1 /*
2 *******************************************************************************
3 * Copyright (C) 1997-2015, 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 #include "unicode/smpdtfmt.h"
33 #include "unicode/dtfmtsym.h"
34 #include "unicode/ures.h"
35 #include "unicode/msgfmt.h"
36 #include "unicode/calendar.h"
37 #include "unicode/gregocal.h"
38 #include "unicode/timezone.h"
39 #include "unicode/decimfmt.h"
40 #include "unicode/dcfmtsym.h"
41 #include "unicode/uchar.h"
42 #include "unicode/uniset.h"
43 #include "unicode/ustring.h"
44 #include "unicode/basictz.h"
45 #include "unicode/simpletz.h"
46 #include "unicode/rbtz.h"
47 #include "unicode/tzfmt.h"
48 #include "unicode/utf16.h"
49 #include "unicode/vtzone.h"
50 #include "unicode/udisplaycontext.h"
51 #include "unicode/brkiter.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 #include "sharednumberformat.h"
64 #include "ustr_imp.h"
65 
66 #if defined( U_DEBUG_CALSVC ) || defined (U_DEBUG_CAL)
67 #include <stdio.h>
68 #endif
69 
70 // *****************************************************************************
71 // class SimpleDateFormat
72 // *****************************************************************************
73 
74 U_NAMESPACE_BEGIN
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     UDAT_RELATED_YEAR_FIELD };
130 static const int8_t kDateFieldsCount = 16;
131 
132 static const UDateFormatField kTimeFields[] = {
133     UDAT_HOUR_OF_DAY1_FIELD,
134     UDAT_HOUR_OF_DAY0_FIELD,
135     UDAT_MINUTE_FIELD,
136     UDAT_SECOND_FIELD,
137     UDAT_FRACTIONAL_SECOND_FIELD,
138     UDAT_HOUR1_FIELD,
139     UDAT_HOUR0_FIELD,
140     UDAT_MILLISECONDS_IN_DAY_FIELD,
141     UDAT_TIMEZONE_RFC_FIELD,
142     UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD };
143 static const int8_t kTimeFieldsCount = 10;
144 
145 
146 // This is a pattern-of-last-resort used when we can't load a usable pattern out
147 // of a resource.
148 static const UChar gDefaultPattern[] =
149 {
150     0x79, 0x79, 0x79, 0x79, 0x4D, 0x4D, 0x64, 0x64, 0x20, 0x68, 0x68, 0x3A, 0x6D, 0x6D, 0x20, 0x61, 0
151 };  /* "yyyyMMdd hh:mm a" */
152 
153 // This prefix is designed to NEVER MATCH real text, in order to
154 // suppress the parsing of negative numbers.  Adjust as needed (if
155 // this becomes valid Unicode).
156 static const UChar SUPPRESS_NEGATIVE_PREFIX[] = {0xAB00, 0};
157 
158 /**
159  * These are the tags we expect to see in normal resource bundle files associated
160  * with a locale.
161  */
162 static const char gDateTimePatternsTag[]="DateTimePatterns";
163 
164 //static const UChar gEtcUTC[] = {0x45, 0x74, 0x63, 0x2F, 0x55, 0x54, 0x43, 0x00}; // "Etc/UTC"
165 static const UChar QUOTE = 0x27; // Single quote
166 
167 /*
168  * The field range check bias for each UDateFormatField.
169  * The bias is added to the minimum and maximum values
170  * before they are compared to the parsed number.
171  * For example, the calendar stores zero-based month numbers
172  * but the parsed month numbers start at 1, so the bias is 1.
173  *
174  * A value of -1 means that the value is not checked.
175  */
176 static const int32_t gFieldRangeBias[] = {
177     -1,  // 'G' - UDAT_ERA_FIELD
178     -1,  // 'y' - UDAT_YEAR_FIELD
179      1,  // 'M' - UDAT_MONTH_FIELD
180      0,  // 'd' - UDAT_DATE_FIELD
181     -1,  // 'k' - UDAT_HOUR_OF_DAY1_FIELD
182     -1,  // 'H' - UDAT_HOUR_OF_DAY0_FIELD
183      0,  // 'm' - UDAT_MINUTE_FIELD
184      0,  // 's' - UDAT_SECOND_FIELD
185     -1,  // 'S' - UDAT_FRACTIONAL_SECOND_FIELD (0-999?)
186     -1,  // 'E' - UDAT_DAY_OF_WEEK_FIELD (1-7?)
187     -1,  // 'D' - UDAT_DAY_OF_YEAR_FIELD (1 - 366?)
188     -1,  // 'F' - UDAT_DAY_OF_WEEK_IN_MONTH_FIELD (1-5?)
189     -1,  // 'w' - UDAT_WEEK_OF_YEAR_FIELD (1-52?)
190     -1,  // 'W' - UDAT_WEEK_OF_MONTH_FIELD (1-5?)
191     -1,  // 'a' - UDAT_AM_PM_FIELD
192     -1,  // 'h' - UDAT_HOUR1_FIELD
193     -1,  // 'K' - UDAT_HOUR0_FIELD
194     -1,  // 'z' - UDAT_TIMEZONE_FIELD
195     -1,  // 'Y' - UDAT_YEAR_WOY_FIELD
196     -1,  // 'e' - UDAT_DOW_LOCAL_FIELD
197     -1,  // 'u' - UDAT_EXTENDED_YEAR_FIELD
198     -1,  // 'g' - UDAT_JULIAN_DAY_FIELD
199     -1,  // 'A' - UDAT_MILLISECONDS_IN_DAY_FIELD
200     -1,  // 'Z' - UDAT_TIMEZONE_RFC_FIELD
201     -1,  // 'v' - UDAT_TIMEZONE_GENERIC_FIELD
202      0,  // 'c' - UDAT_STANDALONE_DAY_FIELD
203      1,  // 'L' - UDAT_STANDALONE_MONTH_FIELD
204     -1,  // 'Q' - UDAT_QUARTER_FIELD (1-4?)
205     -1,  // 'q' - UDAT_STANDALONE_QUARTER_FIELD
206     -1,  // 'V' - UDAT_TIMEZONE_SPECIAL_FIELD
207     -1,  // 'U' - UDAT_YEAR_NAME_FIELD
208     -1,  // 'O' - UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD
209     -1,  // 'X' - UDAT_TIMEZONE_ISO_FIELD
210     -1,  // 'x' - UDAT_TIMEZONE_ISO_LOCAL_FIELD
211     -1,  // 'r' - UDAT_RELATED_YEAR_FIELD
212 #if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
213     -1,  // ':' - UDAT_TIME_SEPARATOR_FIELD
214 #else
215     -1,  // (no pattern character currently) - UDAT_TIME_SEPARATOR_FIELD
216 #endif
217 };
218 
219 // When calendar uses hebr numbering (i.e. he@calendar=hebrew),
220 // offset the years within the current millenium down to 1-999
221 static const int32_t HEBREW_CAL_CUR_MILLENIUM_START_YEAR = 5000;
222 static const int32_t HEBREW_CAL_CUR_MILLENIUM_END_YEAR = 6000;
223 
224 static UMutex LOCK = U_MUTEX_INITIALIZER;
225 
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)226 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)
227 
228 SimpleDateFormat::NSOverride::~NSOverride() {
229     if (snf != NULL) {
230         snf->removeRef();
231     }
232 }
233 
234 
free()235 void SimpleDateFormat::NSOverride::free() {
236     NSOverride *cur = this;
237     while (cur) {
238         NSOverride *next = cur->next;
239         delete cur;
240         cur = next;
241     }
242 }
243 
244 // no matter what the locale's default number format looked like, we want
245 // to modify it so that it doesn't use thousands separators, doesn't always
246 // show the decimal point, and recognizes integers only when parsing
fixNumberFormatForDates(NumberFormat & nf)247 static void fixNumberFormatForDates(NumberFormat &nf) {
248     nf.setGroupingUsed(FALSE);
249     DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(&nf);
250     if (decfmt != NULL) {
251         decfmt->setDecimalSeparatorAlwaysShown(FALSE);
252     }
253     nf.setParseIntegerOnly(TRUE);
254     nf.setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
255 }
256 
createSharedNumberFormat(NumberFormat * nfToAdopt)257 static const SharedNumberFormat *createSharedNumberFormat(
258         NumberFormat *nfToAdopt) {
259     fixNumberFormatForDates(*nfToAdopt);
260     const SharedNumberFormat *result = new SharedNumberFormat(nfToAdopt);
261     if (result == NULL) {
262         delete nfToAdopt;
263     }
264     return result;
265 }
266 
createSharedNumberFormat(const Locale & loc,UErrorCode & status)267 static const SharedNumberFormat *createSharedNumberFormat(
268         const Locale &loc, UErrorCode &status) {
269     NumberFormat *nf = NumberFormat::createInstance(loc, status);
270     if (U_FAILURE(status)) {
271         return NULL;
272     }
273     const SharedNumberFormat *result = createSharedNumberFormat(nf);
274     if (result == NULL) {
275         status = U_MEMORY_ALLOCATION_ERROR;
276     }
277     return result;
278 }
279 
allocSharedNumberFormatters()280 static const SharedNumberFormat **allocSharedNumberFormatters() {
281     const SharedNumberFormat **result = (const SharedNumberFormat**)
282             uprv_malloc(UDAT_FIELD_COUNT * sizeof(const SharedNumberFormat*));
283     if (result == NULL) {
284         return NULL;
285     }
286     for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
287         result[i] = NULL;
288     }
289     return result;
290 }
291 
freeSharedNumberFormatters(const SharedNumberFormat ** list)292 static void freeSharedNumberFormatters(const SharedNumberFormat ** list) {
293     for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
294         SharedObject::clearPtr(list[i]);
295     }
296     uprv_free(list);
297 }
298 
getNumberFormatByIndex(UDateFormatField index) const299 const NumberFormat *SimpleDateFormat::getNumberFormatByIndex(
300         UDateFormatField index) const {
301     if (fSharedNumberFormatters == NULL ||
302         fSharedNumberFormatters[index] == NULL) {
303         return fNumberFormat;
304     }
305     return &(**fSharedNumberFormatters[index]);
306 }
307 
308 class SimpleDateFormatMutableNFNode {
309  public:
310     const NumberFormat *key;
311     NumberFormat *value;
SimpleDateFormatMutableNFNode()312     SimpleDateFormatMutableNFNode()
313             : key(NULL), value(NULL) { }
~SimpleDateFormatMutableNFNode()314     ~SimpleDateFormatMutableNFNode() {
315         delete value;
316     }
317  private:
318     SimpleDateFormatMutableNFNode(const SimpleDateFormatMutableNFNode &);
319     SimpleDateFormatMutableNFNode &operator=(const SimpleDateFormatMutableNFNode &);
320 };
321 
322 // Single threaded cache of non const NumberFormats. Designed to be stack
323 // allocated and used for a single format call.
324 class SimpleDateFormatMutableNFs : public UMemory {
325  public:
SimpleDateFormatMutableNFs()326     SimpleDateFormatMutableNFs() {
327     }
328 
329     // Returns a non-const clone of nf which can be safely modified.
330     // Subsequent calls with same nf will return the same non-const clone.
331     // This object maintains ownership of all returned non-const
332     // NumberFormat objects. On memory allocation error returns NULL.
333     // Caller must check for NULL return value.
get(const NumberFormat * nf)334     NumberFormat *get(const NumberFormat *nf) {
335         if (nf == NULL) {
336             return NULL;
337         }
338         int32_t idx = 0;
339         while (nodes[idx].value) {
340             if (nf == nodes[idx].key) {
341                 return nodes[idx].value;
342             }
343             ++idx;
344         }
345         U_ASSERT(idx < UDAT_FIELD_COUNT);
346         nodes[idx].key = nf;
347         nodes[idx].value = (NumberFormat *) nf->clone();
348         return nodes[idx].value;
349     }
350  private:
351     // +1 extra for sentinel. If each field had its own NumberFormat, this
352     // cache would have to allocate UDAT_FIELD_COUNT mutable versions worst
353     // case.
354     SimpleDateFormatMutableNFNode nodes[UDAT_FIELD_COUNT + 1];
355     SimpleDateFormatMutableNFs(const SimpleDateFormatMutableNFs &);
356     SimpleDateFormatMutableNFs &operator=(const SimpleDateFormatMutableNFs &);
357 };
358 
359 //----------------------------------------------------------------------
360 
~SimpleDateFormat()361 SimpleDateFormat::~SimpleDateFormat()
362 {
363     delete fSymbols;
364     if (fSharedNumberFormatters) {
365         freeSharedNumberFormatters(fSharedNumberFormatters);
366     }
367     if (fTimeZoneFormat) {
368         delete fTimeZoneFormat;
369     }
370 
371 #if !UCONFIG_NO_BREAK_ITERATION
372     delete fCapitalizationBrkIter;
373 #endif
374 }
375 
376 //----------------------------------------------------------------------
377 
SimpleDateFormat(UErrorCode & status)378 SimpleDateFormat::SimpleDateFormat(UErrorCode& status)
379   :   fLocale(Locale::getDefault()),
380       fSymbols(NULL),
381       fTimeZoneFormat(NULL),
382       fSharedNumberFormatters(NULL),
383       fCapitalizationBrkIter(NULL)
384 {
385     initializeBooleanAttributes();
386     construct(kShort, (EStyle) (kShort + kDateOffset), fLocale, status);
387     initializeDefaultCentury();
388 }
389 
390 //----------------------------------------------------------------------
391 
SimpleDateFormat(const UnicodeString & pattern,UErrorCode & status)392 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
393                                    UErrorCode &status)
394 :   fPattern(pattern),
395     fLocale(Locale::getDefault()),
396     fSymbols(NULL),
397     fTimeZoneFormat(NULL),
398     fSharedNumberFormatters(NULL),
399     fCapitalizationBrkIter(NULL)
400 {
401     fDateOverride.setToBogus();
402     fTimeOverride.setToBogus();
403     initializeBooleanAttributes();
404     initializeCalendar(NULL,fLocale,status);
405     fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
406     initialize(fLocale, status);
407     initializeDefaultCentury();
408 
409 }
410 //----------------------------------------------------------------------
411 
SimpleDateFormat(const UnicodeString & pattern,const UnicodeString & override,UErrorCode & status)412 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
413                                    const UnicodeString& override,
414                                    UErrorCode &status)
415 :   fPattern(pattern),
416     fLocale(Locale::getDefault()),
417     fSymbols(NULL),
418     fTimeZoneFormat(NULL),
419     fSharedNumberFormatters(NULL),
420     fCapitalizationBrkIter(NULL)
421 {
422     fDateOverride.setTo(override);
423     fTimeOverride.setToBogus();
424     initializeBooleanAttributes();
425     initializeCalendar(NULL,fLocale,status);
426     fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
427     initialize(fLocale, status);
428     initializeDefaultCentury();
429 
430     processOverrideString(fLocale,override,kOvrStrBoth,status);
431 
432 }
433 
434 //----------------------------------------------------------------------
435 
SimpleDateFormat(const UnicodeString & pattern,const Locale & locale,UErrorCode & status)436 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
437                                    const Locale& locale,
438                                    UErrorCode& status)
439 :   fPattern(pattern),
440     fLocale(locale),
441     fTimeZoneFormat(NULL),
442     fSharedNumberFormatters(NULL),
443     fCapitalizationBrkIter(NULL)
444 {
445 
446     fDateOverride.setToBogus();
447     fTimeOverride.setToBogus();
448     initializeBooleanAttributes();
449 
450     initializeCalendar(NULL,fLocale,status);
451     fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
452     initialize(fLocale, status);
453     initializeDefaultCentury();
454 }
455 
456 //----------------------------------------------------------------------
457 
SimpleDateFormat(const UnicodeString & pattern,const UnicodeString & override,const Locale & locale,UErrorCode & status)458 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
459                                    const UnicodeString& override,
460                                    const Locale& locale,
461                                    UErrorCode& status)
462 :   fPattern(pattern),
463     fLocale(locale),
464     fTimeZoneFormat(NULL),
465     fSharedNumberFormatters(NULL),
466     fCapitalizationBrkIter(NULL)
467 {
468 
469     fDateOverride.setTo(override);
470     fTimeOverride.setToBogus();
471     initializeBooleanAttributes();
472 
473     initializeCalendar(NULL,fLocale,status);
474     fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
475     initialize(fLocale, status);
476     initializeDefaultCentury();
477 
478     processOverrideString(locale,override,kOvrStrBoth,status);
479 
480 }
481 
482 //----------------------------------------------------------------------
483 
SimpleDateFormat(const UnicodeString & pattern,DateFormatSymbols * symbolsToAdopt,UErrorCode & status)484 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
485                                    DateFormatSymbols* symbolsToAdopt,
486                                    UErrorCode& status)
487 :   fPattern(pattern),
488     fLocale(Locale::getDefault()),
489     fSymbols(symbolsToAdopt),
490     fTimeZoneFormat(NULL),
491     fSharedNumberFormatters(NULL),
492     fCapitalizationBrkIter(NULL)
493 {
494 
495     fDateOverride.setToBogus();
496     fTimeOverride.setToBogus();
497     initializeBooleanAttributes();
498 
499     initializeCalendar(NULL,fLocale,status);
500     initialize(fLocale, status);
501     initializeDefaultCentury();
502 }
503 
504 //----------------------------------------------------------------------
505 
SimpleDateFormat(const UnicodeString & pattern,const DateFormatSymbols & symbols,UErrorCode & status)506 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
507                                    const DateFormatSymbols& symbols,
508                                    UErrorCode& status)
509 :   fPattern(pattern),
510     fLocale(Locale::getDefault()),
511     fSymbols(new DateFormatSymbols(symbols)),
512     fTimeZoneFormat(NULL),
513     fSharedNumberFormatters(NULL),
514     fCapitalizationBrkIter(NULL)
515 {
516 
517     fDateOverride.setToBogus();
518     fTimeOverride.setToBogus();
519     initializeBooleanAttributes();
520 
521     initializeCalendar(NULL, fLocale, status);
522     initialize(fLocale, status);
523     initializeDefaultCentury();
524 }
525 
526 //----------------------------------------------------------------------
527 
528 // Not for public consumption; used by DateFormat
SimpleDateFormat(EStyle timeStyle,EStyle dateStyle,const Locale & locale,UErrorCode & status)529 SimpleDateFormat::SimpleDateFormat(EStyle timeStyle,
530                                    EStyle dateStyle,
531                                    const Locale& locale,
532                                    UErrorCode& status)
533 :   fLocale(locale),
534     fSymbols(NULL),
535     fTimeZoneFormat(NULL),
536     fSharedNumberFormatters(NULL),
537     fCapitalizationBrkIter(NULL)
538 {
539     initializeBooleanAttributes();
540     construct(timeStyle, dateStyle, fLocale, status);
541     if(U_SUCCESS(status)) {
542       initializeDefaultCentury();
543     }
544 }
545 
546 //----------------------------------------------------------------------
547 
548 /**
549  * Not for public consumption; used by DateFormat.  This constructor
550  * never fails.  If the resource data is not available, it uses the
551  * the last resort symbols.
552  */
SimpleDateFormat(const Locale & locale,UErrorCode & status)553 SimpleDateFormat::SimpleDateFormat(const Locale& locale,
554                                    UErrorCode& status)
555 :   fPattern(gDefaultPattern),
556     fLocale(locale),
557     fSymbols(NULL),
558     fTimeZoneFormat(NULL),
559     fSharedNumberFormatters(NULL),
560     fCapitalizationBrkIter(NULL)
561 {
562     if (U_FAILURE(status)) return;
563     initializeBooleanAttributes();
564     initializeCalendar(NULL, fLocale, status);
565     fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
566     if (U_FAILURE(status))
567     {
568         status = U_ZERO_ERROR;
569         delete fSymbols;
570         // This constructor doesn't fail; it uses last resort data
571         fSymbols = new DateFormatSymbols(status);
572         /* test for NULL */
573         if (fSymbols == 0) {
574             status = U_MEMORY_ALLOCATION_ERROR;
575             return;
576         }
577     }
578 
579     fDateOverride.setToBogus();
580     fTimeOverride.setToBogus();
581 
582     initialize(fLocale, status);
583     if(U_SUCCESS(status)) {
584       initializeDefaultCentury();
585     }
586 }
587 
588 //----------------------------------------------------------------------
589 
SimpleDateFormat(const SimpleDateFormat & other)590 SimpleDateFormat::SimpleDateFormat(const SimpleDateFormat& other)
591 :   DateFormat(other),
592     fLocale(other.fLocale),
593     fSymbols(NULL),
594     fTimeZoneFormat(NULL),
595     fSharedNumberFormatters(NULL),
596     fCapitalizationBrkIter(NULL)
597 {
598     initializeBooleanAttributes();
599     *this = other;
600 }
601 
602 //----------------------------------------------------------------------
603 
operator =(const SimpleDateFormat & other)604 SimpleDateFormat& SimpleDateFormat::operator=(const SimpleDateFormat& other)
605 {
606     if (this == &other) {
607         return *this;
608     }
609     DateFormat::operator=(other);
610     fDateOverride = other.fDateOverride;
611     fTimeOverride = other.fTimeOverride;
612 
613     delete fSymbols;
614     fSymbols = NULL;
615 
616     if (other.fSymbols)
617         fSymbols = new DateFormatSymbols(*other.fSymbols);
618 
619     fDefaultCenturyStart         = other.fDefaultCenturyStart;
620     fDefaultCenturyStartYear     = other.fDefaultCenturyStartYear;
621     fHaveDefaultCentury          = other.fHaveDefaultCentury;
622 
623     fPattern = other.fPattern;
624 
625     // TimeZoneFormat in ICU4C only depends on a locale for now
626     if (fLocale != other.fLocale) {
627         delete fTimeZoneFormat;
628         fTimeZoneFormat = NULL; // forces lazy instantiation with the other locale
629         fLocale = other.fLocale;
630     }
631 
632 #if !UCONFIG_NO_BREAK_ITERATION
633     if (other.fCapitalizationBrkIter != NULL) {
634         fCapitalizationBrkIter = (other.fCapitalizationBrkIter)->clone();
635     }
636 #endif
637 
638     if (fSharedNumberFormatters != NULL) {
639         freeSharedNumberFormatters(fSharedNumberFormatters);
640         fSharedNumberFormatters = NULL;
641     }
642     if (other.fSharedNumberFormatters != NULL) {
643         fSharedNumberFormatters = allocSharedNumberFormatters();
644         if (fSharedNumberFormatters) {
645             for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
646                 SharedObject::copyPtr(
647                         other.fSharedNumberFormatters[i],
648                         fSharedNumberFormatters[i]);
649             }
650         }
651     }
652 
653     return *this;
654 }
655 
656 //----------------------------------------------------------------------
657 
658 Format*
clone() const659 SimpleDateFormat::clone() const
660 {
661     return new SimpleDateFormat(*this);
662 }
663 
664 //----------------------------------------------------------------------
665 
666 UBool
operator ==(const Format & other) const667 SimpleDateFormat::operator==(const Format& other) const
668 {
669     if (DateFormat::operator==(other)) {
670         // The DateFormat::operator== check for fCapitalizationContext equality above
671         //   is sufficient to check equality of all derived context-related data.
672         // DateFormat::operator== guarantees following cast is safe
673         SimpleDateFormat* that = (SimpleDateFormat*)&other;
674         return (fPattern             == that->fPattern &&
675                 fSymbols             != NULL && // Check for pathological object
676                 that->fSymbols       != NULL && // Check for pathological object
677                 *fSymbols            == *that->fSymbols &&
678                 fHaveDefaultCentury  == that->fHaveDefaultCentury &&
679                 fDefaultCenturyStart == that->fDefaultCenturyStart);
680     }
681     return FALSE;
682 }
683 
684 //----------------------------------------------------------------------
685 
construct(EStyle timeStyle,EStyle dateStyle,const Locale & locale,UErrorCode & status)686 void SimpleDateFormat::construct(EStyle timeStyle,
687                                  EStyle dateStyle,
688                                  const Locale& locale,
689                                  UErrorCode& status)
690 {
691     // called by several constructors to load pattern data from the resources
692     if (U_FAILURE(status)) return;
693 
694     // We will need the calendar to know what type of symbols to load.
695     initializeCalendar(NULL, locale, status);
696     if (U_FAILURE(status)) return;
697 
698     CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
699     UResourceBundle *dateTimePatterns = calData.getByKey(gDateTimePatternsTag, status);
700     UResourceBundle *currentBundle;
701 
702     if (U_FAILURE(status)) return;
703 
704     if (ures_getSize(dateTimePatterns) <= kDateTime)
705     {
706         status = U_INVALID_FORMAT_ERROR;
707         return;
708     }
709 
710     setLocaleIDs(ures_getLocaleByType(dateTimePatterns, ULOC_VALID_LOCALE, &status),
711                  ures_getLocaleByType(dateTimePatterns, ULOC_ACTUAL_LOCALE, &status));
712 
713     // create a symbols object from the locale
714     fSymbols = DateFormatSymbols::createForLocale(locale, status);
715     if (U_FAILURE(status)) return;
716     /* test for NULL */
717     if (fSymbols == 0) {
718         status = U_MEMORY_ALLOCATION_ERROR;
719         return;
720     }
721 
722     const UChar *resStr,*ovrStr;
723     int32_t resStrLen,ovrStrLen = 0;
724     fDateOverride.setToBogus();
725     fTimeOverride.setToBogus();
726 
727     // if the pattern should include both date and time information, use the date/time
728     // pattern string as a guide to tell use how to glue together the appropriate date
729     // and time pattern strings.  The actual gluing-together is handled by a convenience
730     // method on MessageFormat.
731     if ((timeStyle != kNone) && (dateStyle != kNone))
732     {
733         Formattable timeDateArray[2];
734 
735         // use Formattable::adoptString() so that we can use fastCopyFrom()
736         // instead of Formattable::setString()'s unaware, safe, deep string clone
737         // see Jitterbug 2296
738 
739         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
740         if (U_FAILURE(status)) {
741            status = U_INVALID_FORMAT_ERROR;
742            return;
743         }
744         switch (ures_getType(currentBundle)) {
745             case URES_STRING: {
746                resStr = ures_getString(currentBundle, &resStrLen, &status);
747                break;
748             }
749             case URES_ARRAY: {
750                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
751                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
752                fTimeOverride.setTo(TRUE, ovrStr, ovrStrLen);
753                break;
754             }
755             default: {
756                status = U_INVALID_FORMAT_ERROR;
757                ures_close(currentBundle);
758                return;
759             }
760         }
761         ures_close(currentBundle);
762 
763         UnicodeString *tempus1 = new UnicodeString(TRUE, resStr, resStrLen);
764         // NULL pointer check
765         if (tempus1 == NULL) {
766             status = U_MEMORY_ALLOCATION_ERROR;
767             return;
768         }
769         timeDateArray[0].adoptString(tempus1);
770 
771         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)dateStyle, NULL, &status);
772         if (U_FAILURE(status)) {
773            status = U_INVALID_FORMAT_ERROR;
774            return;
775         }
776         switch (ures_getType(currentBundle)) {
777             case URES_STRING: {
778                resStr = ures_getString(currentBundle, &resStrLen, &status);
779                break;
780             }
781             case URES_ARRAY: {
782                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
783                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
784                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
785                break;
786             }
787             default: {
788                status = U_INVALID_FORMAT_ERROR;
789                ures_close(currentBundle);
790                return;
791             }
792         }
793         ures_close(currentBundle);
794 
795         UnicodeString *tempus2 = new UnicodeString(TRUE, resStr, resStrLen);
796         // Null pointer check
797         if (tempus2 == NULL) {
798             status = U_MEMORY_ALLOCATION_ERROR;
799             return;
800         }
801         timeDateArray[1].adoptString(tempus2);
802 
803         int32_t glueIndex = kDateTime;
804         int32_t patternsSize = ures_getSize(dateTimePatterns);
805         if (patternsSize >= (kDateTimeOffset + kShort + 1)) {
806             // Get proper date time format
807             glueIndex = (int32_t)(kDateTimeOffset + (dateStyle - kDateOffset));
808         }
809 
810         resStr = ures_getStringByIndex(dateTimePatterns, glueIndex, &resStrLen, &status);
811         MessageFormat::format(UnicodeString(TRUE, resStr, resStrLen), timeDateArray, 2, fPattern, status);
812     }
813     // if the pattern includes just time data or just date date, load the appropriate
814     // pattern string from the resources
815     // setTo() - see DateFormatSymbols::assignArray comments
816     else if (timeStyle != kNone) {
817         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
818         if (U_FAILURE(status)) {
819            status = U_INVALID_FORMAT_ERROR;
820            return;
821         }
822         switch (ures_getType(currentBundle)) {
823             case URES_STRING: {
824                resStr = ures_getString(currentBundle, &resStrLen, &status);
825                break;
826             }
827             case URES_ARRAY: {
828                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
829                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
830                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
831                break;
832             }
833             default: {
834                status = U_INVALID_FORMAT_ERROR;
835                 ures_close(currentBundle);
836                return;
837             }
838         }
839         fPattern.setTo(TRUE, resStr, resStrLen);
840         ures_close(currentBundle);
841     }
842     else if (dateStyle != kNone) {
843         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)dateStyle, NULL, &status);
844         if (U_FAILURE(status)) {
845            status = U_INVALID_FORMAT_ERROR;
846            return;
847         }
848         switch (ures_getType(currentBundle)) {
849             case URES_STRING: {
850                resStr = ures_getString(currentBundle, &resStrLen, &status);
851                break;
852             }
853             case URES_ARRAY: {
854                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
855                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
856                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
857                break;
858             }
859             default: {
860                status = U_INVALID_FORMAT_ERROR;
861                ures_close(currentBundle);
862                return;
863             }
864         }
865         fPattern.setTo(TRUE, resStr, resStrLen);
866         ures_close(currentBundle);
867     }
868 
869     // and if it includes _neither_, that's an error
870     else
871         status = U_INVALID_FORMAT_ERROR;
872 
873     // finally, finish initializing by creating a Calendar and a NumberFormat
874     initialize(locale, status);
875 }
876 
877 //----------------------------------------------------------------------
878 
879 Calendar*
initializeCalendar(TimeZone * adoptZone,const Locale & locale,UErrorCode & status)880 SimpleDateFormat::initializeCalendar(TimeZone* adoptZone, const Locale& locale, UErrorCode& status)
881 {
882     if(!U_FAILURE(status)) {
883         fCalendar = Calendar::createInstance(adoptZone?adoptZone:TimeZone::createDefault(), locale, status);
884     }
885     return fCalendar;
886 }
887 
888 void
initialize(const Locale & locale,UErrorCode & status)889 SimpleDateFormat::initialize(const Locale& locale,
890                              UErrorCode& status)
891 {
892     if (U_FAILURE(status)) return;
893 
894     // We don't need to check that the row count is >= 1, since all 2d arrays have at
895     // least one row
896     fNumberFormat = NumberFormat::createInstance(locale, status);
897     if (fNumberFormat != NULL && U_SUCCESS(status))
898     {
899         fixNumberFormatForDates(*fNumberFormat);
900         //fNumberFormat->setLenient(TRUE); // Java uses a custom DateNumberFormat to format/parse
901 
902         initNumberFormatters(locale,status);
903 
904     }
905     else if (U_SUCCESS(status))
906     {
907         status = U_MISSING_RESOURCE_ERROR;
908     }
909 }
910 
911 /* Initialize the fields we use to disambiguate ambiguous years. Separate
912  * so we can call it from readObject().
913  */
initializeDefaultCentury()914 void SimpleDateFormat::initializeDefaultCentury()
915 {
916   if(fCalendar) {
917     fHaveDefaultCentury = fCalendar->haveDefaultCentury();
918     if(fHaveDefaultCentury) {
919       fDefaultCenturyStart = fCalendar->defaultCenturyStart();
920       fDefaultCenturyStartYear = fCalendar->defaultCenturyStartYear();
921     } else {
922       fDefaultCenturyStart = DBL_MIN;
923       fDefaultCenturyStartYear = -1;
924     }
925   }
926 }
927 
928 /*
929  * Initialize the boolean attributes. Separate so we can call it from all constructors.
930  */
initializeBooleanAttributes()931 void SimpleDateFormat::initializeBooleanAttributes()
932 {
933     UErrorCode status = U_ZERO_ERROR;
934 
935     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status);
936     setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
937     setBooleanAttribute(UDAT_PARSE_PARTIAL_LITERAL_MATCH, true, status);
938     setBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, true, status);
939 }
940 
941 /* Define one-century window into which to disambiguate dates using
942  * two-digit years. Make public in JDK 1.2.
943  */
parseAmbiguousDatesAsAfter(UDate startDate,UErrorCode & status)944 void SimpleDateFormat::parseAmbiguousDatesAsAfter(UDate startDate, UErrorCode& status)
945 {
946     if(U_FAILURE(status)) {
947         return;
948     }
949     if(!fCalendar) {
950       status = U_ILLEGAL_ARGUMENT_ERROR;
951       return;
952     }
953 
954     fCalendar->setTime(startDate, status);
955     if(U_SUCCESS(status)) {
956         fHaveDefaultCentury = TRUE;
957         fDefaultCenturyStart = startDate;
958         fDefaultCenturyStartYear = fCalendar->get(UCAL_YEAR, status);
959     }
960 }
961 
962 //----------------------------------------------------------------------
963 
964 UnicodeString&
format(Calendar & cal,UnicodeString & appendTo,FieldPosition & pos) const965 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo, FieldPosition& pos) const
966 {
967   UErrorCode status = U_ZERO_ERROR;
968   FieldPositionOnlyHandler handler(pos);
969   return _format(cal, appendTo, handler, status);
970 }
971 
972 //----------------------------------------------------------------------
973 
974 UnicodeString&
format(Calendar & cal,UnicodeString & appendTo,FieldPositionIterator * posIter,UErrorCode & status) const975 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo,
976                          FieldPositionIterator* posIter, UErrorCode& status) const
977 {
978   FieldPositionIteratorHandler handler(posIter, status);
979   return _format(cal, appendTo, handler, status);
980 }
981 
982 //----------------------------------------------------------------------
983 
984 UnicodeString&
_format(Calendar & cal,UnicodeString & appendTo,FieldPositionHandler & handler,UErrorCode & status) const985 SimpleDateFormat::_format(Calendar& cal, UnicodeString& appendTo,
986                             FieldPositionHandler& handler, UErrorCode& status) const
987 {
988     if ( U_FAILURE(status) ) {
989        return appendTo;
990     }
991     Calendar* workCal = &cal;
992     Calendar* calClone = NULL;
993     if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
994         // Different calendar type
995         // We use the time and time zone from the input calendar, but
996         // do not use the input calendar for field calculation.
997         calClone = fCalendar->clone();
998         if (calClone != NULL) {
999             UDate t = cal.getTime(status);
1000             calClone->setTime(t, status);
1001             calClone->setTimeZone(cal.getTimeZone());
1002             workCal = calClone;
1003         } else {
1004             status = U_MEMORY_ALLOCATION_ERROR;
1005             return appendTo;
1006         }
1007     }
1008 
1009     UBool inQuote = FALSE;
1010     UChar prevCh = 0;
1011     int32_t count = 0;
1012     int32_t fieldNum = 0;
1013     UDisplayContext capitalizationContext = getContext(UDISPCTX_TYPE_CAPITALIZATION, status);
1014 
1015     // Create temporary cache of mutable number format objects. This way
1016     // subFormat won't have to clone the const NumberFormat for each field.
1017     // if several fields share the same NumberFormat, which will almost
1018     // always be the case, this is a big save.
1019     SimpleDateFormatMutableNFs mutableNFs;
1020     // loop through the pattern string character by character
1021     for (int32_t i = 0; i < fPattern.length() && U_SUCCESS(status); ++i) {
1022         UChar ch = fPattern[i];
1023 
1024         // Use subFormat() to format a repeated pattern character
1025         // when a different pattern or non-pattern character is seen
1026         if (ch != prevCh && count > 0) {
1027             subFormat(appendTo, prevCh, count, capitalizationContext, fieldNum++, handler, *workCal, mutableNFs, status);
1028             count = 0;
1029         }
1030         if (ch == QUOTE) {
1031             // Consecutive single quotes are a single quote literal,
1032             // either outside of quotes or between quotes
1033             if ((i+1) < fPattern.length() && fPattern[i+1] == QUOTE) {
1034                 appendTo += (UChar)QUOTE;
1035                 ++i;
1036             } else {
1037                 inQuote = ! inQuote;
1038             }
1039         }
1040         else if (!inQuote && isSyntaxChar(ch)) {
1041             // ch is a date-time pattern character to be interpreted
1042             // by subFormat(); count the number of times it is repeated
1043             prevCh = ch;
1044             ++count;
1045         }
1046         else {
1047             // Append quoted characters and unquoted non-pattern characters
1048             appendTo += ch;
1049         }
1050     }
1051 
1052     // Format the last item in the pattern, if any
1053     if (count > 0) {
1054         subFormat(appendTo, prevCh, count, capitalizationContext, fieldNum++, handler, *workCal, mutableNFs, status);
1055     }
1056 
1057     if (calClone != NULL) {
1058         delete calClone;
1059     }
1060 
1061     return appendTo;
1062 }
1063 
1064 //----------------------------------------------------------------------
1065 
1066 /* Map calendar field into calendar field level.
1067  * the larger the level, the smaller the field unit.
1068  * For example, UCAL_ERA level is 0, UCAL_YEAR level is 10,
1069  * UCAL_MONTH level is 20.
1070  * NOTE: if new fields adds in, the table needs to update.
1071  */
1072 const int32_t
1073 SimpleDateFormat::fgCalendarFieldToLevel[] =
1074 {
1075     /*GyM*/ 0, 10, 20,
1076     /*wW*/ 20, 30,
1077     /*dDEF*/ 30, 20, 30, 30,
1078     /*ahHm*/ 40, 50, 50, 60,
1079     /*sS*/ 70, 80,
1080     /*z?Y*/ 0, 0, 10,
1081     /*eug*/ 30, 10, 0,
1082     /*A?.*/ 40, 0, 0
1083 };
1084 
getLevelFromChar(UChar ch)1085 int32_t SimpleDateFormat::getLevelFromChar(UChar ch) {
1086     // Map date field LETTER into calendar field level.
1087     // the larger the level, the smaller the field unit.
1088     // NOTE: if new fields adds in, the table needs to update.
1089     static const int32_t mapCharToLevel[] = {
1090             -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
1091         //
1092             -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
1093         //       !   "   #   $   %   &   '   (   )   *   +   ,   -   .   /
1094             -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
1095 #if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
1096         //   0   1   2   3   4   5   6   7   8   9   :   ;   <   =   >   ?
1097             -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,  0, -1, -1, -1, -1, -1,
1098 #else
1099         //   0   1   2   3   4   5   6   7   8   9   :   ;   <   =   >   ?
1100             -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
1101 #endif
1102         //   @   A   B   C   D   E   F   G   H   I   J   K   L   M   N   O
1103             -1, 40, -1, -1, 20, 30, 30,  0, 50, -1, -1, 50, 20, 20, -1,  0,
1104         //   P   Q   R   S   T   U   V   W   X   Y   Z   [   \   ]   ^   _
1105             -1, 20, -1, 80, -1, 10,  0, 30,  0, 10,  0, -1, -1, -1, -1, -1,
1106         //   `   a   b   c   d   e   f   g   h   i   j   k   l   m   n   o
1107             -1, 40, -1, 30, 30, 30, -1,  0, 50, -1, -1, 50,  0, 60, -1, -1,
1108         //   p   q   r   s   t   u   v   w   x   y   z   {   |   }   ~
1109             -1, 20, 10, 70, -1, 10,  0, 20,  0, 10,  0, -1, -1, -1, -1, -1
1110     };
1111 
1112     return ch < UPRV_LENGTHOF(mapCharToLevel) ? mapCharToLevel[ch] : -1;
1113 }
1114 
isSyntaxChar(UChar ch)1115 UBool SimpleDateFormat::isSyntaxChar(UChar ch) {
1116     static const UBool mapCharToIsSyntax[] = {
1117         //
1118         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1119         //
1120         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1121         //
1122         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1123         //
1124         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1125         //         !      "      #      $      %      &      '
1126         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1127         //  (      )      *      +      ,      -      .      /
1128         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1129         //  0      1      2      3      4      5      6      7
1130         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1131 #if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
1132         //  8      9      :      ;      <      =      >      ?
1133         FALSE, FALSE,  TRUE, FALSE, FALSE, FALSE, FALSE, FALSE,
1134 #else
1135         //  8      9      :      ;      <      =      >      ?
1136         FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1137 #endif
1138         //  @      A      B      C      D      E      F      G
1139         FALSE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,
1140         //  H      I      J      K      L      M      N      O
1141          TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,
1142         //  P      Q      R      S      T      U      V      W
1143          TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,
1144         //  X      Y      Z      [      \      ]      ^      _
1145          TRUE,  TRUE,  TRUE, FALSE, FALSE, FALSE, FALSE, FALSE,
1146         //  `      a      b      c      d      e      f      g
1147         FALSE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,
1148         //  h      i      j      k      l      m      n      o
1149          TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,
1150         //  p      q      r      s      t      u      v      w
1151          TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,  TRUE,
1152         //  x      y      z      {      |      }      ~
1153          TRUE,  TRUE,  TRUE, FALSE, FALSE, FALSE, FALSE, FALSE
1154     };
1155 
1156     return ch < UPRV_LENGTHOF(mapCharToIsSyntax) ? mapCharToIsSyntax[ch] : FALSE;
1157 }
1158 
1159 // Map index into pattern character string to Calendar field number.
1160 const UCalendarDateFields
1161 SimpleDateFormat::fgPatternIndexToCalendarField[] =
1162 {
1163     /*GyM*/ UCAL_ERA, UCAL_YEAR, UCAL_MONTH,
1164     /*dkH*/ UCAL_DATE, UCAL_HOUR_OF_DAY, UCAL_HOUR_OF_DAY,
1165     /*msS*/ UCAL_MINUTE, UCAL_SECOND, UCAL_MILLISECOND,
1166     /*EDF*/ UCAL_DAY_OF_WEEK, UCAL_DAY_OF_YEAR, UCAL_DAY_OF_WEEK_IN_MONTH,
1167     /*wWa*/ UCAL_WEEK_OF_YEAR, UCAL_WEEK_OF_MONTH, UCAL_AM_PM,
1168     /*hKz*/ UCAL_HOUR, UCAL_HOUR, UCAL_ZONE_OFFSET,
1169     /*Yeu*/ UCAL_YEAR_WOY, UCAL_DOW_LOCAL, UCAL_EXTENDED_YEAR,
1170     /*gAZ*/ UCAL_JULIAN_DAY, UCAL_MILLISECONDS_IN_DAY, UCAL_ZONE_OFFSET,
1171     /*v*/   UCAL_ZONE_OFFSET,
1172     /*c*/   UCAL_DOW_LOCAL,
1173     /*L*/   UCAL_MONTH,
1174     /*Q*/   UCAL_MONTH,
1175     /*q*/   UCAL_MONTH,
1176     /*V*/   UCAL_ZONE_OFFSET,
1177     /*U*/   UCAL_YEAR,
1178     /*O*/   UCAL_ZONE_OFFSET,
1179     /*Xx*/  UCAL_ZONE_OFFSET, UCAL_ZONE_OFFSET,
1180     /*r*/   UCAL_EXTENDED_YEAR,
1181 #if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
1182     /*:*/   UCAL_FIELD_COUNT, /* => no useful mapping to any calendar field */
1183 #else
1184     /*no pattern char for UDAT_TIME_SEPARATOR_FIELD*/   UCAL_FIELD_COUNT, /* => no useful mapping to any calendar field */
1185 #endif
1186 };
1187 
1188 // Map index into pattern character string to DateFormat field number
1189 const UDateFormatField
1190 SimpleDateFormat::fgPatternIndexToDateFormatField[] = {
1191     /*GyM*/ UDAT_ERA_FIELD, UDAT_YEAR_FIELD, UDAT_MONTH_FIELD,
1192     /*dkH*/ UDAT_DATE_FIELD, UDAT_HOUR_OF_DAY1_FIELD, UDAT_HOUR_OF_DAY0_FIELD,
1193     /*msS*/ UDAT_MINUTE_FIELD, UDAT_SECOND_FIELD, UDAT_FRACTIONAL_SECOND_FIELD,
1194     /*EDF*/ UDAT_DAY_OF_WEEK_FIELD, UDAT_DAY_OF_YEAR_FIELD, UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
1195     /*wWa*/ UDAT_WEEK_OF_YEAR_FIELD, UDAT_WEEK_OF_MONTH_FIELD, UDAT_AM_PM_FIELD,
1196     /*hKz*/ UDAT_HOUR1_FIELD, UDAT_HOUR0_FIELD, UDAT_TIMEZONE_FIELD,
1197     /*Yeu*/ UDAT_YEAR_WOY_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_EXTENDED_YEAR_FIELD,
1198     /*gAZ*/ UDAT_JULIAN_DAY_FIELD, UDAT_MILLISECONDS_IN_DAY_FIELD, UDAT_TIMEZONE_RFC_FIELD,
1199     /*v*/   UDAT_TIMEZONE_GENERIC_FIELD,
1200     /*c*/   UDAT_STANDALONE_DAY_FIELD,
1201     /*L*/   UDAT_STANDALONE_MONTH_FIELD,
1202     /*Q*/   UDAT_QUARTER_FIELD,
1203     /*q*/   UDAT_STANDALONE_QUARTER_FIELD,
1204     /*V*/   UDAT_TIMEZONE_SPECIAL_FIELD,
1205     /*U*/   UDAT_YEAR_NAME_FIELD,
1206     /*O*/   UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD,
1207     /*Xx*/  UDAT_TIMEZONE_ISO_FIELD, UDAT_TIMEZONE_ISO_LOCAL_FIELD,
1208     /*r*/   UDAT_RELATED_YEAR_FIELD,
1209 #if UDAT_HAS_PATTERN_CHAR_FOR_TIME_SEPARATOR
1210     /*:*/   UDAT_TIME_SEPARATOR_FIELD,
1211 #else
1212     /*no pattern char for UDAT_TIME_SEPARATOR_FIELD*/   UDAT_TIME_SEPARATOR_FIELD,
1213 #endif
1214 };
1215 
1216 //----------------------------------------------------------------------
1217 
1218 /**
1219  * Append symbols[value] to dst.  Make sure the array index is not out
1220  * of bounds.
1221  */
1222 static inline void
_appendSymbol(UnicodeString & dst,int32_t value,const UnicodeString * symbols,int32_t symbolsCount)1223 _appendSymbol(UnicodeString& dst,
1224               int32_t value,
1225               const UnicodeString* symbols,
1226               int32_t symbolsCount) {
1227     U_ASSERT(0 <= value && value < symbolsCount);
1228     if (0 <= value && value < symbolsCount) {
1229         dst += symbols[value];
1230     }
1231 }
1232 
1233 static inline void
_appendSymbolWithMonthPattern(UnicodeString & dst,int32_t value,const UnicodeString * symbols,int32_t symbolsCount,const UnicodeString * monthPattern,UErrorCode & status)1234 _appendSymbolWithMonthPattern(UnicodeString& dst, int32_t value, const UnicodeString* symbols, int32_t symbolsCount,
1235               const UnicodeString* monthPattern, UErrorCode& status) {
1236     U_ASSERT(0 <= value && value < symbolsCount);
1237     if (0 <= value && value < symbolsCount) {
1238         if (monthPattern == NULL) {
1239             dst += symbols[value];
1240         } else {
1241             Formattable monthName((const UnicodeString&)(symbols[value]));
1242             MessageFormat::format(*monthPattern, &monthName, 1, dst, status);
1243         }
1244     }
1245 }
1246 
1247 //----------------------------------------------------------------------
1248 void
initNumberFormatters(const Locale & locale,UErrorCode & status)1249 SimpleDateFormat::initNumberFormatters(const Locale &locale,UErrorCode &status) {
1250     if (U_FAILURE(status)) {
1251         return;
1252     }
1253     if ( fDateOverride.isBogus() && fTimeOverride.isBogus() ) {
1254         return;
1255     }
1256     umtx_lock(&LOCK);
1257     if (fSharedNumberFormatters == NULL) {
1258         fSharedNumberFormatters = allocSharedNumberFormatters();
1259         if (fSharedNumberFormatters == NULL) {
1260             status = U_MEMORY_ALLOCATION_ERROR;
1261         }
1262     }
1263     umtx_unlock(&LOCK);
1264 
1265     if (U_FAILURE(status)) {
1266         return;
1267     }
1268 
1269     processOverrideString(locale,fDateOverride,kOvrStrDate,status);
1270     processOverrideString(locale,fTimeOverride,kOvrStrTime,status);
1271 }
1272 
1273 void
processOverrideString(const Locale & locale,const UnicodeString & str,int8_t type,UErrorCode & status)1274 SimpleDateFormat::processOverrideString(const Locale &locale, const UnicodeString &str, int8_t type, UErrorCode &status) {
1275     if (str.isBogus() || U_FAILURE(status)) {
1276         return;
1277     }
1278 
1279     int32_t start = 0;
1280     int32_t len;
1281     UnicodeString nsName;
1282     UnicodeString ovrField;
1283     UBool moreToProcess = TRUE;
1284     NSOverride *overrideList = NULL;
1285 
1286     while (moreToProcess) {
1287         int32_t delimiterPosition = str.indexOf((UChar)ULOC_KEYWORD_ITEM_SEPARATOR_UNICODE,start);
1288         if (delimiterPosition == -1) {
1289             moreToProcess = FALSE;
1290             len = str.length() - start;
1291         } else {
1292             len = delimiterPosition - start;
1293         }
1294         UnicodeString currentString(str,start,len);
1295         int32_t equalSignPosition = currentString.indexOf((UChar)ULOC_KEYWORD_ASSIGN_UNICODE,0);
1296         if (equalSignPosition == -1) { // Simple override string such as "hebrew"
1297             nsName.setTo(currentString);
1298             ovrField.setToBogus();
1299         } else { // Field specific override string such as "y=hebrew"
1300             nsName.setTo(currentString,equalSignPosition+1);
1301             ovrField.setTo(currentString,0,1); // We just need the first character.
1302         }
1303 
1304         int32_t nsNameHash = nsName.hashCode();
1305         // See if the numbering system is in the override list, if not, then add it.
1306         NSOverride *cur = overrideList;
1307         const SharedNumberFormat *snf = NULL;
1308         UBool found = FALSE;
1309         while ( cur && !found ) {
1310             if ( cur->hash == nsNameHash ) {
1311                 snf = cur->snf;
1312                 found = TRUE;
1313             }
1314             cur = cur->next;
1315         }
1316 
1317         if (!found) {
1318            LocalPointer<NSOverride> cur(new NSOverride);
1319            if (!cur.isNull()) {
1320                char kw[ULOC_KEYWORD_AND_VALUES_CAPACITY];
1321                uprv_strcpy(kw,"numbers=");
1322                nsName.extract(0,len,kw+8,ULOC_KEYWORD_AND_VALUES_CAPACITY-8,US_INV);
1323 
1324                Locale ovrLoc(locale.getLanguage(),locale.getCountry(),locale.getVariant(),kw);
1325                cur->hash = nsNameHash;
1326                cur->next = overrideList;
1327                SharedObject::copyPtr(
1328                        createSharedNumberFormat(ovrLoc, status), cur->snf);
1329                if (U_FAILURE(status)) {
1330                    if (overrideList) {
1331                        overrideList->free();
1332                    }
1333                    return;
1334                }
1335                snf = cur->snf;
1336                overrideList = cur.orphan();
1337            } else {
1338                status = U_MEMORY_ALLOCATION_ERROR;
1339                if (overrideList) {
1340                    overrideList->free();
1341                }
1342                return;
1343            }
1344         }
1345 
1346         // Now that we have an appropriate number formatter, fill in the appropriate spaces in the
1347         // number formatters table.
1348         if (ovrField.isBogus()) {
1349             switch (type) {
1350                 case kOvrStrDate:
1351                 case kOvrStrBoth: {
1352                     for ( int8_t i=0 ; i<kDateFieldsCount; i++ ) {
1353                         SharedObject::copyPtr(snf, fSharedNumberFormatters[kDateFields[i]]);
1354                     }
1355                     if (type==kOvrStrDate) {
1356                         break;
1357                     }
1358                 }
1359                 case kOvrStrTime : {
1360                     for ( int8_t i=0 ; i<kTimeFieldsCount; i++ ) {
1361                         SharedObject::copyPtr(snf, fSharedNumberFormatters[kTimeFields[i]]);
1362                     }
1363                     break;
1364                 }
1365             }
1366         } else {
1367            // if the pattern character is unrecognized, signal an error and bail out
1368            UDateFormatField patternCharIndex =
1369               DateFormatSymbols::getPatternCharIndex(ovrField.charAt(0));
1370            if (patternCharIndex == UDAT_FIELD_COUNT) {
1371                status = U_INVALID_FORMAT_ERROR;
1372                if (overrideList) {
1373                    overrideList->free();
1374                }
1375                return;
1376            }
1377            SharedObject::copyPtr(snf, fSharedNumberFormatters[patternCharIndex]);
1378         }
1379 
1380         start = delimiterPosition + 1;
1381     }
1382     if (overrideList) {
1383         overrideList->free();
1384     }
1385 }
1386 
1387 //---------------------------------------------------------------------
1388 void
subFormat(UnicodeString & appendTo,UChar ch,int32_t count,UDisplayContext capitalizationContext,int32_t fieldNum,FieldPositionHandler & handler,Calendar & cal,SimpleDateFormatMutableNFs & mutableNFs,UErrorCode & status) const1389 SimpleDateFormat::subFormat(UnicodeString &appendTo,
1390                             UChar ch,
1391                             int32_t count,
1392                             UDisplayContext capitalizationContext,
1393                             int32_t fieldNum,
1394                             FieldPositionHandler& handler,
1395                             Calendar& cal,
1396                             SimpleDateFormatMutableNFs &mutableNFs,
1397                             UErrorCode& status) const
1398 {
1399     if (U_FAILURE(status)) {
1400         return;
1401     }
1402 
1403     // this function gets called by format() to produce the appropriate substitution
1404     // text for an individual pattern symbol (e.g., "HH" or "yyyy")
1405 
1406     UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
1407     const int32_t maxIntCount = 10;
1408     int32_t beginOffset = appendTo.length();
1409     NumberFormat *currentNumberFormat;
1410     DateFormatSymbols::ECapitalizationContextUsageType capContextUsageType = DateFormatSymbols::kCapContextUsageOther;
1411 
1412     UBool isHebrewCalendar = (uprv_strcmp(cal.getType(),"hebrew") == 0);
1413     UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0 || uprv_strcmp(cal.getType(),"dangi") == 0);
1414 
1415     // if the pattern character is unrecognized, signal an error and dump out
1416     if (patternCharIndex == UDAT_FIELD_COUNT)
1417     {
1418         if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
1419             status = U_INVALID_FORMAT_ERROR;
1420         }
1421         return;
1422     }
1423 
1424     UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex];
1425     int32_t value = 0;
1426     // Don't get value unless it is useful
1427     if (field < UCAL_FIELD_COUNT) {
1428         value = (patternCharIndex != UDAT_RELATED_YEAR_FIELD)? cal.get(field, status): cal.getRelatedYear(status);
1429     }
1430     if (U_FAILURE(status)) {
1431         return;
1432     }
1433 
1434     currentNumberFormat = mutableNFs.get(getNumberFormatByIndex(patternCharIndex));
1435     if (currentNumberFormat == NULL) {
1436         status = U_MEMORY_ALLOCATION_ERROR;
1437         return;
1438     }
1439     UnicodeString hebr("hebr", 4, US_INV);
1440 
1441     switch (patternCharIndex) {
1442 
1443     // for any "G" symbol, write out the appropriate era string
1444     // "GGGG" is wide era name, "GGGGG" is narrow era name, anything else is abbreviated name
1445     case UDAT_ERA_FIELD:
1446         if (isChineseCalendar) {
1447             zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, 9); // as in ICU4J
1448         } else {
1449             if (count == 5) {
1450                 _appendSymbol(appendTo, value, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount);
1451                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraNarrow;
1452             } else if (count == 4) {
1453                 _appendSymbol(appendTo, value, fSymbols->fEraNames, fSymbols->fEraNamesCount);
1454                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraWide;
1455             } else {
1456                 _appendSymbol(appendTo, value, fSymbols->fEras, fSymbols->fErasCount);
1457                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraAbbrev;
1458             }
1459         }
1460         break;
1461 
1462      case UDAT_YEAR_NAME_FIELD:
1463         if (fSymbols->fShortYearNames != NULL && value <= fSymbols->fShortYearNamesCount) {
1464             // the Calendar YEAR field runs 1 through 60 for cyclic years
1465             _appendSymbol(appendTo, value - 1, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount);
1466             break;
1467         }
1468         // else fall through to numeric year handling, do not break here
1469 
1470    // OLD: for "yyyy", write out the whole year; for "yy", write out the last 2 digits
1471     // NEW: UTS#35:
1472 //Year         y     yy     yyy     yyyy     yyyyy
1473 //AD 1         1     01     001     0001     00001
1474 //AD 12       12     12     012     0012     00012
1475 //AD 123     123     23     123     0123     00123
1476 //AD 1234   1234     34    1234     1234     01234
1477 //AD 12345 12345     45   12345    12345     12345
1478     case UDAT_YEAR_FIELD:
1479     case UDAT_YEAR_WOY_FIELD:
1480         if (fDateOverride.compare(hebr)==0 && value>HEBREW_CAL_CUR_MILLENIUM_START_YEAR && value<HEBREW_CAL_CUR_MILLENIUM_END_YEAR) {
1481             value-=HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
1482         }
1483         if(count == 2)
1484             zeroPaddingNumber(currentNumberFormat, appendTo, value, 2, 2);
1485         else
1486             zeroPaddingNumber(currentNumberFormat, appendTo, value, count, maxIntCount);
1487         break;
1488 
1489     // for "MMMM"/"LLLL", write out the whole month name, for "MMM"/"LLL", write out the month
1490     // abbreviation, for "M"/"L" or "MM"/"LL", write out the month as a number with the
1491     // appropriate number of digits
1492     // for "MMMMM"/"LLLLL", use the narrow form
1493     case UDAT_MONTH_FIELD:
1494     case UDAT_STANDALONE_MONTH_FIELD:
1495         if ( isHebrewCalendar ) {
1496            HebrewCalendar *hc = (HebrewCalendar*)&cal;
1497            if (hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value == 6 && count >= 3 )
1498                value = 13; // Show alternate form for Adar II in leap years in Hebrew calendar.
1499            if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6 && count < 3 )
1500                value--; // Adjust the month number down 1 in Hebrew non-leap years, i.e. Adar is 6, not 7.
1501         }
1502         {
1503             int32_t isLeapMonth = (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount)?
1504                         cal.get(UCAL_IS_LEAP_MONTH, status): 0;
1505             // should consolidate the next section by using arrays of pointers & counts for the right symbols...
1506             if (count == 5) {
1507                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1508                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fNarrowMonths, fSymbols->fNarrowMonthsCount,
1509                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatNarrow]): NULL, status);
1510                 } else {
1511                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneNarrowMonths, fSymbols->fStandaloneNarrowMonthsCount,
1512                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneNarrow]): NULL, status);
1513                 }
1514                 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthNarrow;
1515             } else if (count == 4) {
1516                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1517                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fMonths, fSymbols->fMonthsCount,
1518                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide]): NULL, status);
1519                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1520                 } else {
1521                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount,
1522                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide]): NULL, status);
1523                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1524                 }
1525             } else if (count == 3) {
1526                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1527                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fShortMonths, fSymbols->fShortMonthsCount,
1528                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev]): NULL, status);
1529                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1530                 } else {
1531                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount,
1532                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev]): NULL, status);
1533                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1534                 }
1535             } else {
1536                 UnicodeString monthNumber;
1537                 zeroPaddingNumber(currentNumberFormat,monthNumber, value + 1, count, maxIntCount);
1538                 _appendSymbolWithMonthPattern(appendTo, 0, &monthNumber, 1,
1539                         (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric]): NULL, status);
1540             }
1541         }
1542         break;
1543 
1544     // for "k" and "kk", write out the hour, adjusting midnight to appear as "24"
1545     case UDAT_HOUR_OF_DAY1_FIELD:
1546         if (value == 0)
1547             zeroPaddingNumber(currentNumberFormat,appendTo, cal.getMaximum(UCAL_HOUR_OF_DAY) + 1, count, maxIntCount);
1548         else
1549             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1550         break;
1551 
1552     case UDAT_FRACTIONAL_SECOND_FIELD:
1553         // Fractional seconds left-justify
1554         {
1555             currentNumberFormat->setMinimumIntegerDigits((count > 3) ? 3 : count);
1556             currentNumberFormat->setMaximumIntegerDigits(maxIntCount);
1557             if (count == 1) {
1558                 value /= 100;
1559             } else if (count == 2) {
1560                 value /= 10;
1561             }
1562             FieldPosition p(0);
1563             currentNumberFormat->format(value, appendTo, p);
1564             if (count > 3) {
1565                 currentNumberFormat->setMinimumIntegerDigits(count - 3);
1566                 currentNumberFormat->format((int32_t)0, appendTo, p);
1567             }
1568         }
1569         break;
1570 
1571     // for "ee" or "e", use local numeric day-of-the-week
1572     // for "EEEEEE" or "eeeeee", write out the short day-of-the-week name
1573     // for "EEEEE" or "eeeee", write out the narrow day-of-the-week name
1574     // for "EEEE" or "eeee", write out the wide day-of-the-week name
1575     // for "EEE" or "EE" or "E" or "eee", write out the abbreviated day-of-the-week name
1576     case UDAT_DOW_LOCAL_FIELD:
1577         if ( count < 3 ) {
1578             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1579             break;
1580         }
1581         // fall through to EEEEE-EEE handling, but for that we don't want local day-of-week,
1582         // we want standard day-of-week, so first fix value to work for EEEEE-EEE.
1583         value = cal.get(UCAL_DAY_OF_WEEK, status);
1584         if (U_FAILURE(status)) {
1585             return;
1586         }
1587         // fall through, do not break here
1588     case UDAT_DAY_OF_WEEK_FIELD:
1589         if (count == 5) {
1590             _appendSymbol(appendTo, value, fSymbols->fNarrowWeekdays,
1591                           fSymbols->fNarrowWeekdaysCount);
1592             capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1593         } else if (count == 4) {
1594             _appendSymbol(appendTo, value, fSymbols->fWeekdays,
1595                           fSymbols->fWeekdaysCount);
1596             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1597         } else if (count == 6) {
1598             _appendSymbol(appendTo, value, fSymbols->fShorterWeekdays,
1599                           fSymbols->fShorterWeekdaysCount);
1600             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1601         } else {
1602             _appendSymbol(appendTo, value, fSymbols->fShortWeekdays,
1603                           fSymbols->fShortWeekdaysCount);
1604             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1605         }
1606         break;
1607 
1608     // for "ccc", write out the abbreviated day-of-the-week name
1609     // for "cccc", write out the wide day-of-the-week name
1610     // for "ccccc", use the narrow day-of-the-week name
1611     // for "ccccc", use the short day-of-the-week name
1612     case UDAT_STANDALONE_DAY_FIELD:
1613         if ( count < 3 ) {
1614             zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, maxIntCount);
1615             break;
1616         }
1617         // fall through to alpha DOW handling, but for that we don't want local day-of-week,
1618         // we want standard day-of-week, so first fix value.
1619         value = cal.get(UCAL_DAY_OF_WEEK, status);
1620         if (U_FAILURE(status)) {
1621             return;
1622         }
1623         if (count == 5) {
1624             _appendSymbol(appendTo, value, fSymbols->fStandaloneNarrowWeekdays,
1625                           fSymbols->fStandaloneNarrowWeekdaysCount);
1626             capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1627         } else if (count == 4) {
1628             _appendSymbol(appendTo, value, fSymbols->fStandaloneWeekdays,
1629                           fSymbols->fStandaloneWeekdaysCount);
1630             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1631         } else if (count == 6) {
1632             _appendSymbol(appendTo, value, fSymbols->fStandaloneShorterWeekdays,
1633                           fSymbols->fStandaloneShorterWeekdaysCount);
1634             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1635         } else { // count == 3
1636             _appendSymbol(appendTo, value, fSymbols->fStandaloneShortWeekdays,
1637                           fSymbols->fStandaloneShortWeekdaysCount);
1638             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1639         }
1640         break;
1641 
1642     // for "a" symbol, write out the whole AM/PM string
1643     case UDAT_AM_PM_FIELD:
1644         if (count < 5) {
1645             _appendSymbol(appendTo, value, fSymbols->fAmPms,
1646                           fSymbols->fAmPmsCount);
1647         } else {
1648             _appendSymbol(appendTo, value, fSymbols->fNarrowAmPms,
1649                           fSymbols->fNarrowAmPmsCount);
1650         }
1651         break;
1652 
1653     // if we see pattern character for UDAT_TIME_SEPARATOR_FIELD (none currently defined),
1654     // write out the time separator string. Leave support in for future definition.
1655     case UDAT_TIME_SEPARATOR_FIELD:
1656         {
1657             UnicodeString separator;
1658             appendTo += fSymbols->getTimeSeparatorString(separator);
1659         }
1660         break;
1661 
1662     // for "h" and "hh", write out the hour, adjusting noon and midnight to show up
1663     // as "12"
1664     case UDAT_HOUR1_FIELD:
1665         if (value == 0)
1666             zeroPaddingNumber(currentNumberFormat,appendTo, cal.getLeastMaximum(UCAL_HOUR) + 1, count, maxIntCount);
1667         else
1668             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1669         break;
1670 
1671     case UDAT_TIMEZONE_FIELD: // 'z'
1672     case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
1673     case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
1674     case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
1675     case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
1676     case UDAT_TIMEZONE_ISO_FIELD: // 'X'
1677     case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
1678         {
1679             UChar zsbuf[64];
1680             UnicodeString zoneString(zsbuf, 0, UPRV_LENGTHOF(zsbuf));
1681             const TimeZone& tz = cal.getTimeZone();
1682             UDate date = cal.getTime(status);
1683             if (U_SUCCESS(status)) {
1684                 if (patternCharIndex == UDAT_TIMEZONE_FIELD) {
1685                     if (count < 4) {
1686                         // "z", "zz", "zzz"
1687                         tzFormat()->format(UTZFMT_STYLE_SPECIFIC_SHORT, tz, date, zoneString);
1688                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1689                     } else {
1690                         // "zzzz" or longer
1691                         tzFormat()->format(UTZFMT_STYLE_SPECIFIC_LONG, tz, date, zoneString);
1692                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1693                     }
1694                 }
1695                 else if (patternCharIndex == UDAT_TIMEZONE_RFC_FIELD) {
1696                     if (count < 4) {
1697                         // "Z"
1698                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
1699                     } else if (count == 5) {
1700                         // "ZZZZZ"
1701                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
1702                     } else {
1703                         // "ZZ", "ZZZ", "ZZZZ"
1704                         tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
1705                     }
1706                 }
1707                 else if (patternCharIndex == UDAT_TIMEZONE_GENERIC_FIELD) {
1708                     if (count == 1) {
1709                         // "v"
1710                         tzFormat()->format(UTZFMT_STYLE_GENERIC_SHORT, tz, date, zoneString);
1711                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1712                     } else if (count == 4) {
1713                         // "vvvv"
1714                         tzFormat()->format(UTZFMT_STYLE_GENERIC_LONG, tz, date, zoneString);
1715                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1716                     }
1717                 }
1718                 else if (patternCharIndex == UDAT_TIMEZONE_SPECIAL_FIELD) {
1719                     if (count == 1) {
1720                         // "V"
1721                         tzFormat()->format(UTZFMT_STYLE_ZONE_ID_SHORT, tz, date, zoneString);
1722                     } else if (count == 2) {
1723                         // "VV"
1724                         tzFormat()->format(UTZFMT_STYLE_ZONE_ID, tz, date, zoneString);
1725                     } else if (count == 3) {
1726                         // "VVV"
1727                         tzFormat()->format(UTZFMT_STYLE_EXEMPLAR_LOCATION, tz, date, zoneString);
1728                     } else if (count == 4) {
1729                         // "VVVV"
1730                         tzFormat()->format(UTZFMT_STYLE_GENERIC_LOCATION, tz, date, zoneString);
1731                         capContextUsageType = DateFormatSymbols::kCapContextUsageZoneLong;
1732                     }
1733                 }
1734                 else if (patternCharIndex == UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD) {
1735                     if (count == 1) {
1736                         // "O"
1737                         tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT_SHORT, tz, date, zoneString);
1738                     } else if (count == 4) {
1739                         // "OOOO"
1740                         tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
1741                     }
1742                 }
1743                 else if (patternCharIndex == UDAT_TIMEZONE_ISO_FIELD) {
1744                     if (count == 1) {
1745                         // "X"
1746                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_SHORT, tz, date, zoneString);
1747                     } else if (count == 2) {
1748                         // "XX"
1749                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_FIXED, tz, date, zoneString);
1750                     } else if (count == 3) {
1751                         // "XXX"
1752                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FIXED, tz, date, zoneString);
1753                     } else if (count == 4) {
1754                         // "XXXX"
1755                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_FULL, tz, date, zoneString);
1756                     } else if (count == 5) {
1757                         // "XXXXX"
1758                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
1759                     }
1760                 }
1761                 else if (patternCharIndex == UDAT_TIMEZONE_ISO_LOCAL_FIELD) {
1762                     if (count == 1) {
1763                         // "x"
1764                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT, tz, date, zoneString);
1765                     } else if (count == 2) {
1766                         // "xx"
1767                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED, tz, date, zoneString);
1768                     } else if (count == 3) {
1769                         // "xxx"
1770                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED, tz, date, zoneString);
1771                     } else if (count == 4) {
1772                         // "xxxx"
1773                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
1774                     } else if (count == 5) {
1775                         // "xxxxx"
1776                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL, tz, date, zoneString);
1777                     }
1778                 }
1779                 else {
1780                     U_ASSERT(FALSE);
1781                 }
1782             }
1783             appendTo += zoneString;
1784         }
1785         break;
1786 
1787     case UDAT_QUARTER_FIELD:
1788         if (count >= 4)
1789             _appendSymbol(appendTo, value/3, fSymbols->fQuarters,
1790                           fSymbols->fQuartersCount);
1791         else if (count == 3)
1792             _appendSymbol(appendTo, value/3, fSymbols->fShortQuarters,
1793                           fSymbols->fShortQuartersCount);
1794         else
1795             zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1796         break;
1797 
1798     case UDAT_STANDALONE_QUARTER_FIELD:
1799         if (count >= 4)
1800             _appendSymbol(appendTo, value/3, fSymbols->fStandaloneQuarters,
1801                           fSymbols->fStandaloneQuartersCount);
1802         else if (count == 3)
1803             _appendSymbol(appendTo, value/3, fSymbols->fStandaloneShortQuarters,
1804                           fSymbols->fStandaloneShortQuartersCount);
1805         else
1806             zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1807         break;
1808 
1809 
1810     // all of the other pattern symbols can be formatted as simple numbers with
1811     // appropriate zero padding
1812     default:
1813         zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1814         break;
1815     }
1816 #if !UCONFIG_NO_BREAK_ITERATION
1817     // if first field, check to see whether we need to and are able to titlecase it
1818     if (fieldNum == 0 && u_islower(appendTo.char32At(beginOffset)) && fCapitalizationBrkIter != NULL) {
1819         UBool titlecase = FALSE;
1820         switch (capitalizationContext) {
1821             case UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE:
1822                 titlecase = TRUE;
1823                 break;
1824             case UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU:
1825                 titlecase = fSymbols->fCapitalization[capContextUsageType][0];
1826                 break;
1827             case UDISPCTX_CAPITALIZATION_FOR_STANDALONE:
1828                 titlecase = fSymbols->fCapitalization[capContextUsageType][1];
1829                 break;
1830             default:
1831                 // titlecase = FALSE;
1832                 break;
1833         }
1834         if (titlecase) {
1835             UnicodeString firstField(appendTo, beginOffset);
1836             firstField.toTitle(fCapitalizationBrkIter, fLocale, U_TITLECASE_NO_LOWERCASE | U_TITLECASE_NO_BREAK_ADJUSTMENT);
1837             appendTo.replaceBetween(beginOffset, appendTo.length(), firstField);
1838         }
1839     }
1840 #endif
1841 
1842     handler.addAttribute(fgPatternIndexToDateFormatField[patternCharIndex], beginOffset, appendTo.length());
1843 }
1844 
1845 //----------------------------------------------------------------------
1846 
adoptNumberFormat(NumberFormat * formatToAdopt)1847 void SimpleDateFormat::adoptNumberFormat(NumberFormat *formatToAdopt) {
1848     fixNumberFormatForDates(*formatToAdopt);
1849     delete fNumberFormat;
1850     fNumberFormat = formatToAdopt;
1851 
1852     // We successfully set the default number format. Now delete the overrides
1853     // (can't fail).
1854     if (fSharedNumberFormatters) {
1855         freeSharedNumberFormatters(fSharedNumberFormatters);
1856         fSharedNumberFormatters = NULL;
1857     }
1858 }
1859 
adoptNumberFormat(const UnicodeString & fields,NumberFormat * formatToAdopt,UErrorCode & status)1860 void SimpleDateFormat::adoptNumberFormat(const UnicodeString& fields, NumberFormat *formatToAdopt, UErrorCode &status){
1861     fixNumberFormatForDates(*formatToAdopt);
1862     LocalPointer<NumberFormat> fmt(formatToAdopt);
1863     if (U_FAILURE(status)) {
1864         return;
1865     }
1866 
1867     // We must ensure fSharedNumberFormatters is allocated.
1868     if (fSharedNumberFormatters == NULL) {
1869         fSharedNumberFormatters = allocSharedNumberFormatters();
1870         if (fSharedNumberFormatters == NULL) {
1871             status = U_MEMORY_ALLOCATION_ERROR;
1872             return;
1873         }
1874     }
1875     const SharedNumberFormat *newFormat = createSharedNumberFormat(fmt.orphan());
1876     if (newFormat == NULL) {
1877         status = U_MEMORY_ALLOCATION_ERROR;
1878         return;
1879     }
1880     for (int i=0; i<fields.length(); i++) {
1881         UChar field = fields.charAt(i);
1882         // if the pattern character is unrecognized, signal an error and bail out
1883         UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(field);
1884         if (patternCharIndex == UDAT_FIELD_COUNT) {
1885             status = U_INVALID_FORMAT_ERROR;
1886             newFormat->deleteIfZeroRefCount();
1887             return;
1888         }
1889 
1890         // Set the number formatter in the table
1891         SharedObject::copyPtr(
1892                 newFormat, fSharedNumberFormatters[patternCharIndex]);
1893     }
1894     newFormat->deleteIfZeroRefCount();
1895 }
1896 
1897 const NumberFormat *
getNumberFormatForField(UChar field) const1898 SimpleDateFormat::getNumberFormatForField(UChar field) const {
1899     UDateFormatField index = DateFormatSymbols::getPatternCharIndex(field);
1900     if (index == UDAT_FIELD_COUNT) {
1901         return NULL;
1902     }
1903     return getNumberFormatByIndex(index);
1904 }
1905 
1906 //----------------------------------------------------------------------
1907 void
zeroPaddingNumber(NumberFormat * currentNumberFormat,UnicodeString & appendTo,int32_t value,int32_t minDigits,int32_t maxDigits) const1908 SimpleDateFormat::zeroPaddingNumber(
1909         NumberFormat *currentNumberFormat,
1910         UnicodeString &appendTo,
1911         int32_t value, int32_t minDigits, int32_t maxDigits) const
1912 {
1913     if (currentNumberFormat!=NULL) {
1914         FieldPosition pos(0);
1915 
1916         currentNumberFormat->setMinimumIntegerDigits(minDigits);
1917         currentNumberFormat->setMaximumIntegerDigits(maxDigits);
1918         currentNumberFormat->format(value, appendTo, pos);  // 3rd arg is there to speed up processing
1919     }
1920 }
1921 
1922 //----------------------------------------------------------------------
1923 
1924 /**
1925  * Return true if the given format character, occuring count
1926  * times, represents a numeric field.
1927  */
isNumeric(UChar formatChar,int32_t count)1928 UBool SimpleDateFormat::isNumeric(UChar formatChar, int32_t count) {
1929     return DateFormatSymbols::isNumericPatternChar(formatChar, count);
1930 }
1931 
1932 UBool
isAtNumericField(const UnicodeString & pattern,int32_t patternOffset)1933 SimpleDateFormat::isAtNumericField(const UnicodeString &pattern, int32_t patternOffset) {
1934     if (patternOffset >= pattern.length()) {
1935         // not at any field
1936         return FALSE;
1937     }
1938     UChar ch = pattern.charAt(patternOffset);
1939     UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
1940     if (f == UDAT_FIELD_COUNT) {
1941         // not at any field
1942         return FALSE;
1943     }
1944     int32_t i = patternOffset;
1945     while (pattern.charAt(++i) == ch) {}
1946     return DateFormatSymbols::isNumericField(f, i - patternOffset);
1947 }
1948 
1949 UBool
isAfterNonNumericField(const UnicodeString & pattern,int32_t patternOffset)1950 SimpleDateFormat::isAfterNonNumericField(const UnicodeString &pattern, int32_t patternOffset) {
1951     if (patternOffset <= 0) {
1952         // not after any field
1953         return FALSE;
1954     }
1955     UChar ch = pattern.charAt(--patternOffset);
1956     UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
1957     if (f == UDAT_FIELD_COUNT) {
1958         // not after any field
1959         return FALSE;
1960     }
1961     int32_t i = patternOffset;
1962     while (pattern.charAt(--i) == ch) {}
1963     return !DateFormatSymbols::isNumericField(f, patternOffset - i);
1964 }
1965 
1966 void
parse(const UnicodeString & text,Calendar & cal,ParsePosition & parsePos) const1967 SimpleDateFormat::parse(const UnicodeString& text, Calendar& cal, ParsePosition& parsePos) const
1968 {
1969     UErrorCode status = U_ZERO_ERROR;
1970     int32_t pos = parsePos.getIndex();
1971     if(parsePos.getIndex() < 0) {
1972         parsePos.setErrorIndex(0);
1973         return;
1974     }
1975     int32_t start = pos;
1976 
1977 
1978     UBool ambiguousYear[] = { FALSE };
1979     int32_t saveHebrewMonth = -1;
1980     int32_t count = 0;
1981     UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
1982     SimpleDateFormatMutableNFs mutableNFs;
1983 
1984     // For parsing abutting numeric fields. 'abutPat' is the
1985     // offset into 'pattern' of the first of 2 or more abutting
1986     // numeric fields.  'abutStart' is the offset into 'text'
1987     // where parsing the fields begins. 'abutPass' starts off as 0
1988     // and increments each time we try to parse the fields.
1989     int32_t abutPat = -1; // If >=0, we are in a run of abutting numeric fields
1990     int32_t abutStart = 0;
1991     int32_t abutPass = 0;
1992     UBool inQuote = FALSE;
1993 
1994     MessageFormat * numericLeapMonthFormatter = NULL;
1995 
1996     Calendar* calClone = NULL;
1997     Calendar *workCal = &cal;
1998     if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
1999         // Different calendar type
2000         // We use the time/zone from the input calendar, but
2001         // do not use the input calendar for field calculation.
2002         calClone = fCalendar->clone();
2003         if (calClone != NULL) {
2004             calClone->setTime(cal.getTime(status),status);
2005             if (U_FAILURE(status)) {
2006                 goto ExitParse;
2007             }
2008             calClone->setTimeZone(cal.getTimeZone());
2009             workCal = calClone;
2010         } else {
2011             status = U_MEMORY_ALLOCATION_ERROR;
2012             goto ExitParse;
2013         }
2014     }
2015 
2016     if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
2017         numericLeapMonthFormatter = new MessageFormat(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric], fLocale, status);
2018         if (numericLeapMonthFormatter == NULL) {
2019              status = U_MEMORY_ALLOCATION_ERROR;
2020              goto ExitParse;
2021         } else if (U_FAILURE(status)) {
2022              goto ExitParse; // this will delete numericLeapMonthFormatter
2023         }
2024     }
2025 
2026     for (int32_t i=0; i<fPattern.length(); ++i) {
2027         UChar ch = fPattern.charAt(i);
2028 
2029         // Handle alphabetic field characters.
2030         if (!inQuote && isSyntaxChar(ch)) {
2031             int32_t fieldPat = i;
2032 
2033             // Count the length of this field specifier
2034             count = 1;
2035             while ((i+1)<fPattern.length() &&
2036                    fPattern.charAt(i+1) == ch) {
2037                 ++count;
2038                 ++i;
2039             }
2040 
2041             if (isNumeric(ch, count)) {
2042                 if (abutPat < 0) {
2043                     // Determine if there is an abutting numeric field.
2044                     // Record the start of a set of abutting numeric fields.
2045                     if (isAtNumericField(fPattern, i + 1)) {
2046                         abutPat = fieldPat;
2047                         abutStart = pos;
2048                         abutPass = 0;
2049                     }
2050                 }
2051             } else {
2052                 abutPat = -1; // End of any abutting fields
2053             }
2054 
2055             // Handle fields within a run of abutting numeric fields.  Take
2056             // the pattern "HHmmss" as an example. We will try to parse
2057             // 2/2/2 characters of the input text, then if that fails,
2058             // 1/2/2.  We only adjust the width of the leftmost field; the
2059             // others remain fixed.  This allows "123456" => 12:34:56, but
2060             // "12345" => 1:23:45.  Likewise, for the pattern "yyyyMMdd" we
2061             // try 4/2/2, 3/2/2, 2/2/2, and finally 1/2/2.
2062             if (abutPat >= 0) {
2063                 // If we are at the start of a run of abutting fields, then
2064                 // shorten this field in each pass.  If we can't shorten
2065                 // this field any more, then the parse of this set of
2066                 // abutting numeric fields has failed.
2067                 if (fieldPat == abutPat) {
2068                     count -= abutPass++;
2069                     if (count == 0) {
2070                         status = U_PARSE_ERROR;
2071                         goto ExitParse;
2072                     }
2073                 }
2074 
2075                 pos = subParse(text, pos, ch, count,
2076                                TRUE, FALSE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter, &tzTimeType, mutableNFs);
2077 
2078                 // If the parse fails anywhere in the run, back up to the
2079                 // start of the run and retry.
2080                 if (pos < 0) {
2081                     i = abutPat - 1;
2082                     pos = abutStart;
2083                     continue;
2084                 }
2085             }
2086 
2087             // Handle non-numeric fields and non-abutting numeric
2088             // fields.
2089             else if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
2090                 int32_t s = subParse(text, pos, ch, count,
2091                                FALSE, TRUE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter, &tzTimeType, mutableNFs);
2092 
2093                 if (s == -pos-1) {
2094                     // era not present, in special cases allow this to continue
2095                     // from the position where the era was expected
2096                     s = pos;
2097 
2098                     if (i+1 < fPattern.length()) {
2099                         // move to next pattern character
2100                         UChar ch = fPattern.charAt(i+1);
2101 
2102                         // check for whitespace
2103                         if (PatternProps::isWhiteSpace(ch)) {
2104                             i++;
2105                             // Advance over run in pattern
2106                             while ((i+1)<fPattern.length() &&
2107                                    PatternProps::isWhiteSpace(fPattern.charAt(i+1))) {
2108                                 ++i;
2109                             }
2110                         }
2111                     }
2112                 }
2113                 else if (s <= 0) {
2114                     status = U_PARSE_ERROR;
2115                     goto ExitParse;
2116                 }
2117                 pos = s;
2118             }
2119         }
2120 
2121         // Handle literal pattern characters.  These are any
2122         // quoted characters and non-alphabetic unquoted
2123         // characters.
2124         else {
2125 
2126             abutPat = -1; // End of any abutting fields
2127 
2128             if (! matchLiterals(fPattern, i, text, pos, getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status), getBooleanAttribute(UDAT_PARSE_PARTIAL_LITERAL_MATCH, status), isLenient())) {
2129                 status = U_PARSE_ERROR;
2130                 goto ExitParse;
2131             }
2132         }
2133     }
2134 
2135     // Special hack for trailing "." after non-numeric field.
2136     if (text.charAt(pos) == 0x2e && getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) {
2137         // only do if the last field is not numeric
2138         if (isAfterNonNumericField(fPattern, fPattern.length())) {
2139             pos++; // skip the extra "."
2140         }
2141     }
2142 
2143     // At this point the fields of Calendar have been set.  Calendar
2144     // will fill in default values for missing fields when the time
2145     // is computed.
2146 
2147     parsePos.setIndex(pos);
2148 
2149     // This part is a problem:  When we call parsedDate.after, we compute the time.
2150     // Take the date April 3 2004 at 2:30 am.  When this is first set up, the year
2151     // will be wrong if we're parsing a 2-digit year pattern.  It will be 1904.
2152     // April 3 1904 is a Sunday (unlike 2004) so it is the DST onset day.  2:30 am
2153     // is therefore an "impossible" time, since the time goes from 1:59 to 3:00 am
2154     // on that day.  It is therefore parsed out to fields as 3:30 am.  Then we
2155     // add 100 years, and get April 3 2004 at 3:30 am.  Note that April 3 2004 is
2156     // a Saturday, so it can have a 2:30 am -- and it should. [LIU]
2157     /*
2158         UDate parsedDate = calendar.getTime();
2159         if( ambiguousYear[0] && !parsedDate.after(fDefaultCenturyStart) ) {
2160             calendar.add(Calendar.YEAR, 100);
2161             parsedDate = calendar.getTime();
2162         }
2163     */
2164     // Because of the above condition, save off the fields in case we need to readjust.
2165     // The procedure we use here is not particularly efficient, but there is no other
2166     // way to do this given the API restrictions present in Calendar.  We minimize
2167     // inefficiency by only performing this computation when it might apply, that is,
2168     // when the two-digit year is equal to the start year, and thus might fall at the
2169     // front or the back of the default century.  This only works because we adjust
2170     // the year correctly to start with in other cases -- see subParse().
2171     if (ambiguousYear[0] || tzTimeType != UTZFMT_TIME_TYPE_UNKNOWN) // If this is true then the two-digit year == the default start year
2172     {
2173         // We need a copy of the fields, and we need to avoid triggering a call to
2174         // complete(), which will recalculate the fields.  Since we can't access
2175         // the fields[] array in Calendar, we clone the entire object.  This will
2176         // stop working if Calendar.clone() is ever rewritten to call complete().
2177         Calendar *copy;
2178         if (ambiguousYear[0]) {
2179             copy = cal.clone();
2180             // Check for failed cloning.
2181             if (copy == NULL) {
2182                 status = U_MEMORY_ALLOCATION_ERROR;
2183                 goto ExitParse;
2184             }
2185             UDate parsedDate = copy->getTime(status);
2186             // {sfb} check internalGetDefaultCenturyStart
2187             if (fHaveDefaultCentury && (parsedDate < fDefaultCenturyStart)) {
2188                 // We can't use add here because that does a complete() first.
2189                 cal.set(UCAL_YEAR, fDefaultCenturyStartYear + 100);
2190             }
2191             delete copy;
2192         }
2193 
2194         if (tzTimeType != UTZFMT_TIME_TYPE_UNKNOWN) {
2195             copy = cal.clone();
2196             // Check for failed cloning.
2197             if (copy == NULL) {
2198                 status = U_MEMORY_ALLOCATION_ERROR;
2199                 goto ExitParse;
2200             }
2201             const TimeZone & tz = cal.getTimeZone();
2202             BasicTimeZone *btz = NULL;
2203 
2204             if (dynamic_cast<const OlsonTimeZone *>(&tz) != NULL
2205                 || dynamic_cast<const SimpleTimeZone *>(&tz) != NULL
2206                 || dynamic_cast<const RuleBasedTimeZone *>(&tz) != NULL
2207                 || dynamic_cast<const VTimeZone *>(&tz) != NULL) {
2208                 btz = (BasicTimeZone*)&tz;
2209             }
2210 
2211             // Get local millis
2212             copy->set(UCAL_ZONE_OFFSET, 0);
2213             copy->set(UCAL_DST_OFFSET, 0);
2214             UDate localMillis = copy->getTime(status);
2215 
2216             // Make sure parsed time zone type (Standard or Daylight)
2217             // matches the rule used by the parsed time zone.
2218             int32_t raw, dst;
2219             if (btz != NULL) {
2220                 if (tzTimeType == UTZFMT_TIME_TYPE_STANDARD) {
2221                     btz->getOffsetFromLocal(localMillis,
2222                         BasicTimeZone::kStandard, BasicTimeZone::kStandard, raw, dst, status);
2223                 } else {
2224                     btz->getOffsetFromLocal(localMillis,
2225                         BasicTimeZone::kDaylight, BasicTimeZone::kDaylight, raw, dst, status);
2226                 }
2227             } else {
2228                 // No good way to resolve ambiguous time at transition,
2229                 // but following code work in most case.
2230                 tz.getOffset(localMillis, TRUE, raw, dst, status);
2231             }
2232 
2233             // Now, compare the results with parsed type, either standard or daylight saving time
2234             int32_t resolvedSavings = dst;
2235             if (tzTimeType == UTZFMT_TIME_TYPE_STANDARD) {
2236                 if (dst != 0) {
2237                     // Override DST_OFFSET = 0 in the result calendar
2238                     resolvedSavings = 0;
2239                 }
2240             } else { // tztype == TZTYPE_DST
2241                 if (dst == 0) {
2242                     if (btz != NULL) {
2243                         UDate time = localMillis + raw;
2244                         // We use the nearest daylight saving time rule.
2245                         TimeZoneTransition beforeTrs, afterTrs;
2246                         UDate beforeT = time, afterT = time;
2247                         int32_t beforeSav = 0, afterSav = 0;
2248                         UBool beforeTrsAvail, afterTrsAvail;
2249 
2250                         // Search for DST rule before or on the time
2251                         while (TRUE) {
2252                             beforeTrsAvail = btz->getPreviousTransition(beforeT, TRUE, beforeTrs);
2253                             if (!beforeTrsAvail) {
2254                                 break;
2255                             }
2256                             beforeT = beforeTrs.getTime() - 1;
2257                             beforeSav = beforeTrs.getFrom()->getDSTSavings();
2258                             if (beforeSav != 0) {
2259                                 break;
2260                             }
2261                         }
2262 
2263                         // Search for DST rule after the time
2264                         while (TRUE) {
2265                             afterTrsAvail = btz->getNextTransition(afterT, FALSE, afterTrs);
2266                             if (!afterTrsAvail) {
2267                                 break;
2268                             }
2269                             afterT = afterTrs.getTime();
2270                             afterSav = afterTrs.getTo()->getDSTSavings();
2271                             if (afterSav != 0) {
2272                                 break;
2273                             }
2274                         }
2275 
2276                         if (beforeTrsAvail && afterTrsAvail) {
2277                             if (time - beforeT > afterT - time) {
2278                                 resolvedSavings = afterSav;
2279                             } else {
2280                                 resolvedSavings = beforeSav;
2281                             }
2282                         } else if (beforeTrsAvail && beforeSav != 0) {
2283                             resolvedSavings = beforeSav;
2284                         } else if (afterTrsAvail && afterSav != 0) {
2285                             resolvedSavings = afterSav;
2286                         } else {
2287                             resolvedSavings = btz->getDSTSavings();
2288                         }
2289                     } else {
2290                         resolvedSavings = tz.getDSTSavings();
2291                     }
2292                     if (resolvedSavings == 0) {
2293                         // final fallback
2294                         resolvedSavings = U_MILLIS_PER_HOUR;
2295                     }
2296                 }
2297             }
2298             cal.set(UCAL_ZONE_OFFSET, raw);
2299             cal.set(UCAL_DST_OFFSET, resolvedSavings);
2300             delete copy;
2301         }
2302     }
2303 ExitParse:
2304     // Set the parsed result if local calendar is used
2305     // instead of the input calendar
2306     if (U_SUCCESS(status) && workCal != &cal) {
2307         cal.setTimeZone(workCal->getTimeZone());
2308         cal.setTime(workCal->getTime(status), status);
2309     }
2310 
2311     if (numericLeapMonthFormatter != NULL) {
2312         delete numericLeapMonthFormatter;
2313     }
2314     if (calClone != NULL) {
2315         delete calClone;
2316     }
2317 
2318     // If any Calendar calls failed, we pretend that we
2319     // couldn't parse the string, when in reality this isn't quite accurate--
2320     // we did parse it; the Calendar calls just failed.
2321     if (U_FAILURE(status)) {
2322         parsePos.setErrorIndex(pos);
2323         parsePos.setIndex(start);
2324     }
2325 }
2326 
2327 //----------------------------------------------------------------------
2328 
2329 static int32_t
2330 matchStringWithOptionalDot(const UnicodeString &text,
2331                             int32_t index,
2332                             const UnicodeString &data);
2333 
matchQuarterString(const UnicodeString & text,int32_t start,UCalendarDateFields field,const UnicodeString * data,int32_t dataCount,Calendar & cal) const2334 int32_t SimpleDateFormat::matchQuarterString(const UnicodeString& text,
2335                               int32_t start,
2336                               UCalendarDateFields field,
2337                               const UnicodeString* data,
2338                               int32_t dataCount,
2339                               Calendar& cal) const
2340 {
2341     int32_t i = 0;
2342     int32_t count = dataCount;
2343 
2344     // There may be multiple strings in the data[] array which begin with
2345     // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2346     // We keep track of the longest match, and return that.  Note that this
2347     // unfortunately requires us to test all array elements.
2348     int32_t bestMatchLength = 0, bestMatch = -1;
2349     UnicodeString bestMatchName;
2350 
2351     for (; i < count; ++i) {
2352         int32_t matchLength = 0;
2353         if ((matchLength = matchStringWithOptionalDot(text, start, data[i])) > bestMatchLength) {
2354             bestMatchLength = matchLength;
2355             bestMatch = i;
2356         }
2357     }
2358 
2359     if (bestMatch >= 0) {
2360         cal.set(field, bestMatch * 3);
2361         return start + bestMatchLength;
2362     }
2363 
2364     return -start;
2365 }
2366 
2367 //----------------------------------------------------------------------
matchLiterals(const UnicodeString & pattern,int32_t & patternOffset,const UnicodeString & text,int32_t & textOffset,UBool whitespaceLenient,UBool partialMatchLenient,UBool oldLeniency)2368 UBool SimpleDateFormat::matchLiterals(const UnicodeString &pattern,
2369                                       int32_t &patternOffset,
2370                                       const UnicodeString &text,
2371                                       int32_t &textOffset,
2372                                       UBool whitespaceLenient,
2373                                       UBool partialMatchLenient,
2374                                       UBool oldLeniency)
2375 {
2376     UBool inQuote = FALSE;
2377     UnicodeString literal;
2378     int32_t i = patternOffset;
2379 
2380     // scan pattern looking for contiguous literal characters
2381     for ( ; i < pattern.length(); i += 1) {
2382         UChar ch = pattern.charAt(i);
2383 
2384         if (!inQuote && isSyntaxChar(ch)) {
2385             break;
2386         }
2387 
2388         if (ch == QUOTE) {
2389             // Match a quote literal ('') inside OR outside of quotes
2390             if ((i + 1) < pattern.length() && pattern.charAt(i + 1) == QUOTE) {
2391                 i += 1;
2392             } else {
2393                 inQuote = !inQuote;
2394                 continue;
2395             }
2396         }
2397 
2398         literal += ch;
2399     }
2400 
2401     // at this point, literal contains the literal text
2402     // and i is the index of the next non-literal pattern character.
2403     int32_t p;
2404     int32_t t = textOffset;
2405 
2406     if (whitespaceLenient) {
2407         // trim leading, trailing whitespace from
2408         // the literal text
2409         literal.trim();
2410 
2411         // ignore any leading whitespace in the text
2412         while (t < text.length() && u_isWhitespace(text.charAt(t))) {
2413             t += 1;
2414         }
2415     }
2416 
2417     for (p = 0; p < literal.length() && t < text.length();) {
2418         UBool needWhitespace = FALSE;
2419 
2420         while (p < literal.length() && PatternProps::isWhiteSpace(literal.charAt(p))) {
2421             needWhitespace = TRUE;
2422             p += 1;
2423         }
2424 
2425         if (needWhitespace) {
2426             int32_t tStart = t;
2427 
2428             while (t < text.length()) {
2429                 UChar tch = text.charAt(t);
2430 
2431                 if (!u_isUWhiteSpace(tch) && !PatternProps::isWhiteSpace(tch)) {
2432                     break;
2433                 }
2434 
2435                 t += 1;
2436             }
2437 
2438             // TODO: should we require internal spaces
2439             // in lenient mode? (There won't be any
2440             // leading or trailing spaces)
2441             if (!whitespaceLenient && t == tStart) {
2442                 // didn't find matching whitespace:
2443                 // an error in strict mode
2444                 return FALSE;
2445             }
2446 
2447             // In strict mode, this run of whitespace
2448             // may have been at the end.
2449             if (p >= literal.length()) {
2450                 break;
2451             }
2452         }
2453         if (t >= text.length() || literal.charAt(p) != text.charAt(t)) {
2454             // Ran out of text, or found a non-matching character:
2455             // OK in lenient mode, an error in strict mode.
2456             if (whitespaceLenient) {
2457                 if (t == textOffset && text.charAt(t) == 0x2e &&
2458                         isAfterNonNumericField(pattern, patternOffset)) {
2459                     // Lenient mode and the literal input text begins with a "." and
2460                     // we are after a non-numeric field: We skip the "."
2461                     ++t;
2462                     continue;  // Do not update p.
2463                 }
2464                 // if it is actual whitespace and we're whitespace lenient it's OK
2465 
2466                 UChar wsc = text.charAt(t);
2467                 if(PatternProps::isWhiteSpace(wsc)) {
2468                     // Lenient mode and it's just whitespace we skip it
2469                     ++t;
2470                     continue;  // Do not update p.
2471                 }
2472             }
2473             // hack around oldleniency being a bit of a catch-all bucket and we're just adding support specifically for paritial matches
2474             if(partialMatchLenient && oldLeniency) {
2475                 break;
2476             }
2477 
2478             return FALSE;
2479         }
2480         ++p;
2481         ++t;
2482     }
2483 
2484     // At this point if we're in strict mode we have a complete match.
2485     // If we're in lenient mode we may have a partial match, or no
2486     // match at all.
2487     if (p <= 0) {
2488         // no match. Pretend it matched a run of whitespace
2489         // and ignorables in the text.
2490         const  UnicodeSet *ignorables = NULL;
2491         UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(pattern.charAt(i));
2492         if (patternCharIndex != UDAT_FIELD_COUNT) {
2493             ignorables = SimpleDateFormatStaticSets::getIgnorables(patternCharIndex);
2494         }
2495 
2496         for (t = textOffset; t < text.length(); t += 1) {
2497             UChar ch = text.charAt(t);
2498 
2499             if (ignorables == NULL || !ignorables->contains(ch)) {
2500                 break;
2501             }
2502         }
2503     }
2504 
2505     // if we get here, we've got a complete match.
2506     patternOffset = i - 1;
2507     textOffset = t;
2508 
2509     return TRUE;
2510 }
2511 
2512 //----------------------------------------------------------------------
2513 
matchString(const UnicodeString & text,int32_t start,UCalendarDateFields field,const UnicodeString * data,int32_t dataCount,const UnicodeString * monthPattern,Calendar & cal) const2514 int32_t SimpleDateFormat::matchString(const UnicodeString& text,
2515                               int32_t start,
2516                               UCalendarDateFields field,
2517                               const UnicodeString* data,
2518                               int32_t dataCount,
2519                               const UnicodeString* monthPattern,
2520                               Calendar& cal) const
2521 {
2522     int32_t i = 0;
2523     int32_t count = dataCount;
2524 
2525     if (field == UCAL_DAY_OF_WEEK) i = 1;
2526 
2527     // There may be multiple strings in the data[] array which begin with
2528     // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2529     // We keep track of the longest match, and return that.  Note that this
2530     // unfortunately requires us to test all array elements.
2531     int32_t bestMatchLength = 0, bestMatch = -1;
2532     UnicodeString bestMatchName;
2533     int32_t isLeapMonth = 0;
2534 
2535     for (; i < count; ++i) {
2536         int32_t matchLen = 0;
2537         if ((matchLen = matchStringWithOptionalDot(text, start, data[i])) > bestMatchLength) {
2538             bestMatch = i;
2539             bestMatchLength = matchLen;
2540         }
2541 
2542         if (monthPattern != NULL) {
2543             UErrorCode status = U_ZERO_ERROR;
2544             UnicodeString leapMonthName;
2545             Formattable monthName((const UnicodeString&)(data[i]));
2546             MessageFormat::format(*monthPattern, &monthName, 1, leapMonthName, status);
2547             if (U_SUCCESS(status)) {
2548                 if ((matchLen = matchStringWithOptionalDot(text, start, leapMonthName)) > bestMatchLength) {
2549                     bestMatch = i;
2550                     bestMatchLength = matchLen;
2551                     isLeapMonth = 1;
2552                 }
2553             }
2554         }
2555     }
2556 
2557     if (bestMatch >= 0) {
2558         if (field < UCAL_FIELD_COUNT) {
2559             // Adjustment for Hebrew Calendar month Adar II
2560             if (!strcmp(cal.getType(),"hebrew") && field==UCAL_MONTH && bestMatch==13) {
2561                 cal.set(field,6);
2562             } else {
2563                 if (field == UCAL_YEAR) {
2564                     bestMatch++; // only get here for cyclic year names, which match 1-based years 1-60
2565                 }
2566                 cal.set(field, bestMatch);
2567             }
2568             if (monthPattern != NULL) {
2569                 cal.set(UCAL_IS_LEAP_MONTH, isLeapMonth);
2570             }
2571         }
2572 
2573         return start + bestMatchLength;
2574     }
2575 
2576     return -start;
2577 }
2578 
2579 static int32_t
matchStringWithOptionalDot(const UnicodeString & text,int32_t index,const UnicodeString & data)2580 matchStringWithOptionalDot(const UnicodeString &text,
2581                             int32_t index,
2582                             const UnicodeString &data) {
2583     UErrorCode sts = U_ZERO_ERROR;
2584     int32_t matchLenText = 0;
2585     int32_t matchLenData = 0;
2586 
2587     u_caseInsensitivePrefixMatch(text.getBuffer() + index, text.length() - index,
2588                                  data.getBuffer(), data.length(),
2589                                  0 /* default case option */,
2590                                  &matchLenText, &matchLenData,
2591                                  &sts);
2592     U_ASSERT (U_SUCCESS(sts));
2593 
2594     if (matchLenData == data.length() /* normal match */
2595         || (data.charAt(data.length() - 1) == 0x2e
2596             && matchLenData == data.length() - 1 /* match without trailing dot */)) {
2597         return matchLenText;
2598     }
2599 
2600     return 0;
2601 }
2602 
2603 //----------------------------------------------------------------------
2604 
2605 void
set2DigitYearStart(UDate d,UErrorCode & status)2606 SimpleDateFormat::set2DigitYearStart(UDate d, UErrorCode& status)
2607 {
2608     parseAmbiguousDatesAsAfter(d, status);
2609 }
2610 
2611 /**
2612  * Private member function that converts the parsed date strings into
2613  * timeFields. Returns -start (for ParsePosition) if failed.
2614  */
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,UTimeZoneFormatTimeType * tzTimeType,SimpleDateFormatMutableNFs & mutableNFs) const2615 int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UChar ch, int32_t count,
2616                            UBool obeyCount, UBool allowNegative, UBool ambiguousYear[], int32_t& saveHebrewMonth, Calendar& cal,
2617                            int32_t patLoc, MessageFormat * numericLeapMonthFormatter, UTimeZoneFormatTimeType *tzTimeType, SimpleDateFormatMutableNFs &mutableNFs) const
2618 {
2619     Formattable number;
2620     int32_t value = 0;
2621     int32_t i;
2622     int32_t ps = 0;
2623     UErrorCode status = U_ZERO_ERROR;
2624     ParsePosition pos(0);
2625     UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
2626     NumberFormat *currentNumberFormat;
2627     UnicodeString temp;
2628     UBool gotNumber = FALSE;
2629 
2630 #if defined (U_DEBUG_CAL)
2631     //fprintf(stderr, "%s:%d - [%c]  st=%d \n", __FILE__, __LINE__, (char) ch, start);
2632 #endif
2633 
2634     if (patternCharIndex == UDAT_FIELD_COUNT) {
2635         return -start;
2636     }
2637 
2638     currentNumberFormat = mutableNFs.get(getNumberFormatByIndex(patternCharIndex));
2639     if (currentNumberFormat == NULL) {
2640         return -start;
2641     }
2642     UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex]; // UCAL_FIELD_COUNT if irrelevant
2643     UnicodeString hebr("hebr", 4, US_INV);
2644 
2645     if (numericLeapMonthFormatter != NULL) {
2646         numericLeapMonthFormatter->setFormats((const Format **)&currentNumberFormat, 1);
2647     }
2648     UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0 || uprv_strcmp(cal.getType(),"dangi") == 0);
2649 
2650     // If there are any spaces here, skip over them.  If we hit the end
2651     // of the string, then fail.
2652     for (;;) {
2653         if (start >= text.length()) {
2654             return -start;
2655         }
2656         UChar32 c = text.char32At(start);
2657         if (!u_isUWhiteSpace(c) /*||*/ && !PatternProps::isWhiteSpace(c)) {
2658             break;
2659         }
2660         start += U16_LENGTH(c);
2661     }
2662     pos.setIndex(start);
2663 
2664     // We handle a few special cases here where we need to parse
2665     // a number value.  We handle further, more generic cases below.  We need
2666     // to handle some of them here because some fields require extra processing on
2667     // the parsed value.
2668     if (patternCharIndex == UDAT_HOUR_OF_DAY1_FIELD ||                       // k
2669         patternCharIndex == UDAT_HOUR_OF_DAY0_FIELD ||                       // H
2670         patternCharIndex == UDAT_HOUR1_FIELD ||                              // h
2671         patternCharIndex == UDAT_HOUR0_FIELD ||                              // K
2672         (patternCharIndex == UDAT_DOW_LOCAL_FIELD && count <= 2) ||          // e
2673         (patternCharIndex == UDAT_STANDALONE_DAY_FIELD && count <= 2) ||     // c
2674         (patternCharIndex == UDAT_MONTH_FIELD && count <= 2) ||              // M
2675         (patternCharIndex == UDAT_STANDALONE_MONTH_FIELD && count <= 2) ||   // L
2676         (patternCharIndex == UDAT_QUARTER_FIELD && count <= 2) ||            // Q
2677         (patternCharIndex == UDAT_STANDALONE_QUARTER_FIELD && count <= 2) || // q
2678         patternCharIndex == UDAT_YEAR_FIELD ||                               // y
2679         patternCharIndex == UDAT_YEAR_WOY_FIELD ||                           // Y
2680         patternCharIndex == UDAT_YEAR_NAME_FIELD ||                          // U (falls back to numeric)
2681         (patternCharIndex == UDAT_ERA_FIELD && isChineseCalendar) ||         // G
2682         patternCharIndex == UDAT_FRACTIONAL_SECOND_FIELD)                    // S
2683     {
2684         int32_t parseStart = pos.getIndex();
2685         // It would be good to unify this with the obeyCount logic below,
2686         // but that's going to be difficult.
2687         const UnicodeString* src;
2688 
2689         UBool parsedNumericLeapMonth = FALSE;
2690         if (numericLeapMonthFormatter != NULL && (patternCharIndex == UDAT_MONTH_FIELD || patternCharIndex == UDAT_STANDALONE_MONTH_FIELD)) {
2691             int32_t argCount;
2692             Formattable * args = numericLeapMonthFormatter->parse(text, pos, argCount);
2693             if (args != NULL && argCount == 1 && pos.getIndex() > parseStart && args[0].isNumeric()) {
2694                 parsedNumericLeapMonth = TRUE;
2695                 number.setLong(args[0].getLong());
2696                 cal.set(UCAL_IS_LEAP_MONTH, 1);
2697                 delete[] args;
2698             } else {
2699                 pos.setIndex(parseStart);
2700                 cal.set(UCAL_IS_LEAP_MONTH, 0);
2701             }
2702         }
2703 
2704         if (!parsedNumericLeapMonth) {
2705             if (obeyCount) {
2706                 if ((start+count) > text.length()) {
2707                     return -start;
2708                 }
2709 
2710                 text.extractBetween(0, start + count, temp);
2711                 src = &temp;
2712             } else {
2713                 src = &text;
2714             }
2715 
2716             parseInt(*src, number, pos, allowNegative,currentNumberFormat);
2717         }
2718 
2719         int32_t txtLoc = pos.getIndex();
2720 
2721         if (txtLoc > parseStart) {
2722             value = number.getLong();
2723             gotNumber = TRUE;
2724 
2725             // suffix processing
2726             if (value < 0 ) {
2727                 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, TRUE);
2728                 if (txtLoc != pos.getIndex()) {
2729                     value *= -1;
2730                 }
2731             }
2732             else {
2733                 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, FALSE);
2734             }
2735 
2736             if (!getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) {
2737                 // Check the range of the value
2738                 int32_t bias = gFieldRangeBias[patternCharIndex];
2739                 if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
2740                     return -start;
2741                 }
2742             }
2743 
2744             pos.setIndex(txtLoc);
2745         }
2746     }
2747 
2748     // Make sure that we got a number if
2749     // we want one, and didn't get one
2750     // if we don't want one.
2751     switch (patternCharIndex) {
2752         case UDAT_HOUR_OF_DAY1_FIELD:
2753         case UDAT_HOUR_OF_DAY0_FIELD:
2754         case UDAT_HOUR1_FIELD:
2755         case UDAT_HOUR0_FIELD:
2756             // special range check for hours:
2757             if (value < 0 || value > 24) {
2758                 return -start;
2759             }
2760 
2761             // fall through to gotNumber check
2762 
2763         case UDAT_YEAR_FIELD:
2764         case UDAT_YEAR_WOY_FIELD:
2765         case UDAT_FRACTIONAL_SECOND_FIELD:
2766             // these must be a number
2767             if (! gotNumber) {
2768                 return -start;
2769             }
2770 
2771             break;
2772 
2773         default:
2774             // we check the rest of the fields below.
2775             break;
2776     }
2777 
2778     switch (patternCharIndex) {
2779     case UDAT_ERA_FIELD:
2780         if (isChineseCalendar) {
2781             if (!gotNumber) {
2782                 return -start;
2783             }
2784             cal.set(UCAL_ERA, value);
2785             return pos.getIndex();
2786         }
2787         if (count == 5) {
2788             ps = matchString(text, start, UCAL_ERA, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount, NULL, cal);
2789         } else if (count == 4) {
2790             ps = matchString(text, start, UCAL_ERA, fSymbols->fEraNames, fSymbols->fEraNamesCount, NULL, cal);
2791         } else {
2792             ps = matchString(text, start, UCAL_ERA, fSymbols->fEras, fSymbols->fErasCount, NULL, cal);
2793         }
2794 
2795         // check return position, if it equals -start, then matchString error
2796         // special case the return code so we don't necessarily fail out until we
2797         // verify no year information also
2798         if (ps == -start)
2799             ps--;
2800 
2801         return ps;
2802 
2803     case UDAT_YEAR_FIELD:
2804         // If there are 3 or more YEAR pattern characters, this indicates
2805         // that the year value is to be treated literally, without any
2806         // two-digit year adjustments (e.g., from "01" to 2001).  Otherwise
2807         // we made adjustments to place the 2-digit year in the proper
2808         // century, for parsed strings from "00" to "99".  Any other string
2809         // is treated literally:  "2250", "-1", "1", "002".
2810         if (fDateOverride.compare(hebr)==0 && value < 1000) {
2811             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2812         } else if ((pos.getIndex() - start) == 2 && !isChineseCalendar
2813             && u_isdigit(text.charAt(start))
2814             && u_isdigit(text.charAt(start+1)))
2815         {
2816             // only adjust year for patterns less than 3.
2817             if(count < 3) {
2818                 // Assume for example that the defaultCenturyStart is 6/18/1903.
2819                 // This means that two-digit years will be forced into the range
2820                 // 6/18/1903 to 6/17/2003.  As a result, years 00, 01, and 02
2821                 // correspond to 2000, 2001, and 2002.  Years 04, 05, etc. correspond
2822                 // to 1904, 1905, etc.  If the year is 03, then it is 2003 if the
2823                 // other fields specify a date before 6/18, or 1903 if they specify a
2824                 // date afterwards.  As a result, 03 is an ambiguous year.  All other
2825                 // two-digit years are unambiguous.
2826                 if(fHaveDefaultCentury) { // check if this formatter even has a pivot year
2827                     int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2828                     ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2829                     value += (fDefaultCenturyStartYear/100)*100 +
2830                             (value < ambiguousTwoDigitYear ? 100 : 0);
2831                 }
2832             }
2833         }
2834         cal.set(UCAL_YEAR, value);
2835 
2836         // Delayed checking for adjustment of Hebrew month numbers in non-leap years.
2837         if (saveHebrewMonth >= 0) {
2838             HebrewCalendar *hc = (HebrewCalendar*)&cal;
2839             if (!hc->isLeapYear(value) && saveHebrewMonth >= 6) {
2840                cal.set(UCAL_MONTH,saveHebrewMonth);
2841             } else {
2842                cal.set(UCAL_MONTH,saveHebrewMonth-1);
2843             }
2844             saveHebrewMonth = -1;
2845         }
2846         return pos.getIndex();
2847 
2848     case UDAT_YEAR_WOY_FIELD:
2849         // Comment is the same as for UDAT_Year_FIELDs - look above
2850         if (fDateOverride.compare(hebr)==0 && value < 1000) {
2851             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2852         } else if ((pos.getIndex() - start) == 2
2853             && u_isdigit(text.charAt(start))
2854             && u_isdigit(text.charAt(start+1))
2855             && fHaveDefaultCentury )
2856         {
2857             int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2858             ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2859             value += (fDefaultCenturyStartYear/100)*100 +
2860                 (value < ambiguousTwoDigitYear ? 100 : 0);
2861         }
2862         cal.set(UCAL_YEAR_WOY, value);
2863         return pos.getIndex();
2864 
2865     case UDAT_YEAR_NAME_FIELD:
2866         if (fSymbols->fShortYearNames != NULL) {
2867             int32_t newStart = matchString(text, start, UCAL_YEAR, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount, NULL, cal);
2868             if (newStart > 0) {
2869                 return newStart;
2870             }
2871         }
2872         if (gotNumber && (getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC,status) || value > fSymbols->fShortYearNamesCount)) {
2873             cal.set(UCAL_YEAR, value);
2874             return pos.getIndex();
2875         }
2876         return -start;
2877 
2878     case UDAT_MONTH_FIELD:
2879     case UDAT_STANDALONE_MONTH_FIELD:
2880         if (gotNumber) // i.e., M or MM.
2881         {
2882             // When parsing month numbers from the Hebrew Calendar, we might need to adjust the month depending on whether
2883             // 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
2884             // the year is parsed.
2885             if (!strcmp(cal.getType(),"hebrew")) {
2886                 HebrewCalendar *hc = (HebrewCalendar*)&cal;
2887                 if (cal.isSet(UCAL_YEAR)) {
2888                    UErrorCode status = U_ZERO_ERROR;
2889                    if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
2890                        cal.set(UCAL_MONTH, value);
2891                    } else {
2892                        cal.set(UCAL_MONTH, value - 1);
2893                    }
2894                 } else {
2895                     saveHebrewMonth = value;
2896                 }
2897             } else {
2898                 // Don't want to parse the month if it is a string
2899                 // while pattern uses numeric style: M/MM, L/LL
2900                 // [We computed 'value' above.]
2901                 cal.set(UCAL_MONTH, value - 1);
2902             }
2903             return pos.getIndex();
2904         } else {
2905             // count >= 3 // i.e., MMM/MMMM, LLL/LLLL
2906             // Want to be able to parse both short and long forms.
2907             // Try count == 4 first:
2908             UnicodeString * wideMonthPat = NULL;
2909             UnicodeString * shortMonthPat = NULL;
2910             if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
2911                 if (patternCharIndex==UDAT_MONTH_FIELD) {
2912                     wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide];
2913                     shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev];
2914                 } else {
2915                     wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide];
2916                     shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev];
2917                 }
2918             }
2919             int32_t newStart = 0;
2920             if (patternCharIndex==UDAT_MONTH_FIELD) {
2921                 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
2922                     newStart = matchString(text, start, UCAL_MONTH, fSymbols->fMonths, fSymbols->fMonthsCount, wideMonthPat, cal); // try MMMM
2923                     if (newStart > 0) {
2924                         return newStart;
2925                     }
2926                 }
2927                 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
2928                     newStart = matchString(text, start, UCAL_MONTH, fSymbols->fShortMonths, fSymbols->fShortMonthsCount, shortMonthPat, cal); // try MMM
2929                 }
2930             } else {
2931                 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
2932                     newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount, wideMonthPat, cal); // try LLLL
2933                     if (newStart > 0) {
2934                         return newStart;
2935                     }
2936                 }
2937                 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
2938                     newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount, shortMonthPat, cal); // try LLL
2939                 }
2940             }
2941             if (newStart > 0 || !getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))  // currently we do not try to parse MMMMM/LLLLL: #8860
2942                 return newStart;
2943             // else we allowing parsing as number, below
2944         }
2945         break;
2946 
2947     case UDAT_HOUR_OF_DAY1_FIELD:
2948         // [We computed 'value' above.]
2949         if (value == cal.getMaximum(UCAL_HOUR_OF_DAY) + 1)
2950             value = 0;
2951 
2952         // fall through to set field
2953 
2954     case UDAT_HOUR_OF_DAY0_FIELD:
2955         cal.set(UCAL_HOUR_OF_DAY, value);
2956         return pos.getIndex();
2957 
2958     case UDAT_FRACTIONAL_SECOND_FIELD:
2959         // Fractional seconds left-justify
2960         i = pos.getIndex() - start;
2961         if (i < 3) {
2962             while (i < 3) {
2963                 value *= 10;
2964                 i++;
2965             }
2966         } else {
2967             int32_t a = 1;
2968             while (i > 3) {
2969                 a *= 10;
2970                 i--;
2971             }
2972             value /= a;
2973         }
2974         cal.set(UCAL_MILLISECOND, value);
2975         return pos.getIndex();
2976 
2977     case UDAT_DOW_LOCAL_FIELD:
2978         if (gotNumber) // i.e., e or ee
2979         {
2980             // [We computed 'value' above.]
2981             cal.set(UCAL_DOW_LOCAL, value);
2982             return pos.getIndex();
2983         }
2984         // else for eee-eeeee fall through to handling of EEE-EEEEE
2985         // fall through, do not break here
2986     case UDAT_DAY_OF_WEEK_FIELD:
2987         {
2988             // Want to be able to parse both short and long forms.
2989             // Try count == 4 (EEEE) wide first:
2990             int32_t newStart = 0;
2991             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
2992                 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2993                                           fSymbols->fWeekdays, fSymbols->fWeekdaysCount, NULL, cal)) > 0)
2994                     return newStart;
2995             }
2996             // EEEE wide failed, now try EEE abbreviated
2997             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
2998                 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2999                                        fSymbols->fShortWeekdays, fSymbols->fShortWeekdaysCount, NULL, cal)) > 0)
3000                     return newStart;
3001             }
3002             // EEE abbreviated failed, now try EEEEEE short
3003             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) {
3004                 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3005                                        fSymbols->fShorterWeekdays, fSymbols->fShorterWeekdaysCount, NULL, cal)) > 0)
3006                     return newStart;
3007             }
3008             // EEEEEE short failed, now try EEEEE narrow
3009             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 5) {
3010                 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3011                                        fSymbols->fNarrowWeekdays, fSymbols->fNarrowWeekdaysCount, NULL, cal)) > 0)
3012                     return newStart;
3013             }
3014             if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status) || patternCharIndex == UDAT_DAY_OF_WEEK_FIELD)
3015                 return newStart;
3016             // else we allowing parsing as number, below
3017         }
3018         break;
3019 
3020     case UDAT_STANDALONE_DAY_FIELD:
3021         {
3022             if (gotNumber) // c or cc
3023             {
3024                 // [We computed 'value' above.]
3025                 cal.set(UCAL_DOW_LOCAL, value);
3026                 return pos.getIndex();
3027             }
3028             // Want to be able to parse both short and long forms.
3029             // Try count == 4 (cccc) first:
3030             int32_t newStart = 0;
3031             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3032                 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3033                                       fSymbols->fStandaloneWeekdays, fSymbols->fStandaloneWeekdaysCount, NULL, cal)) > 0)
3034                     return newStart;
3035             }
3036             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3037                 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3038                                           fSymbols->fStandaloneShortWeekdays, fSymbols->fStandaloneShortWeekdaysCount, NULL, cal)) > 0)
3039                     return newStart;
3040             }
3041             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) {
3042                 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3043                                           fSymbols->fStandaloneShorterWeekdays, fSymbols->fStandaloneShorterWeekdaysCount, NULL, cal)) > 0)
3044                     return newStart;
3045             }
3046             if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
3047                 return newStart;
3048             // else we allowing parsing as number, below
3049         }
3050         break;
3051 
3052     case UDAT_AM_PM_FIELD:
3053         {
3054             // optionally try both wide/abbrev and narrow forms
3055             int32_t newStart = 0;
3056             // try wide/abbrev
3057             if( getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count < 5 ) {
3058                 if ((newStart = matchString(text, start, UCAL_AM_PM, fSymbols->fAmPms, fSymbols->fAmPmsCount, NULL, cal)) > 0) {
3059                     return newStart;
3060                 }
3061             }
3062             // try narrow
3063             if( getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count >= 5 ) {
3064                 if ((newStart = matchString(text, start, UCAL_AM_PM, fSymbols->fNarrowAmPms, fSymbols->fNarrowAmPmsCount, NULL, cal)) > 0) {
3065                     return newStart;
3066                 }
3067             }
3068             // no matches for given options
3069             return -start;
3070         }
3071 
3072     case UDAT_HOUR1_FIELD:
3073         // [We computed 'value' above.]
3074         if (value == cal.getLeastMaximum(UCAL_HOUR)+1)
3075             value = 0;
3076 
3077         // fall through to set field
3078 
3079     case UDAT_HOUR0_FIELD:
3080         cal.set(UCAL_HOUR, value);
3081         return pos.getIndex();
3082 
3083     case UDAT_QUARTER_FIELD:
3084         if (gotNumber) // i.e., Q or QQ.
3085         {
3086             // Don't want to parse the month if it is a string
3087             // while pattern uses numeric style: Q or QQ.
3088             // [We computed 'value' above.]
3089             cal.set(UCAL_MONTH, (value - 1) * 3);
3090             return pos.getIndex();
3091         } else {
3092             // count >= 3 // i.e., QQQ or QQQQ
3093             // Want to be able to parse both short and long forms.
3094             // Try count == 4 first:
3095             int32_t newStart = 0;
3096 
3097             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3098                 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3099                                       fSymbols->fQuarters, fSymbols->fQuartersCount, cal)) > 0)
3100                     return newStart;
3101             }
3102             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3103                 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3104                                           fSymbols->fShortQuarters, fSymbols->fShortQuartersCount, cal)) > 0)
3105                     return newStart;
3106             }
3107             if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
3108                 return newStart;
3109             // else we allowing parsing as number, below
3110             if(!getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status))
3111                 return -start;
3112         }
3113         break;
3114 
3115     case UDAT_STANDALONE_QUARTER_FIELD:
3116         if (gotNumber) // i.e., q or qq.
3117         {
3118             // Don't want to parse the month if it is a string
3119             // while pattern uses numeric style: q or q.
3120             // [We computed 'value' above.]
3121             cal.set(UCAL_MONTH, (value - 1) * 3);
3122             return pos.getIndex();
3123         } else {
3124             // count >= 3 // i.e., qqq or qqqq
3125             // Want to be able to parse both short and long forms.
3126             // Try count == 4 first:
3127             int32_t newStart = 0;
3128 
3129             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3130                 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3131                                       fSymbols->fStandaloneQuarters, fSymbols->fStandaloneQuartersCount, cal)) > 0)
3132                     return newStart;
3133             }
3134             if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3135                 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3136                                           fSymbols->fStandaloneShortQuarters, fSymbols->fStandaloneShortQuartersCount, cal)) > 0)
3137                     return newStart;
3138             }
3139             if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
3140                 return newStart;
3141             // else we allowing parsing as number, below
3142             if(!getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status))
3143                 return -start;
3144         }
3145         break;
3146 
3147     case UDAT_TIMEZONE_FIELD: // 'z'
3148         {
3149             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_SPECIFIC_SHORT : UTZFMT_STYLE_SPECIFIC_LONG;
3150             TimeZone *tz  = tzFormat()->parse(style, text, pos, tzTimeType);
3151             if (tz != NULL) {
3152                 cal.adoptTimeZone(tz);
3153                 return pos.getIndex();
3154             }
3155         }
3156         break;
3157     case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
3158         {
3159             UTimeZoneFormatStyle style = (count < 4) ?
3160                 UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL : ((count == 5) ? UTZFMT_STYLE_ISO_EXTENDED_FULL: UTZFMT_STYLE_LOCALIZED_GMT);
3161             TimeZone *tz  = tzFormat()->parse(style, text, pos, tzTimeType);
3162             if (tz != NULL) {
3163                 cal.adoptTimeZone(tz);
3164                 return pos.getIndex();
3165             }
3166             return -start;
3167         }
3168     case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
3169         {
3170             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_GENERIC_SHORT : UTZFMT_STYLE_GENERIC_LONG;
3171             TimeZone *tz  = tzFormat()->parse(style, text, pos, tzTimeType);
3172             if (tz != NULL) {
3173                 cal.adoptTimeZone(tz);
3174                 return pos.getIndex();
3175             }
3176             return -start;
3177         }
3178     case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
3179         {
3180             UTimeZoneFormatStyle style;
3181             switch (count) {
3182             case 1:
3183                 style = UTZFMT_STYLE_ZONE_ID_SHORT;
3184                 break;
3185             case 2:
3186                 style = UTZFMT_STYLE_ZONE_ID;
3187                 break;
3188             case 3:
3189                 style = UTZFMT_STYLE_EXEMPLAR_LOCATION;
3190                 break;
3191             default:
3192                 style = UTZFMT_STYLE_GENERIC_LOCATION;
3193                 break;
3194             }
3195             TimeZone *tz  = tzFormat()->parse(style, text, pos, tzTimeType);
3196             if (tz != NULL) {
3197                 cal.adoptTimeZone(tz);
3198                 return pos.getIndex();
3199             }
3200             return -start;
3201         }
3202     case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
3203         {
3204             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_LOCALIZED_GMT_SHORT : UTZFMT_STYLE_LOCALIZED_GMT;
3205             TimeZone *tz  = tzFormat()->parse(style, text, pos, tzTimeType);
3206             if (tz != NULL) {
3207                 cal.adoptTimeZone(tz);
3208                 return pos.getIndex();
3209             }
3210             return -start;
3211         }
3212     case UDAT_TIMEZONE_ISO_FIELD: // 'X'
3213         {
3214             UTimeZoneFormatStyle style;
3215             switch (count) {
3216             case 1:
3217                 style = UTZFMT_STYLE_ISO_BASIC_SHORT;
3218                 break;
3219             case 2:
3220                 style = UTZFMT_STYLE_ISO_BASIC_FIXED;
3221                 break;
3222             case 3:
3223                 style = UTZFMT_STYLE_ISO_EXTENDED_FIXED;
3224                 break;
3225             case 4:
3226                 style = UTZFMT_STYLE_ISO_BASIC_FULL;
3227                 break;
3228             default:
3229                 style = UTZFMT_STYLE_ISO_EXTENDED_FULL;
3230                 break;
3231             }
3232             TimeZone *tz  = tzFormat()->parse(style, text, pos, tzTimeType);
3233             if (tz != NULL) {
3234                 cal.adoptTimeZone(tz);
3235                 return pos.getIndex();
3236             }
3237             return -start;
3238         }
3239     case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
3240         {
3241             UTimeZoneFormatStyle style;
3242             switch (count) {
3243             case 1:
3244                 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT;
3245                 break;
3246             case 2:
3247                 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED;
3248                 break;
3249             case 3:
3250                 style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED;
3251                 break;
3252             case 4:
3253                 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL;
3254                 break;
3255             default:
3256                 style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL;
3257                 break;
3258             }
3259             TimeZone *tz  = tzFormat()->parse(style, text, pos, tzTimeType);
3260             if (tz != NULL) {
3261                 cal.adoptTimeZone(tz);
3262                 return pos.getIndex();
3263             }
3264             return -start;
3265         }
3266     // currently no pattern character is defined for UDAT_TIME_SEPARATOR_FIELD
3267     // so we should not get here. Leave support in for future definition.
3268     case UDAT_TIME_SEPARATOR_FIELD: //
3269         {
3270             static const UChar def_sep = DateFormatSymbols::DEFAULT_TIME_SEPARATOR;
3271             static const UChar alt_sep = DateFormatSymbols::ALTERNATE_TIME_SEPARATOR;
3272 
3273             // Try matching a time separator.
3274             int32_t count = 1;
3275             UnicodeString data[3];
3276             fSymbols->getTimeSeparatorString(data[0]);
3277 
3278             // Add the default, if different from the locale.
3279             if (data[0].compare(&def_sep, 1) != 0) {
3280                 data[count++].setTo(def_sep);
3281             }
3282 
3283             // If lenient, add also the alternate, if different from the locale.
3284             if (isLenient() && data[0].compare(&alt_sep, 1) != 0) {
3285                 data[count++].setTo(alt_sep);
3286             }
3287 
3288             return matchString(text, start, UCAL_FIELD_COUNT /* => nothing to set */, data, count, NULL, cal);
3289         }
3290 
3291     default:
3292         // Handle "generic" fields
3293         // this is now handled below, outside the switch block
3294         break;
3295     }
3296     // Handle "generic" fields:
3297     // switch default case now handled here (outside switch block) to allow
3298     // parsing of some string fields as digits for lenient case
3299 
3300     int32_t parseStart = pos.getIndex();
3301     const UnicodeString* src;
3302     if (obeyCount) {
3303         if ((start+count) > text.length()) {
3304             return -start;
3305         }
3306         text.extractBetween(0, start + count, temp);
3307         src = &temp;
3308     } else {
3309         src = &text;
3310     }
3311     parseInt(*src, number, pos, allowNegative,currentNumberFormat);
3312     if (pos.getIndex() != parseStart) {
3313         int32_t value = number.getLong();
3314 
3315         // Don't need suffix processing here (as in number processing at the beginning of the function);
3316         // the new fields being handled as numeric values (month, weekdays, quarters) should not have suffixes.
3317 
3318         if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status)) {
3319             // Check the range of the value
3320             int32_t bias = gFieldRangeBias[patternCharIndex];
3321             if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
3322                 return -start;
3323             }
3324         }
3325 
3326         // For the following, need to repeat some of the "if (gotNumber)" code above:
3327         // UDAT_[STANDALONE_]MONTH_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_STANDALONE_DAY_FIELD,
3328         // UDAT_[STANDALONE_]QUARTER_FIELD
3329         switch (patternCharIndex) {
3330         case UDAT_MONTH_FIELD:
3331             // See notes under UDAT_MONTH_FIELD case above
3332             if (!strcmp(cal.getType(),"hebrew")) {
3333                 HebrewCalendar *hc = (HebrewCalendar*)&cal;
3334                 if (cal.isSet(UCAL_YEAR)) {
3335                    UErrorCode status = U_ZERO_ERROR;
3336                    if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
3337                        cal.set(UCAL_MONTH, value);
3338                    } else {
3339                        cal.set(UCAL_MONTH, value - 1);
3340                    }
3341                 } else {
3342                     saveHebrewMonth = value;
3343                 }
3344             } else {
3345                 cal.set(UCAL_MONTH, value - 1);
3346             }
3347             break;
3348         case UDAT_STANDALONE_MONTH_FIELD:
3349             cal.set(UCAL_MONTH, value - 1);
3350             break;
3351         case UDAT_DOW_LOCAL_FIELD:
3352         case UDAT_STANDALONE_DAY_FIELD:
3353             cal.set(UCAL_DOW_LOCAL, value);
3354             break;
3355         case UDAT_QUARTER_FIELD:
3356         case UDAT_STANDALONE_QUARTER_FIELD:
3357              cal.set(UCAL_MONTH, (value - 1) * 3);
3358              break;
3359         case UDAT_RELATED_YEAR_FIELD:
3360             cal.setRelatedYear(value);
3361             break;
3362         default:
3363             cal.set(field, value);
3364             break;
3365         }
3366         return pos.getIndex();
3367     }
3368     return -start;
3369 }
3370 
3371 /**
3372  * Parse an integer using fNumberFormat.  This method is semantically
3373  * const, but actually may modify fNumberFormat.
3374  */
parseInt(const UnicodeString & text,Formattable & number,ParsePosition & pos,UBool allowNegative,NumberFormat * fmt) const3375 void SimpleDateFormat::parseInt(const UnicodeString& text,
3376                                 Formattable& number,
3377                                 ParsePosition& pos,
3378                                 UBool allowNegative,
3379                                 NumberFormat *fmt) const {
3380     parseInt(text, number, -1, pos, allowNegative,fmt);
3381 }
3382 
3383 /**
3384  * Parse an integer using fNumberFormat up to maxDigits.
3385  */
parseInt(const UnicodeString & text,Formattable & number,int32_t maxDigits,ParsePosition & pos,UBool allowNegative,NumberFormat * fmt) const3386 void SimpleDateFormat::parseInt(const UnicodeString& text,
3387                                 Formattable& number,
3388                                 int32_t maxDigits,
3389                                 ParsePosition& pos,
3390                                 UBool allowNegative,
3391                                 NumberFormat *fmt) const {
3392     UnicodeString oldPrefix;
3393     DecimalFormat* df = NULL;
3394     if (!allowNegative && (df = dynamic_cast<DecimalFormat*>(fmt)) != NULL) {
3395         df->getNegativePrefix(oldPrefix);
3396         df->setNegativePrefix(UnicodeString(TRUE, SUPPRESS_NEGATIVE_PREFIX, -1));
3397     }
3398     int32_t oldPos = pos.getIndex();
3399     fmt->parse(text, number, pos);
3400     if (df != NULL) {
3401         df->setNegativePrefix(oldPrefix);
3402     }
3403 
3404     if (maxDigits > 0) {
3405         // adjust the result to fit into
3406         // the maxDigits and move the position back
3407         int32_t nDigits = pos.getIndex() - oldPos;
3408         if (nDigits > maxDigits) {
3409             int32_t val = number.getLong();
3410             nDigits -= maxDigits;
3411             while (nDigits > 0) {
3412                 val /= 10;
3413                 nDigits--;
3414             }
3415             pos.setIndex(oldPos + maxDigits);
3416             number.setLong(val);
3417         }
3418     }
3419 }
3420 
3421 //----------------------------------------------------------------------
3422 
translatePattern(const UnicodeString & originalPattern,UnicodeString & translatedPattern,const UnicodeString & from,const UnicodeString & to,UErrorCode & status)3423 void SimpleDateFormat::translatePattern(const UnicodeString& originalPattern,
3424                                         UnicodeString& translatedPattern,
3425                                         const UnicodeString& from,
3426                                         const UnicodeString& to,
3427                                         UErrorCode& status)
3428 {
3429     // run through the pattern and convert any pattern symbols from the version
3430     // in "from" to the corresponding character ion "to".  This code takes
3431     // quoted strings into account (it doesn't try to translate them), and it signals
3432     // an error if a particular "pattern character" doesn't appear in "from".
3433     // Depending on the values of "from" and "to" this can convert from generic
3434     // to localized patterns or localized to generic.
3435     if (U_FAILURE(status)) {
3436         return;
3437     }
3438 
3439     translatedPattern.remove();
3440     UBool inQuote = FALSE;
3441     for (int32_t i = 0; i < originalPattern.length(); ++i) {
3442         UChar c = originalPattern[i];
3443         if (inQuote) {
3444             if (c == QUOTE) {
3445                 inQuote = FALSE;
3446             }
3447         } else {
3448             if (c == QUOTE) {
3449                 inQuote = TRUE;
3450             } else if (isSyntaxChar(c)) {
3451                 int32_t ci = from.indexOf(c);
3452                 if (ci == -1) {
3453                     status = U_INVALID_FORMAT_ERROR;
3454                     return;
3455                 }
3456                 c = to[ci];
3457             }
3458         }
3459         translatedPattern += c;
3460     }
3461     if (inQuote) {
3462         status = U_INVALID_FORMAT_ERROR;
3463         return;
3464     }
3465 }
3466 
3467 //----------------------------------------------------------------------
3468 
3469 UnicodeString&
toPattern(UnicodeString & result) const3470 SimpleDateFormat::toPattern(UnicodeString& result) const
3471 {
3472     result = fPattern;
3473     return result;
3474 }
3475 
3476 //----------------------------------------------------------------------
3477 
3478 UnicodeString&
toLocalizedPattern(UnicodeString & result,UErrorCode & status) const3479 SimpleDateFormat::toLocalizedPattern(UnicodeString& result,
3480                                      UErrorCode& status) const
3481 {
3482     translatePattern(fPattern, result,
3483                      UnicodeString(DateFormatSymbols::getPatternUChars()),
3484                      fSymbols->fLocalPatternChars, status);
3485     return result;
3486 }
3487 
3488 //----------------------------------------------------------------------
3489 
3490 void
applyPattern(const UnicodeString & pattern)3491 SimpleDateFormat::applyPattern(const UnicodeString& pattern)
3492 {
3493     fPattern = pattern;
3494 }
3495 
3496 //----------------------------------------------------------------------
3497 
3498 void
applyLocalizedPattern(const UnicodeString & pattern,UErrorCode & status)3499 SimpleDateFormat::applyLocalizedPattern(const UnicodeString& pattern,
3500                                         UErrorCode &status)
3501 {
3502     translatePattern(pattern, fPattern,
3503                      fSymbols->fLocalPatternChars,
3504                      UnicodeString(DateFormatSymbols::getPatternUChars()), status);
3505 }
3506 
3507 //----------------------------------------------------------------------
3508 
3509 const DateFormatSymbols*
getDateFormatSymbols() const3510 SimpleDateFormat::getDateFormatSymbols() const
3511 {
3512     return fSymbols;
3513 }
3514 
3515 //----------------------------------------------------------------------
3516 
3517 void
adoptDateFormatSymbols(DateFormatSymbols * newFormatSymbols)3518 SimpleDateFormat::adoptDateFormatSymbols(DateFormatSymbols* newFormatSymbols)
3519 {
3520     delete fSymbols;
3521     fSymbols = newFormatSymbols;
3522 }
3523 
3524 //----------------------------------------------------------------------
3525 void
setDateFormatSymbols(const DateFormatSymbols & newFormatSymbols)3526 SimpleDateFormat::setDateFormatSymbols(const DateFormatSymbols& newFormatSymbols)
3527 {
3528     delete fSymbols;
3529     fSymbols = new DateFormatSymbols(newFormatSymbols);
3530 }
3531 
3532 //----------------------------------------------------------------------
3533 const TimeZoneFormat*
getTimeZoneFormat(void) const3534 SimpleDateFormat::getTimeZoneFormat(void) const {
3535     return (const TimeZoneFormat*)tzFormat();
3536 }
3537 
3538 //----------------------------------------------------------------------
3539 void
adoptTimeZoneFormat(TimeZoneFormat * timeZoneFormatToAdopt)3540 SimpleDateFormat::adoptTimeZoneFormat(TimeZoneFormat* timeZoneFormatToAdopt)
3541 {
3542     delete fTimeZoneFormat;
3543     fTimeZoneFormat = timeZoneFormatToAdopt;
3544 }
3545 
3546 //----------------------------------------------------------------------
3547 void
setTimeZoneFormat(const TimeZoneFormat & newTimeZoneFormat)3548 SimpleDateFormat::setTimeZoneFormat(const TimeZoneFormat& newTimeZoneFormat)
3549 {
3550     delete fTimeZoneFormat;
3551     fTimeZoneFormat = new TimeZoneFormat(newTimeZoneFormat);
3552 }
3553 
3554 //----------------------------------------------------------------------
3555 
3556 
adoptCalendar(Calendar * calendarToAdopt)3557 void SimpleDateFormat::adoptCalendar(Calendar* calendarToAdopt)
3558 {
3559   UErrorCode status = U_ZERO_ERROR;
3560   Locale calLocale(fLocale);
3561   calLocale.setKeywordValue("calendar", calendarToAdopt->getType(), status);
3562   DateFormatSymbols *newSymbols =
3563           DateFormatSymbols::createForLocale(calLocale, status);
3564   if (U_FAILURE(status)) {
3565       return;
3566   }
3567   DateFormat::adoptCalendar(calendarToAdopt);
3568   delete fSymbols;
3569   fSymbols = newSymbols;
3570   initializeDefaultCentury();  // we need a new century (possibly)
3571 }
3572 
3573 
3574 //----------------------------------------------------------------------
3575 
3576 
3577 // override the DateFormat implementation in order to
3578 // lazily initialize fCapitalizationBrkIter
3579 void
setContext(UDisplayContext value,UErrorCode & status)3580 SimpleDateFormat::setContext(UDisplayContext value, UErrorCode& status)
3581 {
3582     DateFormat::setContext(value, status);
3583 #if !UCONFIG_NO_BREAK_ITERATION
3584     if (U_SUCCESS(status)) {
3585         if ( fCapitalizationBrkIter == NULL && (value==UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE ||
3586                 value==UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU || value==UDISPCTX_CAPITALIZATION_FOR_STANDALONE) ) {
3587             UErrorCode status = U_ZERO_ERROR;
3588             fCapitalizationBrkIter = BreakIterator::createSentenceInstance(fLocale, status);
3589             if (U_FAILURE(status)) {
3590                 delete fCapitalizationBrkIter;
3591                 fCapitalizationBrkIter = NULL;
3592             }
3593         }
3594     }
3595 #endif
3596 }
3597 
3598 
3599 //----------------------------------------------------------------------
3600 
3601 
3602 UBool
isFieldUnitIgnored(UCalendarDateFields field) const3603 SimpleDateFormat::isFieldUnitIgnored(UCalendarDateFields field) const {
3604     return isFieldUnitIgnored(fPattern, field);
3605 }
3606 
3607 
3608 UBool
isFieldUnitIgnored(const UnicodeString & pattern,UCalendarDateFields field)3609 SimpleDateFormat::isFieldUnitIgnored(const UnicodeString& pattern,
3610                                      UCalendarDateFields field) {
3611     int32_t fieldLevel = fgCalendarFieldToLevel[field];
3612     int32_t level;
3613     UChar ch;
3614     UBool inQuote = FALSE;
3615     UChar prevCh = 0;
3616     int32_t count = 0;
3617 
3618     for (int32_t i = 0; i < pattern.length(); ++i) {
3619         ch = pattern[i];
3620         if (ch != prevCh && count > 0) {
3621             level = getLevelFromChar(prevCh);
3622             // the larger the level, the smaller the field unit.
3623             if (fieldLevel <= level) {
3624                 return FALSE;
3625             }
3626             count = 0;
3627         }
3628         if (ch == QUOTE) {
3629             if ((i+1) < pattern.length() && pattern[i+1] == QUOTE) {
3630                 ++i;
3631             } else {
3632                 inQuote = ! inQuote;
3633             }
3634         }
3635         else if (!inQuote && isSyntaxChar(ch)) {
3636             prevCh = ch;
3637             ++count;
3638         }
3639     }
3640     if (count > 0) {
3641         // last item
3642         level = getLevelFromChar(prevCh);
3643         if (fieldLevel <= level) {
3644             return FALSE;
3645         }
3646     }
3647     return TRUE;
3648 }
3649 
3650 //----------------------------------------------------------------------
3651 
3652 const Locale&
getSmpFmtLocale(void) const3653 SimpleDateFormat::getSmpFmtLocale(void) const {
3654     return fLocale;
3655 }
3656 
3657 //----------------------------------------------------------------------
3658 
3659 int32_t
checkIntSuffix(const UnicodeString & text,int32_t start,int32_t patLoc,UBool isNegative) const3660 SimpleDateFormat::checkIntSuffix(const UnicodeString& text, int32_t start,
3661                                  int32_t patLoc, UBool isNegative) const {
3662     // local variables
3663     UnicodeString suf;
3664     int32_t patternMatch;
3665     int32_t textPreMatch;
3666     int32_t textPostMatch;
3667 
3668     // check that we are still in range
3669     if ( (start > text.length()) ||
3670          (start < 0) ||
3671          (patLoc < 0) ||
3672          (patLoc > fPattern.length())) {
3673         // out of range, don't advance location in text
3674         return start;
3675     }
3676 
3677     // get the suffix
3678     DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(fNumberFormat);
3679     if (decfmt != NULL) {
3680         if (isNegative) {
3681             suf = decfmt->getNegativeSuffix(suf);
3682         }
3683         else {
3684             suf = decfmt->getPositiveSuffix(suf);
3685         }
3686     }
3687 
3688     // check for suffix
3689     if (suf.length() <= 0) {
3690         return start;
3691     }
3692 
3693     // check suffix will be encountered in the pattern
3694     patternMatch = compareSimpleAffix(suf,fPattern,patLoc);
3695 
3696     // check if a suffix will be encountered in the text
3697     textPreMatch = compareSimpleAffix(suf,text,start);
3698 
3699     // check if a suffix was encountered in the text
3700     textPostMatch = compareSimpleAffix(suf,text,start-suf.length());
3701 
3702     // check for suffix match
3703     if ((textPreMatch >= 0) && (patternMatch >= 0) && (textPreMatch == patternMatch)) {
3704         return start;
3705     }
3706     else if ((textPostMatch >= 0) && (patternMatch >= 0) && (textPostMatch == patternMatch)) {
3707         return  start - suf.length();
3708     }
3709 
3710     // should not get here
3711     return start;
3712 }
3713 
3714 //----------------------------------------------------------------------
3715 
3716 int32_t
compareSimpleAffix(const UnicodeString & affix,const UnicodeString & input,int32_t pos) const3717 SimpleDateFormat::compareSimpleAffix(const UnicodeString& affix,
3718                    const UnicodeString& input,
3719                    int32_t pos) const {
3720     int32_t start = pos;
3721     for (int32_t i=0; i<affix.length(); ) {
3722         UChar32 c = affix.char32At(i);
3723         int32_t len = U16_LENGTH(c);
3724         if (PatternProps::isWhiteSpace(c)) {
3725             // We may have a pattern like: \u200F \u0020
3726             //        and input text like: \u200F \u0020
3727             // Note that U+200F and U+0020 are Pattern_White_Space but only
3728             // U+0020 is UWhiteSpace.  So we have to first do a direct
3729             // match of the run of Pattern_White_Space in the pattern,
3730             // then match any extra characters.
3731             UBool literalMatch = FALSE;
3732             while (pos < input.length() &&
3733                    input.char32At(pos) == c) {
3734                 literalMatch = TRUE;
3735                 i += len;
3736                 pos += len;
3737                 if (i == affix.length()) {
3738                     break;
3739                 }
3740                 c = affix.char32At(i);
3741                 len = U16_LENGTH(c);
3742                 if (!PatternProps::isWhiteSpace(c)) {
3743                     break;
3744                 }
3745             }
3746 
3747             // Advance over run in pattern
3748             i = skipPatternWhiteSpace(affix, i);
3749 
3750             // Advance over run in input text
3751             // Must see at least one white space char in input,
3752             // unless we've already matched some characters literally.
3753             int32_t s = pos;
3754             pos = skipUWhiteSpace(input, pos);
3755             if (pos == s && !literalMatch) {
3756                 return -1;
3757             }
3758 
3759             // If we skip UWhiteSpace in the input text, we need to skip it in the pattern.
3760             // Otherwise, the previous lines may have skipped over text (such as U+00A0) that
3761             // is also in the affix.
3762             i = skipUWhiteSpace(affix, i);
3763         } else {
3764             if (pos < input.length() &&
3765                 input.char32At(pos) == c) {
3766                 i += len;
3767                 pos += len;
3768             } else {
3769                 return -1;
3770             }
3771         }
3772     }
3773     return pos - start;
3774 }
3775 
3776 //----------------------------------------------------------------------
3777 
3778 int32_t
skipPatternWhiteSpace(const UnicodeString & text,int32_t pos) const3779 SimpleDateFormat::skipPatternWhiteSpace(const UnicodeString& text, int32_t pos) const {
3780     const UChar* s = text.getBuffer();
3781     return (int32_t)(PatternProps::skipWhiteSpace(s + pos, text.length() - pos) - s);
3782 }
3783 
3784 //----------------------------------------------------------------------
3785 
3786 int32_t
skipUWhiteSpace(const UnicodeString & text,int32_t pos) const3787 SimpleDateFormat::skipUWhiteSpace(const UnicodeString& text, int32_t pos) const {
3788     while (pos < text.length()) {
3789         UChar32 c = text.char32At(pos);
3790         if (!u_isUWhiteSpace(c)) {
3791             break;
3792         }
3793         pos += U16_LENGTH(c);
3794     }
3795     return pos;
3796 }
3797 
3798 //----------------------------------------------------------------------
3799 
3800 // Lazy TimeZoneFormat instantiation, semantically const.
3801 TimeZoneFormat *
tzFormat() const3802 SimpleDateFormat::tzFormat() const {
3803     if (fTimeZoneFormat == NULL) {
3804         umtx_lock(&LOCK);
3805         {
3806             if (fTimeZoneFormat == NULL) {
3807                 UErrorCode status = U_ZERO_ERROR;
3808                 TimeZoneFormat *tzfmt = TimeZoneFormat::createInstance(fLocale, status);
3809                 if (U_FAILURE(status)) {
3810                     return NULL;
3811                 }
3812 
3813                 const_cast<SimpleDateFormat *>(this)->fTimeZoneFormat = tzfmt;
3814             }
3815         }
3816         umtx_unlock(&LOCK);
3817     }
3818     return fTimeZoneFormat;
3819 }
3820 
3821 U_NAMESPACE_END
3822 
3823 #endif /* #if !UCONFIG_NO_FORMATTING */
3824 
3825 //eof
3826