, lcase, 0, length) == 0) {
2311         // normal match
2312         bestMatchName = lcase;
2313         bestMatchLength = length;
2314         return TRUE;
2315     }
2316     if (lcase.charAt(--length) == 0x2e) {
2317         if (lcaseText.compareBetween(0, length, lcase, 0, length) == 0) {
2318             // The input text matches the data except for data's trailing dot.
2319             bestMatchName = lcase;
2320             bestMatchName.truncate(length);
2321             bestMatchLength = length;
2322             return TRUE;
2323         }
2324     }
2325     return FALSE;
2326 }
2327 
2328 //----------------------------------------------------------------------
2329 
2330 void
set2DigitYearStart(UDate d,UErrorCode & status)2331 SimpleDateFormat::set2DigitYearStart(UDate d, UErrorCode& status)
2332 {
2333     parseAmbiguousDatesAsAfter(d, status);
2334 }
2335 
2336 /**
2337  * Private member function that converts the parsed date strings into
2338  * timeFields. Returns -start (for ParsePosition) if failed.
2339  * @param text the time text to be parsed.
2340  * @param start where to start parsing.
2341  * @param ch the pattern character for the date field text to be parsed.
2342  * @param count the count of a pattern character.
2343  * @return the new start position if matching succeeded; a negative number
2344  * indicating matching failure, otherwise.
2345  */
subParse(const UnicodeString & text,int32_t & start,UChar ch,int32_t count,UBool obeyCount,UBool allowNegative,UBool ambiguousYear[],int32_t & saveHebrewMonth,Calendar & cal,int32_t patLoc,MessageFormat * numericLeapMonthFormatter) const2346 int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UChar ch, int32_t count,
2347                            UBool obeyCount, UBool allowNegative, UBool ambiguousYear[], int32_t& saveHebrewMonth, Calendar& cal,
2348                            int32_t patLoc, MessageFormat * numericLeapMonthFormatter) const
2349 {
2350     Formattable number;
2351     int32_t value = 0;
2352     int32_t i;
2353     int32_t ps = 0;
2354     ParsePosition pos(0);
2355     UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
2356     NumberFormat *currentNumberFormat;
2357     UnicodeString temp;
2358     UBool lenient = isLenient();
2359     UBool gotNumber = FALSE;
2360 
2361 #if defined (U_DEBUG_CAL)
2362     //fprintf(stderr, "%s:%d - [%c]  st=%d \n", __FILE__, __LINE__, (char) ch, start);
2363 #endif
2364 
2365     if (patternCharIndex == UDAT_FIELD_COUNT) {
2366         return -start;
2367     }
2368 
2369     currentNumberFormat = getNumberFormatByIndex(patternCharIndex);
2370     UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex];
2371     UnicodeString hebr("hebr", 4, US_INV);
2372 
2373     if (numericLeapMonthFormatter != NULL) {
2374         numericLeapMonthFormatter->setFormats((const Format **)&currentNumberFormat, 1);
2375     }
2376     UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0);
2377 
2378     // If there are any spaces here, skip over them.  If we hit the end
2379     // of the string, then fail.
2380     for (;;) {
2381         if (start >= text.length()) {
2382             return -start;
2383         }
2384         UChar32 c = text.char32At(start);
2385         if (!u_isUWhiteSpace(c) /*||*/ && !PatternProps::isWhiteSpace(c)) {
2386             break;
2387         }
2388         start += U16_LENGTH(c);
2389     }
2390     pos.setIndex(start);
2391 
2392     // We handle a few special cases here where we need to parse
2393     // a number value.  We handle further, more generic cases below.  We need
2394     // to handle some of them here because some fields require extra processing on
2395     // the parsed value.
2396     if (patternCharIndex == UDAT_HOUR_OF_DAY1_FIELD ||                       // k
2397         patternCharIndex == UDAT_HOUR_OF_DAY0_FIELD ||                       // H
2398         patternCharIndex == UDAT_HOUR1_FIELD ||                              // h
2399         patternCharIndex == UDAT_HOUR0_FIELD ||                              // K
2400         (patternCharIndex == UDAT_DOW_LOCAL_FIELD && count <= 2) ||          // e
2401         (patternCharIndex == UDAT_STANDALONE_DAY_FIELD && count <= 2) ||     // c
2402         (patternCharIndex == UDAT_MONTH_FIELD && count <= 2) ||              // M
2403         (patternCharIndex == UDAT_STANDALONE_MONTH_FIELD && count <= 2) ||   // L
2404         (patternCharIndex == UDAT_QUARTER_FIELD && count <= 2) ||            // Q
2405         (patternCharIndex == UDAT_STANDALONE_QUARTER_FIELD && count <= 2) || // q
2406         patternCharIndex == UDAT_YEAR_FIELD ||                               // y
2407         patternCharIndex == UDAT_YEAR_WOY_FIELD ||                           // Y
2408         patternCharIndex == UDAT_YEAR_NAME_FIELD ||                          // U (falls back to numeric)
2409         (patternCharIndex == UDAT_ERA_FIELD && isChineseCalendar) ||         // G
2410         patternCharIndex == UDAT_FRACTIONAL_SECOND_FIELD)                    // S
2411     {
2412         int32_t parseStart = pos.getIndex();
2413         // It would be good to unify this with the obeyCount logic below,
2414         // but that's going to be difficult.
2415         const UnicodeString* src;
2416 
2417         UBool parsedNumericLeapMonth = FALSE;
2418         if (numericLeapMonthFormatter != NULL && (patternCharIndex == UDAT_MONTH_FIELD || patternCharIndex == UDAT_STANDALONE_MONTH_FIELD)) {
2419             int32_t argCount;
2420             Formattable * args = numericLeapMonthFormatter->parse(text, pos, argCount);
2421             if (args != NULL && argCount == 1 && pos.getIndex() > parseStart && args[0].isNumeric()) {
2422                 parsedNumericLeapMonth = TRUE;
2423                 number.setLong(args[0].getLong());
2424                 cal.set(UCAL_IS_LEAP_MONTH, 1);
2425                 delete[] args;
2426             } else {
2427                 pos.setIndex(parseStart);
2428                 cal.set(UCAL_IS_LEAP_MONTH, 0);
2429             }
2430         }
2431 
2432         if (!parsedNumericLeapMonth) {
2433             if (obeyCount) {
2434                 if ((start+count) > text.length()) {
2435                     return -start;
2436                 }
2437 
2438                 text.extractBetween(0, start + count, temp);
2439                 src = &temp;
2440             } else {
2441                 src = &text;
2442             }
2443 
2444             parseInt(*src, number, pos, allowNegative,currentNumberFormat);
2445         }
2446 
2447         int32_t txtLoc = pos.getIndex();
2448 
2449         if (txtLoc > parseStart) {
2450             value = number.getLong();
2451             gotNumber = TRUE;
2452 
2453             // suffix processing
2454             if (value < 0 ) {
2455                 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, TRUE);
2456                 if (txtLoc != pos.getIndex()) {
2457                     value *= -1;
2458                 }
2459             }
2460             else {
2461                 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, FALSE);
2462             }
2463 
2464             if (!lenient) {
2465                 // Check the range of the value
2466                 int32_t bias = gFieldRangeBias[patternCharIndex];
2467                 if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
2468                     return -start;
2469                 }
2470             }
2471 
2472             pos.setIndex(txtLoc);
2473         }
2474     }
2475 
2476     // Make sure that we got a number if
2477     // we want one, and didn't get one
2478     // if we don't want one.
2479     switch (patternCharIndex) {
2480         case UDAT_HOUR_OF_DAY1_FIELD:
2481         case UDAT_HOUR_OF_DAY0_FIELD:
2482         case UDAT_HOUR1_FIELD:
2483         case UDAT_HOUR0_FIELD:
2484             // special range check for hours:
2485             if (value < 0 || value > 24) {
2486                 return -start;
2487             }
2488 
2489             // fall through to gotNumber check
2490 
2491         case UDAT_YEAR_FIELD:
2492         case UDAT_YEAR_WOY_FIELD:
2493         case UDAT_FRACTIONAL_SECOND_FIELD:
2494             // these must be a number
2495             if (! gotNumber) {
2496                 return -start;
2497             }
2498 
2499             break;
2500 
2501         default:
2502             // we check the rest of the fields below.
2503             break;
2504     }
2505 
2506     switch (patternCharIndex) {
2507     case UDAT_ERA_FIELD:
2508         if (isChineseCalendar) {
2509             if (!gotNumber) {
2510                 return -start;
2511             }
2512             cal.set(UCAL_ERA, value);
2513             return pos.getIndex();
2514         }
2515         if (count == 5) {
2516             ps = matchString(text, start, UCAL_ERA, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount, NULL, cal);
2517         } else if (count == 4) {
2518             ps = matchString(text, start, UCAL_ERA, fSymbols->fEraNames, fSymbols->fEraNamesCount, NULL, cal);
2519         } else {
2520             ps = matchString(text, start, UCAL_ERA, fSymbols->fEras, fSymbols->fErasCount, NULL, cal);
2521         }
2522 
2523         // check return position, if it equals -start, then matchString error
2524         // special case the return code so we don't necessarily fail out until we
2525         // verify no year information also
2526         if (ps == -start)
2527             ps--;
2528 
2529         return ps;
2530 
2531     case UDAT_YEAR_FIELD:
2532         // If there are 3 or more YEAR pattern characters, this indicates
2533         // that the year value is to be treated literally, without any
2534         // two-digit year adjustments (e.g., from "01" to 2001).  Otherwise
2535         // we made adjustments to place the 2-digit year in the proper
2536         // century, for parsed strings from "00" to "99".  Any other string
2537         // is treated literally:  "2250", "-1", "1", "002".
2538         if (fDateOverride.compare(hebr)==0 && value < 1000) {
2539             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2540         } else if ((pos.getIndex() - start) == 2 && !isChineseCalendar
2541             && u_isdigit(text.charAt(start))
2542             && u_isdigit(text.charAt(start+1)))
2543         {
2544             // Assume for example that the defaultCenturyStart is 6/18/1903.
2545             // This means that two-digit years will be forced into the range
2546             // 6/18/1903 to 6/17/2003.  As a result, years 00, 01, and 02
2547             // correspond to 2000, 2001, and 2002.  Years 04, 05, etc. correspond
2548             // to 1904, 1905, etc.  If the year is 03, then it is 2003 if the
2549             // other fields specify a date before 6/18, or 1903 if they specify a
2550             // date afterwards.  As a result, 03 is an ambiguous year.  All other
2551             // two-digit years are unambiguous.
2552           if(fHaveDefaultCentury) { // check if this formatter even has a pivot year
2553               int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2554               ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2555               value += (fDefaultCenturyStartYear/100)*100 +
2556                 (value < ambiguousTwoDigitYear ? 100 : 0);
2557             }
2558         }
2559         cal.set(UCAL_YEAR, value);
2560 
2561         // Delayed checking for adjustment of Hebrew month numbers in non-leap years.
2562         if (saveHebrewMonth >= 0) {
2563             HebrewCalendar *hc = (HebrewCalendar*)&cal;
2564             if (!hc->isLeapYear(value) && saveHebrewMonth >= 6) {
2565                cal.set(UCAL_MONTH,saveHebrewMonth);
2566             } else {
2567                cal.set(UCAL_MONTH,saveHebrewMonth-1);
2568             }
2569             saveHebrewMonth = -1;
2570         }
2571         return pos.getIndex();
2572 
2573     case UDAT_YEAR_WOY_FIELD:
2574         // Comment is the same as for UDAT_Year_FIELDs - look above
2575         if (fDateOverride.compare(hebr)==0 && value < 1000) {
2576             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2577         } else if ((pos.getIndex() - start) == 2
2578             && u_isdigit(text.charAt(start))
2579             && u_isdigit(text.charAt(start+1))
2580             && fHaveDefaultCentury )
2581         {
2582             int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2583             ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2584             value += (fDefaultCenturyStartYear/100)*100 +
2585                 (value < ambiguousTwoDigitYear ? 100 : 0);
2586         }
2587         cal.set(UCAL_YEAR_WOY, value);
2588         return pos.getIndex();
2589 
2590     case UDAT_YEAR_NAME_FIELD:
2591         if (fSymbols->fShortYearNames != NULL) {
2592             int32_t newStart = matchString(text, start, UCAL_YEAR, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount, NULL, cal);
2593             if (newStart > 0) {
2594                 return newStart;
2595             }
2596         }
2597         if (gotNumber && (lenient || value > fSymbols->fShortYearNamesCount)) {
2598             cal.set(UCAL_YEAR, value);
2599             return pos.getIndex();
2600         }
2601         return -start;
2602 
2603     case UDAT_MONTH_FIELD:
2604     case UDAT_STANDALONE_MONTH_FIELD:
2605         if (gotNumber) // i.e., M or MM.
2606         {
2607             // When parsing month numbers from the Hebrew Calendar, we might need to adjust the month depending on whether
2608             // or not it was a leap year.  We may or may not yet know what year it is, so might have to delay checking until
2609             // the year is parsed.
2610             if (!strcmp(cal.getType(),"hebrew")) {
2611                 HebrewCalendar *hc = (HebrewCalendar*)&cal;
2612                 if (cal.isSet(UCAL_YEAR)) {
2613                    UErrorCode status = U_ZERO_ERROR;
2614                    if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
2615                        cal.set(UCAL_MONTH, value);
2616                    } else {
2617                        cal.set(UCAL_MONTH, value - 1);
2618                    }
2619                 } else {
2620                     saveHebrewMonth = value;
2621                 }
2622             } else {
2623                 // Don't want to parse the month if it is a string
2624                 // while pattern uses numeric style: M/MM, L/LL
2625                 // [We computed 'value' above.]
2626                 cal.set(UCAL_MONTH, value - 1);
2627             }
2628             return pos.getIndex();
2629         } else {
2630             // count >= 3 // i.e., MMM/MMMM, LLL/LLLL
2631             // Want to be able to parse both short and long forms.
2632             // Try count == 4 first:
2633             UnicodeString * wideMonthPat = NULL;
2634             UnicodeString * shortMonthPat = NULL;
2635             if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
2636                 if (patternCharIndex==UDAT_MONTH_FIELD) {
2637                     wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide];
2638                     shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev];
2639                 } else {
2640                     wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide];
2641                     shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev];
2642                 }
2643             }
2644             int32_t newStart = 0;
2645             if (patternCharIndex==UDAT_MONTH_FIELD) {
2646                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fMonths, fSymbols->fMonthsCount, wideMonthPat, cal); // try MMMM
2647                 if (newStart > 0) {
2648                     return newStart;
2649                 }
2650                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fShortMonths, fSymbols->fShortMonthsCount, shortMonthPat, cal); // try MMM
2651             } else {
2652                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount, wideMonthPat, cal); // try LLLL
2653                 if (newStart > 0) {
2654                     return newStart;
2655                 }
2656                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount, shortMonthPat, cal); // try LLL
2657             }
2658             if (newStart > 0 || !lenient)  // currently we do not try to parse MMMMM/LLLLL: #8860
2659                 return newStart;
2660             // else we allowing parsing as number, below
2661         }
2662         break;
2663 
2664     case UDAT_HOUR_OF_DAY1_FIELD:
2665         // [We computed 'value' above.]
2666         if (value == cal.getMaximum(UCAL_HOUR_OF_DAY) + 1)
2667             value = 0;
2668 
2669         // fall through to set field
2670 
2671     case UDAT_HOUR_OF_DAY0_FIELD:
2672         cal.set(UCAL_HOUR_OF_DAY, value);
2673         return pos.getIndex();
2674 
2675     case UDAT_FRACTIONAL_SECOND_FIELD:
2676         // Fractional seconds left-justify
2677         i = pos.getIndex() - start;
2678         if (i < 3) {
2679             while (i < 3) {
2680                 value *= 10;
2681                 i++;
2682             }
2683         } else {
2684             int32_t a = 1;
2685             while (i > 3) {
2686                 a *= 10;
2687                 i--;
2688             }
2689             value = (value + (a>>1)) / a;
2690         }
2691         cal.set(UCAL_MILLISECOND, value);
2692         return pos.getIndex();
2693 
2694     case UDAT_DOW_LOCAL_FIELD:
2695         if (gotNumber) // i.e., e or ee
2696         {
2697             // [We computed 'value' above.]
2698             cal.set(UCAL_DOW_LOCAL, value);
2699             return pos.getIndex();
2700         }
2701         // else for eee-eeeee fall through to handling of EEE-EEEEE
2702         // fall through, do not break here
2703     case UDAT_DAY_OF_WEEK_FIELD:
2704         {
2705             // Want to be able to parse both short and long forms.
2706             // Try count == 4 (EEEE) first:
2707             int32_t newStart = 0;
2708             if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2709                                       fSymbols->fWeekdays, fSymbols->fWeekdaysCount, NULL, cal)) > 0)
2710                 return newStart;
2711             // EEEE failed, now try EEE
2712             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2713                                    fSymbols->fShortWeekdays, fSymbols->fShortWeekdaysCount, NULL, cal)) > 0)
2714                 return newStart;
2715             // EEE failed, now try EEEEE
2716             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2717                                    fSymbols->fNarrowWeekdays, fSymbols->fNarrowWeekdaysCount, NULL, cal)) > 0)
2718                 return newStart;
2719             else if (!lenient || patternCharIndex == UDAT_DAY_OF_WEEK_FIELD)
2720                 return newStart;
2721             // else we allowing parsing as number, below
2722         }
2723         break;
2724 
2725     case UDAT_STANDALONE_DAY_FIELD:
2726         {
2727             if (gotNumber) // c or cc
2728             {
2729                 // [We computed 'value' above.]
2730                 cal.set(UCAL_DOW_LOCAL, value);
2731                 return pos.getIndex();
2732             }
2733             // Want to be able to parse both short and long forms.
2734             // Try count == 4 (cccc) first:
2735             int32_t newStart = 0;
2736             if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2737                                       fSymbols->fStandaloneWeekdays, fSymbols->fStandaloneWeekdaysCount, NULL, cal)) > 0)
2738                 return newStart;
2739             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2740                                           fSymbols->fStandaloneShortWeekdays, fSymbols->fStandaloneShortWeekdaysCount, NULL, cal)) > 0)
2741                 return newStart;
2742             else if (!lenient)
2743                 return newStart;
2744             // else we allowing parsing as number, below
2745         }
2746         break;
2747 
2748     case UDAT_AM_PM_FIELD:
2749         return matchString(text, start, UCAL_AM_PM, fSymbols->fAmPms, fSymbols->fAmPmsCount, NULL, cal);
2750 
2751     case UDAT_HOUR1_FIELD:
2752         // [We computed 'value' above.]
2753         if (value == cal.getLeastMaximum(UCAL_HOUR)+1)
2754             value = 0;
2755 
2756         // fall through to set field
2757 
2758     case UDAT_HOUR0_FIELD:
2759         cal.set(UCAL_HOUR, value);
2760         return pos.getIndex();
2761 
2762     case UDAT_QUARTER_FIELD:
2763         if (gotNumber) // i.e., Q or QQ.
2764         {
2765             // Don't want to parse the month if it is a string
2766             // while pattern uses numeric style: Q or QQ.
2767             // [We computed 'value' above.]
2768             cal.set(UCAL_MONTH, (value - 1) * 3);
2769             return pos.getIndex();
2770         } else {
2771             // count >= 3 // i.e., QQQ or QQQQ
2772             // Want to be able to parse both short and long forms.
2773             // Try count == 4 first:
2774             int32_t newStart = 0;
2775 
2776             if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2777                                       fSymbols->fQuarters, fSymbols->fQuartersCount, cal)) > 0)
2778                 return newStart;
2779             else if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2780                                           fSymbols->fShortQuarters, fSymbols->fShortQuartersCount, cal)) > 0)
2781                 return newStart;
2782             else if (!lenient)
2783                 return newStart;
2784             // else we allowing parsing as number, below
2785         }
2786         break;
2787 
2788     case UDAT_STANDALONE_QUARTER_FIELD:
2789         if (gotNumber) // i.e., q or qq.
2790         {
2791             // Don't want to parse the month if it is a string
2792             // while pattern uses numeric style: q or q.
2793             // [We computed 'value' above.]
2794             cal.set(UCAL_MONTH, (value - 1) * 3);
2795             return pos.getIndex();
2796         } else {
2797             // count >= 3 // i.e., qqq or qqqq
2798             // Want to be able to parse both short and long forms.
2799             // Try count == 4 first:
2800             int32_t newStart = 0;
2801 
2802             if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2803                                       fSymbols->fStandaloneQuarters, fSymbols->fStandaloneQuartersCount, cal)) > 0)
2804                 return newStart;
2805             else if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2806                                           fSymbols->fStandaloneShortQuarters, fSymbols->fStandaloneShortQuartersCount, cal)) > 0)
2807                 return newStart;
2808             else if (!lenient)
2809                 return newStart;
2810             // else we allowing parsing as number, below
2811         }
2812         break;
2813 
2814     case UDAT_TIMEZONE_FIELD:
2815         {
2816             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2817             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_SPECIFIC_SHORT : UTZFMT_STYLE_SPECIFIC_LONG;
2818             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2819             if (tz != NULL) {
2820                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2821                 cal.adoptTimeZone(tz);
2822                 return pos.getIndex();
2823             }
2824         }
2825         break;
2826     case UDAT_TIMEZONE_RFC_FIELD:
2827         {
2828             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2829             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_RFC822 : ((count == 5) ? UTZFMT_STYLE_ISO8601: UTZFMT_STYLE_LOCALIZED_GMT);
2830             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2831             if (tz != NULL) {
2832                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2833                 cal.adoptTimeZone(tz);
2834                 return pos.getIndex();
2835             }
2836             return -start;
2837         }
2838     case UDAT_TIMEZONE_GENERIC_FIELD:
2839         {
2840             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2841             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_GENERIC_SHORT : UTZFMT_STYLE_GENERIC_LONG;
2842             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2843             if (tz != NULL) {
2844                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2845                 cal.adoptTimeZone(tz);
2846                 return pos.getIndex();
2847             }
2848             return -start;
2849         }
2850     case UDAT_TIMEZONE_SPECIAL_FIELD:
2851         {
2852             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2853             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_SPECIFIC_SHORT : UTZFMT_STYLE_GENERIC_LOCATION;
2854             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2855             if (tz != NULL) {
2856                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2857                 cal.adoptTimeZone(tz);
2858                 return pos.getIndex();
2859             }
2860             return -start;
2861         }
2862 
2863     default:
2864         // Handle "generic" fields
2865         // this is now handled below, outside the switch block
2866         break;
2867     }
2868     // Handle "generic" fields:
2869     // switch default case now handled here (outside switch block) to allow
2870     // parsing of some string fields as digits for lenient case
2871 
2872     int32_t parseStart = pos.getIndex();
2873     const UnicodeString* src;
2874     if (obeyCount) {
2875         if ((start+count) > text.length()) {
2876             return -start;
2877         }
2878         text.extractBetween(0, start + count, temp);
2879         src = &temp;
2880     } else {
2881         src = &text;
2882     }
2883     parseInt(*src, number, pos, allowNegative,currentNumberFormat);
2884     if (pos.getIndex() != parseStart) {
2885         int32_t value = number.getLong();
2886 
2887         // Don't need suffix processing here (as in number processing at the beginning of the function);
2888         // the new fields being handled as numeric values (month, weekdays, quarters) should not have suffixes.
2889 
2890         if (!lenient) {
2891             // Check the range of the value
2892             int32_t bias = gFieldRangeBias[patternCharIndex];
2893             if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
2894                 return -start;
2895             }
2896         }
2897 
2898         // For the following, need to repeat some of the "if (gotNumber)" code above:
2899         // UDAT_[STANDALONE_]MONTH_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_STANDALONE_DAY_FIELD,
2900         // UDAT_[STANDALONE_]QUARTER_FIELD
2901         switch (patternCharIndex) {
2902         case UDAT_MONTH_FIELD:
2903             // See notes under UDAT_MONTH_FIELD case above
2904             if (!strcmp(cal.getType(),"hebrew")) {
2905                 HebrewCalendar *hc = (HebrewCalendar*)&cal;
2906                 if (cal.isSet(UCAL_YEAR)) {
2907                    UErrorCode status = U_ZERO_ERROR;
2908                    if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
2909                        cal.set(UCAL_MONTH, value);
2910                    } else {
2911                        cal.set(UCAL_MONTH, value - 1);
2912                    }
2913                 } else {
2914                     saveHebrewMonth = value;
2915                 }
2916             } else {
2917                 cal.set(UCAL_MONTH, value - 1);
2918             }
2919             break;
2920         case UDAT_STANDALONE_MONTH_FIELD:
2921             cal.set(UCAL_MONTH, value - 1);
2922             break;
2923         case UDAT_DOW_LOCAL_FIELD:
2924         case UDAT_STANDALONE_DAY_FIELD:
2925             cal.set(UCAL_DOW_LOCAL, value);
2926             break;
2927         case UDAT_QUARTER_FIELD:
2928         case UDAT_STANDALONE_QUARTER_FIELD:
2929              cal.set(UCAL_MONTH, (value - 1) * 3);
2930              break;
2931         default:
2932             cal.set(field, value);
2933             break;
2934         }
2935         return pos.getIndex();
2936     }
2937     return -start;
2938 }
2939 
2940 /**
2941  * Parse an integer using fNumberFormat.  This method is semantically
2942  * const, but actually may modify fNumberFormat.
2943  */
parseInt(const UnicodeString & text,Formattable & number,ParsePosition & pos,UBool allowNegative,NumberFormat * fmt) const2944 void SimpleDateFormat::parseInt(const UnicodeString& text,
2945                                 Formattable& number,
2946                                 ParsePosition& pos,
2947                                 UBool allowNegative,
2948                                 NumberFormat *fmt) const {
2949     parseInt(text, number, -1, pos, allowNegative,fmt);
2950 }
2951 
2952 /**
2953  * Parse an integer using fNumberFormat up to maxDigits.
2954  */
parseInt(const UnicodeString & text,Formattable & number,int32_t maxDigits,ParsePosition & pos,UBool allowNegative,NumberFormat * fmt) const2955 void SimpleDateFormat::parseInt(const UnicodeString& text,
2956                                 Formattable& number,
2957                                 int32_t maxDigits,
2958                                 ParsePosition& pos,
2959                                 UBool allowNegative,
2960                                 NumberFormat *fmt) const {
2961     UnicodeString oldPrefix;
2962     DecimalFormat* df = NULL;
2963     if (!allowNegative && (df = dynamic_cast<DecimalFormat*>(fmt)) != NULL) {
2964         df->getNegativePrefix(oldPrefix);
2965         df->setNegativePrefix(UnicodeString(TRUE, SUPPRESS_NEGATIVE_PREFIX, -1));
2966     }
2967     int32_t oldPos = pos.getIndex();
2968     fmt->parse(text, number, pos);
2969     if (df != NULL) {
2970         df->setNegativePrefix(oldPrefix);
2971     }
2972 
2973     if (maxDigits > 0) {
2974         // adjust the result to fit into
2975         // the maxDigits and move the position back
2976         int32_t nDigits = pos.getIndex() - oldPos;
2977         if (nDigits > maxDigits) {
2978             int32_t val = number.getLong();
2979             nDigits -= maxDigits;
2980             while (nDigits > 0) {
2981                 val /= 10;
2982                 nDigits--;
2983             }
2984             pos.setIndex(oldPos + maxDigits);
2985             number.setLong(val);
2986         }
2987     }
2988 }
2989 
2990 //----------------------------------------------------------------------
2991 
translatePattern(const UnicodeString & originalPattern,UnicodeString & translatedPattern,const UnicodeString & from,const UnicodeString & to,UErrorCode & status)2992 void SimpleDateFormat::translatePattern(const UnicodeString& originalPattern,
2993                                         UnicodeString& translatedPattern,
2994                                         const UnicodeString& from,
2995                                         const UnicodeString& to,
2996                                         UErrorCode& status)
2997 {
2998   // run through the pattern and convert any pattern symbols from the version
2999   // in "from" to the corresponding character ion "to".  This code takes
3000   // quoted strings into account (it doesn't try to translate them), and it signals
3001   // an error if a particular "pattern character" doesn't appear in "from".
3002   // Depending on the values of "from" and "to" this can convert from generic
3003   // to localized patterns or localized to generic.
3004   if (U_FAILURE(status))
3005     return;
3006 
3007   translatedPattern.remove();
3008   UBool inQuote = FALSE;
3009   for (int32_t i = 0; i < originalPattern.length(); ++i) {
3010     UChar c = originalPattern[i];
3011     if (inQuote) {
3012       if (c == QUOTE)
3013     inQuote = FALSE;
3014     }
3015     else {
3016       if (c == QUOTE)
3017     inQuote = TRUE;
3018       else if ((c >= 0x0061 /*'a'*/ && c <= 0x007A) /*'z'*/
3019            || (c >= 0x0041 /*'A'*/ && c <= 0x005A /*'Z'*/)) {
3020     int32_t ci = from.indexOf(c);
3021     if (ci == -1) {
3022       status = U_INVALID_FORMAT_ERROR;
3023       return;
3024     }
3025     c = to[ci];
3026       }
3027     }
3028     translatedPattern += c;
3029   }
3030   if (inQuote) {
3031     status = U_INVALID_FORMAT_ERROR;
3032     return;
3033   }
3034 }
3035 
3036 //----------------------------------------------------------------------
3037 
3038 UnicodeString&
toPattern(UnicodeString & result) const3039 SimpleDateFormat::toPattern(UnicodeString& result) const
3040 {
3041     result = fPattern;
3042     return result;
3043 }
3044 
3045 //----------------------------------------------------------------------
3046 
3047 UnicodeString&
toLocalizedPattern(UnicodeString & result,UErrorCode & status) const3048 SimpleDateFormat::toLocalizedPattern(UnicodeString& result,
3049                                      UErrorCode& status) const
3050 {
3051     translatePattern(fPattern, result,
3052                      UnicodeString(DateFormatSymbols::getPatternUChars()),
3053                      fSymbols->fLocalPatternChars, status);
3054     return result;
3055 }
3056 
3057 //----------------------------------------------------------------------
3058 
3059 void
applyPattern(const UnicodeString & pattern)3060 SimpleDateFormat::applyPattern(const UnicodeString& pattern)
3061 {
3062     fPattern = pattern;
3063 }
3064 
3065 //----------------------------------------------------------------------
3066 
3067 void
applyLocalizedPattern(const UnicodeString & pattern,UErrorCode & status)3068 SimpleDateFormat::applyLocalizedPattern(const UnicodeString& pattern,
3069                                         UErrorCode &status)
3070 {
3071     translatePattern(pattern, fPattern,
3072                      fSymbols->fLocalPatternChars,
3073                      UnicodeString(DateFormatSymbols::getPatternUChars()), status);
3074 }
3075 
3076 //----------------------------------------------------------------------
3077 
3078 const DateFormatSymbols*
getDateFormatSymbols() const3079 SimpleDateFormat::getDateFormatSymbols() const
3080 {
3081     return fSymbols;
3082 }
3083 
3084 //----------------------------------------------------------------------
3085 
3086 void
adoptDateFormatSymbols(DateFormatSymbols * newFormatSymbols)3087 SimpleDateFormat::adoptDateFormatSymbols(DateFormatSymbols* newFormatSymbols)
3088 {
3089     delete fSymbols;
3090     fSymbols = newFormatSymbols;
3091 }
3092 
3093 //----------------------------------------------------------------------
3094 void
setDateFormatSymbols(const DateFormatSymbols & newFormatSymbols)3095 SimpleDateFormat::setDateFormatSymbols(const DateFormatSymbols& newFormatSymbols)
3096 {
3097     delete fSymbols;
3098     fSymbols = new DateFormatSymbols(newFormatSymbols);
3099 }
3100 
3101 //----------------------------------------------------------------------
3102 const TimeZoneFormat*
getTimeZoneFormat(void) const3103 SimpleDateFormat::getTimeZoneFormat(void) const {
3104     return (const TimeZoneFormat*)tzFormat();
3105 }
3106 
3107 //----------------------------------------------------------------------
3108 void
adoptTimeZoneFormat(TimeZoneFormat * timeZoneFormatToAdopt)3109 SimpleDateFormat::adoptTimeZoneFormat(TimeZoneFormat* timeZoneFormatToAdopt)
3110 {
3111     delete fTimeZoneFormat;
3112     fTimeZoneFormat = timeZoneFormatToAdopt;
3113 }
3114 
3115 //----------------------------------------------------------------------
3116 void
setTimeZoneFormat(const TimeZoneFormat & newTimeZoneFormat)3117 SimpleDateFormat::setTimeZoneFormat(const TimeZoneFormat& newTimeZoneFormat)
3118 {
3119     delete fTimeZoneFormat;
3120     fTimeZoneFormat = new TimeZoneFormat(newTimeZoneFormat);
3121 }
3122 
3123 //----------------------------------------------------------------------
3124 
3125 
adoptCalendar(Calendar * calendarToAdopt)3126 void SimpleDateFormat::adoptCalendar(Calendar* calendarToAdopt)
3127 {
3128   UErrorCode status = U_ZERO_ERROR;
3129   DateFormat::adoptCalendar(calendarToAdopt);
3130   delete fSymbols;
3131   fSymbols=NULL;
3132   initializeSymbols(fLocale, fCalendar, status);  // we need new symbols
3133   initializeDefaultCentury();  // we need a new century (possibly)
3134 }
3135 
3136 
3137 //----------------------------------------------------------------------
3138 
3139 
setContext(UDisplayContext value,UErrorCode & status)3140 void SimpleDateFormat::setContext(UDisplayContext value, UErrorCode& status)
3141 {
3142     if (U_FAILURE(status))
3143         return;
3144     if ( (UDisplayContextType)((uint32_t)value >> 8) == UDISPCTX_TYPE_CAPITALIZATION ) {
3145         fCapitalizationContext = value;
3146     } else {
3147         status = U_ILLEGAL_ARGUMENT_ERROR;
3148    }
3149 }
3150 
3151 
3152 //----------------------------------------------------------------------
3153 
3154 
getContext(UDisplayContextType type,UErrorCode & status) const3155 UDisplayContext SimpleDateFormat::getContext(UDisplayContextType type, UErrorCode& status) const
3156 {
3157     if (U_FAILURE(status))
3158         return (UDisplayContext)0;
3159     if (type != UDISPCTX_TYPE_CAPITALIZATION) {
3160         status = U_ILLEGAL_ARGUMENT_ERROR;
3161         return (UDisplayContext)0;
3162     }
3163     return fCapitalizationContext;
3164 }
3165 
3166 
3167 //----------------------------------------------------------------------
3168 
3169 
3170 UBool
isFieldUnitIgnored(UCalendarDateFields field) const3171 SimpleDateFormat::isFieldUnitIgnored(UCalendarDateFields field) const {
3172     return isFieldUnitIgnored(fPattern, field);
3173 }
3174 
3175 
3176 UBool
isFieldUnitIgnored(const UnicodeString & pattern,UCalendarDateFields field)3177 SimpleDateFormat::isFieldUnitIgnored(const UnicodeString& pattern,
3178                                      UCalendarDateFields field) {
3179     int32_t fieldLevel = fgCalendarFieldToLevel[field];
3180     int32_t level;
3181     UChar ch;
3182     UBool inQuote = FALSE;
3183     UChar prevCh = 0;
3184     int32_t count = 0;
3185 
3186     for (int32_t i = 0; i < pattern.length(); ++i) {
3187         ch = pattern[i];
3188         if (ch != prevCh && count > 0) {
3189             level = fgPatternCharToLevel[prevCh - PATTERN_CHAR_BASE];
3190             // the larger the level, the smaller the field unit.
3191             if ( fieldLevel <= level ) {
3192                 return FALSE;
3193             }
3194             count = 0;
3195         }
3196         if (ch == QUOTE) {
3197             if ((i+1) < pattern.length() && pattern[i+1] == QUOTE) {
3198                 ++i;
3199             } else {
3200                 inQuote = ! inQuote;
3201             }
3202         }
3203         else if ( ! inQuote && ((ch >= 0x0061 /*'a'*/ && ch <= 0x007A /*'z'*/)
3204                     || (ch >= 0x0041 /*'A'*/ && ch <= 0x005A /*'Z'*/))) {
3205             prevCh = ch;
3206             ++count;
3207         }
3208     }
3209     if ( count > 0 ) {
3210         // last item
3211         level = fgPatternCharToLevel[prevCh - PATTERN_CHAR_BASE];
3212             if ( fieldLevel <= level ) {
3213                 return FALSE;
3214             }
3215     }
3216     return TRUE;
3217 }
3218 
3219 //----------------------------------------------------------------------
3220 
3221 const Locale&
getSmpFmtLocale(void) const3222 SimpleDateFormat::getSmpFmtLocale(void) const {
3223     return fLocale;
3224 }
3225 
3226 //----------------------------------------------------------------------
3227 
3228 int32_t
checkIntSuffix(const UnicodeString & text,int32_t start,int32_t patLoc,UBool isNegative) const3229 SimpleDateFormat::checkIntSuffix(const UnicodeString& text, int32_t start,
3230                                  int32_t patLoc, UBool isNegative) const {
3231     // local variables
3232     UnicodeString suf;
3233     int32_t patternMatch;
3234     int32_t textPreMatch;
3235     int32_t textPostMatch;
3236 
3237     // check that we are still in range
3238     if ( (start > text.length()) ||
3239          (start < 0) ||
3240          (patLoc < 0) ||
3241          (patLoc > fPattern.length())) {
3242         // out of range, don't advance location in text
3243         return start;
3244     }
3245 
3246     // get the suffix
3247     DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(fNumberFormat);
3248     if (decfmt != NULL) {
3249         if (isNegative) {
3250             suf = decfmt->getNegativeSuffix(suf);
3251         }
3252         else {
3253             suf = decfmt->getPositiveSuffix(suf);
3254         }
3255     }
3256 
3257     // check for suffix
3258     if (suf.length() <= 0) {
3259         return start;
3260     }
3261 
3262     // check suffix will be encountered in the pattern
3263     patternMatch = compareSimpleAffix(suf,fPattern,patLoc);
3264 
3265     // check if a suffix will be encountered in the text
3266     textPreMatch = compareSimpleAffix(suf,text,start);
3267 
3268     // check if a suffix was encountered in the text
3269     textPostMatch = compareSimpleAffix(suf,text,start-suf.length());
3270 
3271     // check for suffix match
3272     if ((textPreMatch >= 0) && (patternMatch >= 0) && (textPreMatch == patternMatch)) {
3273         return start;
3274     }
3275     else if ((textPostMatch >= 0) && (patternMatch >= 0) && (textPostMatch == patternMatch)) {
3276         return  start - suf.length();
3277     }
3278 
3279     // should not get here
3280     return start;
3281 }
3282 
3283 //----------------------------------------------------------------------
3284 
3285 int32_t
compareSimpleAffix(const UnicodeString & affix,const UnicodeString & input,int32_t pos) const3286 SimpleDateFormat::compareSimpleAffix(const UnicodeString& affix,
3287                    const UnicodeString& input,
3288                    int32_t pos) const {
3289     int32_t start = pos;
3290     for (int32_t i=0; i<affix.length(); ) {
3291         UChar32 c = affix.char32At(i);
3292         int32_t len = U16_LENGTH(c);
3293         if (PatternProps::isWhiteSpace(c)) {
3294             // We may have a pattern like: \u200F \u0020
3295             //        and input text like: \u200F \u0020
3296             // Note that U+200F and U+0020 are Pattern_White_Space but only
3297             // U+0020 is UWhiteSpace.  So we have to first do a direct
3298             // match of the run of Pattern_White_Space in the pattern,
3299             // then match any extra characters.
3300             UBool literalMatch = FALSE;
3301             while (pos < input.length() &&
3302                    input.char32At(pos) == c) {
3303                 literalMatch = TRUE;
3304                 i += len;
3305                 pos += len;
3306                 if (i == affix.length()) {
3307                     break;
3308                 }
3309                 c = affix.char32At(i);
3310                 len = U16_LENGTH(c);
3311                 if (!PatternProps::isWhiteSpace(c)) {
3312                     break;
3313                 }
3314             }
3315 
3316             // Advance over run in pattern
3317             i = skipPatternWhiteSpace(affix, i);
3318 
3319             // Advance over run in input text
3320             // Must see at least one white space char in input,
3321             // unless we've already matched some characters literally.
3322             int32_t s = pos;
3323             pos = skipUWhiteSpace(input, pos);
3324             if (pos == s && !literalMatch) {
3325                 return -1;
3326             }
3327 
3328             // If we skip UWhiteSpace in the input text, we need to skip it in the pattern.
3329             // Otherwise, the previous lines may have skipped over text (such as U+00A0) that
3330             // is also in the affix.
3331             i = skipUWhiteSpace(affix, i);
3332         } else {
3333             if (pos < input.length() &&
3334                 input.char32At(pos) == c) {
3335                 i += len;
3336                 pos += len;
3337             } else {
3338                 return -1;
3339             }
3340         }
3341     }
3342     return pos - start;
3343 }
3344 
3345 //----------------------------------------------------------------------
3346 
3347 int32_t
skipPatternWhiteSpace(const UnicodeString & text,int32_t pos) const3348 SimpleDateFormat::skipPatternWhiteSpace(const UnicodeString& text, int32_t pos) const {
3349     const UChar* s = text.getBuffer();
3350     return (int32_t)(PatternProps::skipWhiteSpace(s + pos, text.length() - pos) - s);
3351 }
3352 
3353 //----------------------------------------------------------------------
3354 
3355 int32_t
skipUWhiteSpace(const UnicodeString & text,int32_t pos) const3356 SimpleDateFormat::skipUWhiteSpace(const UnicodeString& text, int32_t pos) const {
3357     while (pos < text.length()) {
3358         UChar32 c = text.char32At(pos);
3359         if (!u_isUWhiteSpace(c)) {
3360             break;
3361         }
3362         pos += U16_LENGTH(c);
3363     }
3364     return pos;
3365 }
3366 
3367 //----------------------------------------------------------------------
3368 
3369 // Lazy TimeZoneFormat instantiation, semantically const.
3370 TimeZoneFormat *
tzFormat() const3371 SimpleDateFormat::tzFormat() const {
3372     if (fTimeZoneFormat == NULL) {
3373         umtx_lock(&LOCK);
3374         {
3375             if (fTimeZoneFormat == NULL) {
3376                 UErrorCode status = U_ZERO_ERROR;
3377                 TimeZoneFormat *tzfmt = TimeZoneFormat::createInstance(fLocale, status);
3378                 if (U_FAILURE(status)) {
3379                     return NULL;
3380                 }
3381 
3382                 const_cast<SimpleDateFormat *>(this)->fTimeZoneFormat = tzfmt;
3383             }
3384         }
3385         umtx_unlock(&LOCK);
3386     }
3387     return fTimeZoneFormat;
3388 }
3389 
3390 U_NAMESPACE_END
3391 
3392 #endif /* #if !UCONFIG_NO_FORMATTING */
3393 
3394 //eof
3395