]> git.saurik.com Git - apple/icu.git/blob - icuSources/i18n/smpdtfmt.cpp
projects / apple / icu.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
ICU-551.51.4.tar.gz
[apple/icu.git] / icuSources / i18n / smpdtfmt.cpp
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 -1, // ':' - UDAT_TIME_SEPARATOR_FIELD
213 };
214 // A slightly looser range check for lenient parsing
215 static const int32_t gFieldRangeBiasLenient[] = {
216 -1, // 'G' - UDAT_ERA_FIELD
217 -1, // 'y' - UDAT_YEAR_FIELD
218 8, // 'M' - UDAT_MONTH_FIELD (allow calendar max + 7, e.g. 19 for grego 1-based month)
219 18, // 'd' - UDAT_DATE_FIELD (allow calendar max + 18, e.g. 49 for grego; tests require at least 40 for grego)
220 -1, // 'k' - UDAT_HOUR_OF_DAY1_FIELD
221 -1, // 'H' - UDAT_HOUR_OF_DAY0_FIELD
222 40, // 'm' - UDAT_MINUTE_FIELD (allow calendar max + 40, e.g. 99)
223 40, // 's' - UDAT_SECOND_FIELD (allow calendar max + 40, e.g. 99)
224 -1, // 'S' - UDAT_FRACTIONAL_SECOND_FIELD (0-999?)
225 -1, // 'E' - UDAT_DAY_OF_WEEK_FIELD (1-7?)
226 -1, // 'D' - UDAT_DAY_OF_YEAR_FIELD (1 - 366?)
227 -1, // 'F' - UDAT_DAY_OF_WEEK_IN_MONTH_FIELD (1-5?)
228 -1, // 'w' - UDAT_WEEK_OF_YEAR_FIELD (1-52?)
229 -1, // 'W' - UDAT_WEEK_OF_MONTH_FIELD (1-5?)
230 -1, // 'a' - UDAT_AM_PM_FIELD
231 -1, // 'h' - UDAT_HOUR1_FIELD
232 -1, // 'K' - UDAT_HOUR0_FIELD
233 -1, // 'z' - UDAT_TIMEZONE_FIELD
234 -1, // 'Y' - UDAT_YEAR_WOY_FIELD
235 -1, // 'e' - UDAT_DOW_LOCAL_FIELD
236 -1, // 'u' - UDAT_EXTENDED_YEAR_FIELD
237 -1, // 'g' - UDAT_JULIAN_DAY_FIELD
238 -1, // 'A' - UDAT_MILLISECONDS_IN_DAY_FIELD
239 -1, // 'Z' - UDAT_TIMEZONE_RFC_FIELD
240 -1, // 'v' - UDAT_TIMEZONE_GENERIC_FIELD
241 18, // 'c' - UDAT_STANDALONE_DAY_FIELD (allow calendar max + 18, e.g. 49 for grego)
242 8, // 'L' - UDAT_STANDALONE_MONTH_FIELD (allow calendar max + 7, e.g. 19 for grego 1-based month)
243 -1, // 'Q' - UDAT_QUARTER_FIELD (1-4?)
244 -1, // 'q' - UDAT_STANDALONE_QUARTER_FIELD
245 -1, // 'V' - UDAT_TIMEZONE_SPECIAL_FIELD
246 -1, // 'U' - UDAT_YEAR_NAME_FIELD
247 -1, // 'O' - UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD
248 -1, // 'X' - UDAT_TIMEZONE_ISO_FIELD
249 -1, // 'x' - UDAT_TIMEZONE_ISO_LOCAL_FIELD
250 -1, // 'r' - UDAT_RELATED_YEAR_FIELD
251 -1, // ':' - UDAT_TIME_SEPARATOR_FIELD
252 };
253
254 // When calendar uses hebr numbering (i.e. he@calendar=hebrew),
255 // offset the years within the current millenium down to 1-999
256 static const int32_t HEBREW_CAL_CUR_MILLENIUM_START_YEAR = 5000;
257 static const int32_t HEBREW_CAL_CUR_MILLENIUM_END_YEAR = 6000;
258
259 static UMutex LOCK = U_MUTEX_INITIALIZER;
260
261 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)
262
263 SimpleDateFormat::NSOverride::~NSOverride() {
264 if (snf != NULL) {
265 snf->removeRef();
266 }
267 }
268
269
270 void SimpleDateFormat::NSOverride::free() {
271 NSOverride *cur = this;
272 while (cur) {
273 NSOverride *next = cur->next;
274 delete cur;
275 cur = next;
276 }
277 }
278
279 // no matter what the locale's default number format looked like, we want
280 // to modify it so that it doesn't use thousands separators, doesn't always
281 // show the decimal point, and recognizes integers only when parsing
282 static void fixNumberFormatForDates(NumberFormat &nf) {
283 nf.setGroupingUsed(FALSE);
284 DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(&nf);
285 if (decfmt != NULL) {
286 decfmt->setDecimalSeparatorAlwaysShown(FALSE);
287 }
288 nf.setParseIntegerOnly(TRUE);
289 nf.setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
290 }
291
292 static const SharedNumberFormat *createSharedNumberFormat(
293 NumberFormat *nfToAdopt) {
294 fixNumberFormatForDates(*nfToAdopt);
295 const SharedNumberFormat *result = new SharedNumberFormat(nfToAdopt);
296 if (result == NULL) {
297 delete nfToAdopt;
298 }
299 return result;
300 }
301
302 static const SharedNumberFormat *createSharedNumberFormat(
303 const Locale &loc, UErrorCode &status) {
304 NumberFormat *nf = NumberFormat::createInstance(loc, status);
305 if (U_FAILURE(status)) {
306 return NULL;
307 }
308 const SharedNumberFormat *result = createSharedNumberFormat(nf);
309 if (result == NULL) {
310 status = U_MEMORY_ALLOCATION_ERROR;
311 }
312 return result;
313 }
314
315 static const SharedNumberFormat **allocSharedNumberFormatters() {
316 const SharedNumberFormat **result = (const SharedNumberFormat**)
317 uprv_malloc(UDAT_FIELD_COUNT * sizeof(const SharedNumberFormat*));
318 if (result == NULL) {
319 return NULL;
320 }
321 for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
322 result[i] = NULL;
323 }
324 return result;
325 }
326
327 static void freeSharedNumberFormatters(const SharedNumberFormat ** list) {
328 for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
329 SharedObject::clearPtr(list[i]);
330 }
331 uprv_free(list);
332 }
333
334 const NumberFormat *SimpleDateFormat::getNumberFormatByIndex(
335 UDateFormatField index) const {
336 if (fSharedNumberFormatters == NULL ||
337 fSharedNumberFormatters[index] == NULL) {
338 return fNumberFormat;
339 }
340 return &(**fSharedNumberFormatters[index]);
341 }
342
343 class SimpleDateFormatMutableNFNode {
344 public:
345 const NumberFormat *key;
346 NumberFormat *value;
347 SimpleDateFormatMutableNFNode()
348 : key(NULL), value(NULL) { }
349 ~SimpleDateFormatMutableNFNode() {
350 delete value;
351 }
352 private:
353 SimpleDateFormatMutableNFNode(const SimpleDateFormatMutableNFNode &);
354 SimpleDateFormatMutableNFNode &operator=(const SimpleDateFormatMutableNFNode &);
355 };
356
357 // Single threaded cache of non const NumberFormats. Designed to be stack
358 // allocated and used for a single format call.
359 class SimpleDateFormatMutableNFs : public UMemory {
360 public:
361 SimpleDateFormatMutableNFs() {
362 }
363
364 // Returns a non-const clone of nf which can be safely modified.
365 // Subsequent calls with same nf will return the same non-const clone.
366 // This object maintains ownership of all returned non-const
367 // NumberFormat objects. On memory allocation error returns NULL.
368 // Caller must check for NULL return value.
369 NumberFormat *get(const NumberFormat *nf) {
370 if (nf == NULL) {
371 return NULL;
372 }
373 int32_t idx = 0;
374 while (nodes[idx].value) {
375 if (nf == nodes[idx].key) {
376 return nodes[idx].value;
377 }
378 ++idx;
379 }
380 U_ASSERT(idx < UDAT_FIELD_COUNT);
381 nodes[idx].key = nf;
382 nodes[idx].value = (NumberFormat *) nf->clone();
383 return nodes[idx].value;
384 }
385 private:
386 // +1 extra for sentinel. If each field had its own NumberFormat, this
387 // cache would have to allocate UDAT_FIELD_COUNT mutable versions worst
388 // case.
389 SimpleDateFormatMutableNFNode nodes[UDAT_FIELD_COUNT + 1];
390 SimpleDateFormatMutableNFs(const SimpleDateFormatMutableNFs &);
391 SimpleDateFormatMutableNFs &operator=(const SimpleDateFormatMutableNFs &);
392 };
393
394 //----------------------------------------------------------------------
395
396 static void updateTimeSepFromPattern(
397 const UnicodeString& pattern,
398 DateFormatSymbols* symbols ) {
399 UnicodeString hourMinChars("hHKkm", -1, US_INV); // pattern chars for hours, minutes
400 UnicodeString colon(":", -1, US_INV);
401 UBool inQuoted = FALSE;
402 UBool lastPatCharWasHourMin = FALSE;
403 int32_t patPos, patLen = pattern.length();
404 for (patPos = 0; patPos < patLen; patPos++) {
405 UChar patChr = pattern.charAt(patPos);
406 if (patChr == 0x27 /* ASCII-range single quote */) {
407 inQuoted = !inQuoted;
408 } else if (!inQuoted) {
409 if (patChr == 0x3A /*colon*/ && lastPatCharWasHourMin) {
410 symbols->setTimeSeparatorString(colon);
411 break;
412 }
413 if ((patChr >= 0x41 && patChr <= 0x5A) || (patChr >= 0x61 && patChr <= 0x7A)) {
414 lastPatCharWasHourMin = (hourMinChars.indexOf(patChr) >= 0);
415 }
416 }
417 }
418 }
419
420 //----------------------------------------------------------------------
421
422 SimpleDateFormat::~SimpleDateFormat()
423 {
424 delete fSymbols;
425 if (fSharedNumberFormatters) {
426 freeSharedNumberFormatters(fSharedNumberFormatters);
427 }
428 if (fTimeZoneFormat) {
429 delete fTimeZoneFormat;
430 }
431
432 #if !UCONFIG_NO_BREAK_ITERATION
433 delete fCapitalizationBrkIter;
434 #endif
435 }
436
437 //----------------------------------------------------------------------
438
439 SimpleDateFormat::SimpleDateFormat(UErrorCode& status)
440 : fLocale(Locale::getDefault()),
441 fSymbols(NULL),
442 fTimeZoneFormat(NULL),
443 fSharedNumberFormatters(NULL),
444 fCapitalizationBrkIter(NULL)
445 {
446 initializeBooleanAttributes();
447 construct(kShort, (EStyle) (kShort + kDateOffset), fLocale, status);
448 initializeDefaultCentury();
449 }
450
451 //----------------------------------------------------------------------
452
453 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
454 UErrorCode &status)
455 : fPattern(pattern),
456 fLocale(Locale::getDefault()),
457 fSymbols(NULL),
458 fTimeZoneFormat(NULL),
459 fSharedNumberFormatters(NULL),
460 fCapitalizationBrkIter(NULL)
461 {
462 fDateOverride.setToBogus();
463 fTimeOverride.setToBogus();
464 initializeBooleanAttributes();
465 initializeCalendar(NULL,fLocale,status);
466 fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
467 initialize(fLocale, status);
468 initializeDefaultCentury();
469 updateTimeSepFromPattern(fPattern, fSymbols);
470 }
471
472 //----------------------------------------------------------------------
473
474 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
475 const UnicodeString& override,
476 UErrorCode &status)
477 : fPattern(pattern),
478 fLocale(Locale::getDefault()),
479 fSymbols(NULL),
480 fTimeZoneFormat(NULL),
481 fSharedNumberFormatters(NULL),
482 fCapitalizationBrkIter(NULL)
483 {
484 fDateOverride.setTo(override);
485 fTimeOverride.setToBogus();
486 initializeBooleanAttributes();
487 initializeCalendar(NULL,fLocale,status);
488 fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
489 initialize(fLocale, status);
490 initializeDefaultCentury();
491
492 processOverrideString(fLocale,override,kOvrStrBoth,status);
493
494 }
495
496 //----------------------------------------------------------------------
497
498 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
499 const Locale& locale,
500 UErrorCode& status)
501 : fPattern(pattern),
502 fLocale(locale),
503 fTimeZoneFormat(NULL),
504 fSharedNumberFormatters(NULL),
505 fCapitalizationBrkIter(NULL)
506 {
507
508 fDateOverride.setToBogus();
509 fTimeOverride.setToBogus();
510 initializeBooleanAttributes();
511
512 initializeCalendar(NULL,fLocale,status);
513 fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
514 initialize(fLocale, status);
515 initializeDefaultCentury();
516 updateTimeSepFromPattern(fPattern, fSymbols);
517 }
518
519 //----------------------------------------------------------------------
520
521 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
522 const UnicodeString& override,
523 const Locale& locale,
524 UErrorCode& status)
525 : fPattern(pattern),
526 fLocale(locale),
527 fTimeZoneFormat(NULL),
528 fSharedNumberFormatters(NULL),
529 fCapitalizationBrkIter(NULL)
530 {
531
532 fDateOverride.setTo(override);
533 fTimeOverride.setToBogus();
534 initializeBooleanAttributes();
535
536 initializeCalendar(NULL,fLocale,status);
537 fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
538 initialize(fLocale, status);
539 initializeDefaultCentury();
540
541 processOverrideString(locale,override,kOvrStrBoth,status);
542
543 }
544
545 //----------------------------------------------------------------------
546
547 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
548 DateFormatSymbols* symbolsToAdopt,
549 UErrorCode& status)
550 : fPattern(pattern),
551 fLocale(Locale::getDefault()),
552 fSymbols(symbolsToAdopt),
553 fTimeZoneFormat(NULL),
554 fSharedNumberFormatters(NULL),
555 fCapitalizationBrkIter(NULL)
556 {
557
558 fDateOverride.setToBogus();
559 fTimeOverride.setToBogus();
560 initializeBooleanAttributes();
561
562 initializeCalendar(NULL,fLocale,status);
563 initialize(fLocale, status);
564 initializeDefaultCentury();
565 }
566
567 //----------------------------------------------------------------------
568
569 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
570 const DateFormatSymbols& symbols,
571 UErrorCode& status)
572 : fPattern(pattern),
573 fLocale(Locale::getDefault()),
574 fSymbols(new DateFormatSymbols(symbols)),
575 fTimeZoneFormat(NULL),
576 fSharedNumberFormatters(NULL),
577 fCapitalizationBrkIter(NULL)
578 {
579
580 fDateOverride.setToBogus();
581 fTimeOverride.setToBogus();
582 initializeBooleanAttributes();
583
584 initializeCalendar(NULL, fLocale, status);
585 initialize(fLocale, status);
586 initializeDefaultCentury();
587 }
588
589 //----------------------------------------------------------------------
590
591 // Not for public consumption; used by DateFormat
592 SimpleDateFormat::SimpleDateFormat(EStyle timeStyle,
593 EStyle dateStyle,
594 const Locale& locale,
595 UErrorCode& status)
596 : fLocale(locale),
597 fSymbols(NULL),
598 fTimeZoneFormat(NULL),
599 fSharedNumberFormatters(NULL),
600 fCapitalizationBrkIter(NULL)
601 {
602 initializeBooleanAttributes();
603 construct(timeStyle, dateStyle, fLocale, status);
604 if(U_SUCCESS(status)) {
605 initializeDefaultCentury();
606 }
607 }
608
609 //----------------------------------------------------------------------
610
611 /**
612 * Not for public consumption; used by DateFormat. This constructor
613 * never fails. If the resource data is not available, it uses the
614 * the last resort symbols.
615 */
616 SimpleDateFormat::SimpleDateFormat(const Locale& locale,
617 UErrorCode& status)
618 : fPattern(gDefaultPattern),
619 fLocale(locale),
620 fSymbols(NULL),
621 fTimeZoneFormat(NULL),
622 fSharedNumberFormatters(NULL),
623 fCapitalizationBrkIter(NULL)
624 {
625 if (U_FAILURE(status)) return;
626 initializeBooleanAttributes();
627 initializeCalendar(NULL, fLocale, status);
628 fSymbols = DateFormatSymbols::createForLocale(fLocale, status);
629 if (U_FAILURE(status))
630 {
631 status = U_ZERO_ERROR;
632 delete fSymbols;
633 // This constructor doesn't fail; it uses last resort data
634 fSymbols = new DateFormatSymbols(status);
635 /* test for NULL */
636 if (fSymbols == 0) {
637 status = U_MEMORY_ALLOCATION_ERROR;
638 return;
639 }
640 }
641
642 fDateOverride.setToBogus();
643 fTimeOverride.setToBogus();
644
645 initialize(fLocale, status);
646 if(U_SUCCESS(status)) {
647 initializeDefaultCentury();
648 }
649 }
650
651 //----------------------------------------------------------------------
652
653 SimpleDateFormat::SimpleDateFormat(const SimpleDateFormat& other)
654 : DateFormat(other),
655 fLocale(other.fLocale),
656 fSymbols(NULL),
657 fTimeZoneFormat(NULL),
658 fSharedNumberFormatters(NULL),
659 fCapitalizationBrkIter(NULL)
660 {
661 initializeBooleanAttributes();
662 *this = other;
663 }
664
665 //----------------------------------------------------------------------
666
667 SimpleDateFormat& SimpleDateFormat::operator=(const SimpleDateFormat& other)
668 {
669 if (this == &other) {
670 return *this;
671 }
672 DateFormat::operator=(other);
673 fDateOverride = other.fDateOverride;
674 fTimeOverride = other.fTimeOverride;
675
676 delete fSymbols;
677 fSymbols = NULL;
678
679 if (other.fSymbols)
680 fSymbols = new DateFormatSymbols(*other.fSymbols);
681
682 fDefaultCenturyStart = other.fDefaultCenturyStart;
683 fDefaultCenturyStartYear = other.fDefaultCenturyStartYear;
684 fHaveDefaultCentury = other.fHaveDefaultCentury;
685
686 fPattern = other.fPattern;
687
688 fLocale = other.fLocale;
689 // TimeZoneFormat can now be set independently via setter.
690 // If it is NULL, it will be lazily initialized from locale
691 delete fTimeZoneFormat;
692 fTimeZoneFormat = NULL;
693 if (other.fTimeZoneFormat) {
694 fTimeZoneFormat = new TimeZoneFormat(*other.fTimeZoneFormat);
695 }
696
697 #if !UCONFIG_NO_BREAK_ITERATION
698 if (other.fCapitalizationBrkIter != NULL) {
699 fCapitalizationBrkIter = (other.fCapitalizationBrkIter)->clone();
700 }
701 #endif
702
703 if (fSharedNumberFormatters != NULL) {
704 freeSharedNumberFormatters(fSharedNumberFormatters);
705 fSharedNumberFormatters = NULL;
706 }
707 if (other.fSharedNumberFormatters != NULL) {
708 fSharedNumberFormatters = allocSharedNumberFormatters();
709 if (fSharedNumberFormatters) {
710 for (int32_t i = 0; i < UDAT_FIELD_COUNT; ++i) {
711 SharedObject::copyPtr(
712 other.fSharedNumberFormatters[i],
713 fSharedNumberFormatters[i]);
714 }
715 }
716 }
717
718 return *this;
719 }
720
721 //----------------------------------------------------------------------
722
723 Format*
724 SimpleDateFormat::clone() const
725 {
726 return new SimpleDateFormat(*this);
727 }
728
729 //----------------------------------------------------------------------
730
731 UBool
732 SimpleDateFormat::operator==(const Format& other) const
733 {
734 if (DateFormat::operator==(other)) {
735 // The DateFormat::operator== check for fCapitalizationContext equality above
736 // is sufficient to check equality of all derived context-related data.
737 // DateFormat::operator== guarantees following cast is safe
738 SimpleDateFormat* that = (SimpleDateFormat*)&other;
739 return (fPattern == that->fPattern &&
740 fSymbols != NULL && // Check for pathological object
741 that->fSymbols != NULL && // Check for pathological object
742 *fSymbols == *that->fSymbols &&
743 fHaveDefaultCentury == that->fHaveDefaultCentury &&
744 fDefaultCenturyStart == that->fDefaultCenturyStart &&
745 // Check fTimeZoneFormat, it can be set independently via setter
746 ((fTimeZoneFormat == NULL && that->fTimeZoneFormat == NULL) ||
747 (fTimeZoneFormat != NULL && that->fTimeZoneFormat != NULL && *fTimeZoneFormat == *that->fTimeZoneFormat)) &&
748 // Check override strings (these also indicate any relevant
749 // differences in fNumberFormatters, fOverrideList)
750 fDateOverride == that->fDateOverride &&
751 fTimeOverride == that->fTimeOverride);
752 }
753 return FALSE;
754 }
755
756 //----------------------------------------------------------------------
757
758 void SimpleDateFormat::construct(EStyle timeStyle,
759 EStyle dateStyle,
760 const Locale& locale,
761 UErrorCode& status)
762 {
763 // called by several constructors to load pattern data from the resources
764 if (U_FAILURE(status)) return;
765
766 // We will need the calendar to know what type of symbols to load.
767 initializeCalendar(NULL, locale, status);
768 if (U_FAILURE(status)) return;
769
770 CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
771 UResourceBundle *dateTimePatterns = calData.getByKey(gDateTimePatternsTag, status);
772 UResourceBundle *currentBundle;
773
774 if (U_FAILURE(status)) return;
775
776 if (ures_getSize(dateTimePatterns) <= kDateTime)
777 {
778 status = U_INVALID_FORMAT_ERROR;
779 return;
780 }
781
782 setLocaleIDs(ures_getLocaleByType(dateTimePatterns, ULOC_VALID_LOCALE, &status),
783 ures_getLocaleByType(dateTimePatterns, ULOC_ACTUAL_LOCALE, &status));
784
785 // create a symbols object from the locale
786 fSymbols = DateFormatSymbols::createForLocale(locale, status);
787 if (U_FAILURE(status)) return;
788 /* test for NULL */
789 if (fSymbols == 0) {
790 status = U_MEMORY_ALLOCATION_ERROR;
791 return;
792 }
793
794 const UChar *resStr,*ovrStr;
795 int32_t resStrLen,ovrStrLen = 0;
796 fDateOverride.setToBogus();
797 fTimeOverride.setToBogus();
798
799 // if the pattern should include both date and time information, use the date/time
800 // pattern string as a guide to tell use how to glue together the appropriate date
801 // and time pattern strings. The actual gluing-together is handled by a convenience
802 // method on MessageFormat.
803 if ((timeStyle != kNone) && (dateStyle != kNone))
804 {
805 Formattable timeDateArray[2];
806
807 // use Formattable::adoptString() so that we can use fastCopyFrom()
808 // instead of Formattable::setString()'s unaware, safe, deep string clone
809 // see Jitterbug 2296
810
811 currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
812 if (U_FAILURE(status)) {
813 status = U_INVALID_FORMAT_ERROR;
814 return;
815 }
816 switch (ures_getType(currentBundle)) {
817 case URES_STRING: {
818 resStr = ures_getString(currentBundle, &resStrLen, &status);
819 break;
820 }
821 case URES_ARRAY: {
822 resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
823 ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
824 fTimeOverride.setTo(TRUE, ovrStr, ovrStrLen);
825 break;
826 }
827 default: {
828 status = U_INVALID_FORMAT_ERROR;
829 ures_close(currentBundle);
830 return;
831 }
832 }
833 ures_close(currentBundle);
834
835 UnicodeString *tempus1 = new UnicodeString(TRUE, resStr, resStrLen);
836 // NULL pointer check
837 if (tempus1 == NULL) {
838 status = U_MEMORY_ALLOCATION_ERROR;
839 return;
840 }
841 timeDateArray[0].adoptString(tempus1);
842
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 ures_close(currentBundle);
866
867 UnicodeString *tempus2 = new UnicodeString(TRUE, resStr, resStrLen);
868 // Null pointer check
869 if (tempus2 == NULL) {
870 status = U_MEMORY_ALLOCATION_ERROR;
871 return;
872 }
873 timeDateArray[1].adoptString(tempus2);
874
875 int32_t glueIndex = kDateTime;
876 int32_t patternsSize = ures_getSize(dateTimePatterns);
877 if (patternsSize >= (kDateTimeOffset + kShort + 1)) {
878 // Get proper date time format
879 glueIndex = (int32_t)(kDateTimeOffset + (dateStyle - kDateOffset));
880 }
881
882 resStr = ures_getStringByIndex(dateTimePatterns, glueIndex, &resStrLen, &status);
883 MessageFormat::format(UnicodeString(TRUE, resStr, resStrLen), timeDateArray, 2, fPattern, status);
884 }
885 // if the pattern includes just time data or just date date, load the appropriate
886 // pattern string from the resources
887 // setTo() - see DateFormatSymbols::assignArray comments
888 else if (timeStyle != kNone) {
889 currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
890 if (U_FAILURE(status)) {
891 status = U_INVALID_FORMAT_ERROR;
892 return;
893 }
894 switch (ures_getType(currentBundle)) {
895 case URES_STRING: {
896 resStr = ures_getString(currentBundle, &resStrLen, &status);
897 break;
898 }
899 case URES_ARRAY: {
900 resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
901 ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
902 fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
903 break;
904 }
905 default: {
906 status = U_INVALID_FORMAT_ERROR;
907 ures_close(currentBundle);
908 return;
909 }
910 }
911 fPattern.setTo(TRUE, resStr, resStrLen);
912 ures_close(currentBundle);
913 }
914 else if (dateStyle != kNone) {
915 currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)dateStyle, NULL, &status);
916 if (U_FAILURE(status)) {
917 status = U_INVALID_FORMAT_ERROR;
918 return;
919 }
920 switch (ures_getType(currentBundle)) {
921 case URES_STRING: {
922 resStr = ures_getString(currentBundle, &resStrLen, &status);
923 break;
924 }
925 case URES_ARRAY: {
926 resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
927 ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
928 fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
929 break;
930 }
931 default: {
932 status = U_INVALID_FORMAT_ERROR;
933 ures_close(currentBundle);
934 return;
935 }
936 }
937 fPattern.setTo(TRUE, resStr, resStrLen);
938 ures_close(currentBundle);
939 }
940
941 // and if it includes _neither_, that's an error
942 else
943 status = U_INVALID_FORMAT_ERROR;
944
945 // finally, finish initializing by creating a Calendar and a NumberFormat
946 initialize(locale, status);
947 }
948
949 //----------------------------------------------------------------------
950
951 Calendar*
952 SimpleDateFormat::initializeCalendar(TimeZone* adoptZone, const Locale& locale, UErrorCode& status)
953 {
954 if(!U_FAILURE(status)) {
955 fCalendar = Calendar::createInstance(adoptZone?adoptZone:TimeZone::createDefault(), locale, status);
956 }
957 return fCalendar;
958 }
959
960 void
961 SimpleDateFormat::initialize(const Locale& locale,
962 UErrorCode& status)
963 {
964 if (U_FAILURE(status)) return;
965
966 // If the locale has @[....]numbers=hanidays we want to *delete* that (so it
967 // it is not used for every field) and then set fDateOverride to "d=hanidays"
968 // (as with std formats for zh@calendar=chinese) to use hanidays for d field.
969 static const UChar hanidaysOverride[] = {0x64,0x3D,0x68,0x61,0x6E,0x69,0x64,0x61,0x79,0x73,0}; // "d=hanidays"
970 char numbersValue[ULOC_KEYWORDS_CAPACITY];
971 UErrorCode numbersStatus = U_ZERO_ERROR;
972 Locale localeNoHanidays(locale);
973 int32_t numbersLen = localeNoHanidays.getKeywordValue("numbers", numbersValue, ULOC_KEYWORDS_CAPACITY, numbersStatus);
974 if ( U_SUCCESS(numbersStatus) && numbersLen > 0 ) {
975 if ( uprv_strcmp(numbersValue, "hanidays") == 0 ) {
976 localeNoHanidays.setKeywordValue("numbers", NULL, numbersStatus);
977 fDateOverride.setTo(hanidaysOverride,-1);
978 }
979 }
980
981 // We don't need to check that the row count is >= 1, since all 2d arrays have at
982 // least one row
983 fNumberFormat = NumberFormat::createInstance(localeNoHanidays, status);
984 if (fNumberFormat != NULL && U_SUCCESS(status))
985 {
986 fixNumberFormatForDates(*fNumberFormat);
987 //fNumberFormat->setLenient(TRUE); // Java uses a custom DateNumberFormat to format/parse
988
989 initNumberFormatters(locale,status);
990
991 }
992 else if (U_SUCCESS(status))
993 {
994 status = U_MISSING_RESOURCE_ERROR;
995 }
996 }
997
998 /* Initialize the fields we use to disambiguate ambiguous years. Separate
999 * so we can call it from readObject().
1000 */
1001 void SimpleDateFormat::initializeDefaultCentury()
1002 {
1003 if(fCalendar) {
1004 fHaveDefaultCentury = fCalendar->haveDefaultCentury();
1005 if(fHaveDefaultCentury) {
1006 fDefaultCenturyStart = fCalendar->defaultCenturyStart();
1007 fDefaultCenturyStartYear = fCalendar->defaultCenturyStartYear();
1008 } else {
1009 fDefaultCenturyStart = DBL_MIN;
1010 fDefaultCenturyStartYear = -1;
1011 }
1012 }
1013 }
1014
1015 /*
1016 * Initialize the boolean attributes. Separate so we can call it from all constructors.
1017 */
1018 void SimpleDateFormat::initializeBooleanAttributes()
1019 {
1020 UErrorCode status = U_ZERO_ERROR;
1021
1022 setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status);
1023 setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
1024 setBooleanAttribute(UDAT_PARSE_PARTIAL_MATCH, true, status);
1025 setBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, true, status);
1026 }
1027
1028 /* Define one-century window into which to disambiguate dates using
1029 * two-digit years. Make public in JDK 1.2.
1030 */
1031 void SimpleDateFormat::parseAmbiguousDatesAsAfter(UDate startDate, UErrorCode& status)
1032 {
1033 if(U_FAILURE(status)) {
1034 return;
1035 }
1036 if(!fCalendar) {
1037 status = U_ILLEGAL_ARGUMENT_ERROR;
1038 return;
1039 }
1040
1041 fCalendar->setTime(startDate, status);
1042 if(U_SUCCESS(status)) {
1043 fHaveDefaultCentury = TRUE;
1044 fDefaultCenturyStart = startDate;
1045 fDefaultCenturyStartYear = fCalendar->get(UCAL_YEAR, status);
1046 }
1047 }
1048
1049 //----------------------------------------------------------------------
1050
1051 UnicodeString&
1052 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo, FieldPosition& pos) const
1053 {
1054 UErrorCode status = U_ZERO_ERROR;
1055 FieldPositionOnlyHandler handler(pos);
1056 return _format(cal, appendTo, handler, status);
1057 }
1058
1059 //----------------------------------------------------------------------
1060
1061 UnicodeString&
1062 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo,
1063 FieldPositionIterator* posIter, UErrorCode& status) const
1064 {
1065 FieldPositionIteratorHandler handler(posIter, status);
1066 return _format(cal, appendTo, handler, status);
1067 }
1068
1069 //----------------------------------------------------------------------
1070
1071 UnicodeString&
1072 SimpleDateFormat::_format(Calendar& cal, UnicodeString& appendTo,
1073 FieldPositionHandler& handler, UErrorCode& status) const
1074 {
1075 if ( U_FAILURE(status) ) {
1076 return appendTo;
1077 }
1078 Calendar* workCal = &cal;
1079 Calendar* calClone = NULL;
1080 if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
1081 // Different calendar type
1082 // We use the time and time zone from the input calendar, but
1083 // do not use the input calendar for field calculation.
1084 calClone = fCalendar->clone();
1085 if (calClone != NULL) {
1086 UDate t = cal.getTime(status);
1087 calClone->setTime(t, status);
1088 calClone->setTimeZone(cal.getTimeZone());
1089 workCal = calClone;
1090 } else {
1091 status = U_MEMORY_ALLOCATION_ERROR;
1092 return appendTo;
1093 }
1094 }
1095
1096 UBool inQuote = FALSE;
1097 UChar prevCh = 0;
1098 int32_t count = 0;
1099 int32_t fieldNum = 0;
1100 UDisplayContext capitalizationContext = getContext(UDISPCTX_TYPE_CAPITALIZATION, status);
1101
1102 // Create temporary cache of mutable number format objects. This way
1103 // subFormat won't have to clone the const NumberFormat for each field.
1104 // if several fields share the same NumberFormat, which will almost
1105 // always be the case, this is a big save.
1106 SimpleDateFormatMutableNFs mutableNFs;
1107 // loop through the pattern string character by character
1108 for (int32_t i = 0; i < fPattern.length() && U_SUCCESS(status); ++i) {
1109 UChar ch = fPattern[i];
1110
1111 // Use subFormat() to format a repeated pattern character
1112 // when a different pattern or non-pattern character is seen
1113 if (ch != prevCh && count > 0) {
1114 subFormat(appendTo, prevCh, count, capitalizationContext, fieldNum++, handler, *workCal, mutableNFs, status);
1115 count = 0;
1116 }
1117 if (ch == QUOTE) {
1118 // Consecutive single quotes are a single quote literal,
1119 // either outside of quotes or between quotes
1120 if ((i+1) < fPattern.length() && fPattern[i+1] == QUOTE) {
1121 appendTo += (UChar)QUOTE;
1122 ++i;
1123 } else {
1124 inQuote = ! inQuote;
1125 }
1126 }
1127 else if (!inQuote && isSyntaxChar(ch)) {
1128 // ch is a date-time pattern character to be interpreted
1129 // by subFormat(); count the number of times it is repeated
1130 prevCh = ch;
1131 ++count;
1132 }
1133 else {
1134 // Append quoted characters and unquoted non-pattern characters
1135 appendTo += ch;
1136 }
1137 }
1138
1139 // Format the last item in the pattern, if any
1140 if (count > 0) {
1141 subFormat(appendTo, prevCh, count, capitalizationContext, fieldNum++, handler, *workCal, mutableNFs, status);
1142 }
1143
1144 if (calClone != NULL) {
1145 delete calClone;
1146 }
1147
1148 return appendTo;
1149 }
1150
1151 //----------------------------------------------------------------------
1152
1153 /* Map calendar field into calendar field level.
1154 * the larger the level, the smaller the field unit.
1155 * For example, UCAL_ERA level is 0, UCAL_YEAR level is 10,
1156 * UCAL_MONTH level is 20.
1157 * NOTE: if new fields adds in, the table needs to update.
1158 */
1159 const int32_t
1160 SimpleDateFormat::fgCalendarFieldToLevel[] =
1161 {
1162 /*GyM*/ 0, 10, 20,
1163 /*wW*/ 20, 30,
1164 /*dDEF*/ 30, 20, 30, 30,
1165 /*ahHm*/ 40, 50, 50, 60,
1166 /*sS*/ 70, 80,
1167 /*z?Y*/ 0, 0, 10,
1168 /*eug*/ 30, 10, 0,
1169 /*A?.*/ 40, 0, 0
1170 };
1171
1172 int32_t SimpleDateFormat::getLevelFromChar(UChar ch) {
1173 // Map date field LETTER into calendar field level.
1174 // the larger the level, the smaller the field unit.
1175 // NOTE: if new fields adds in, the table needs to update.
1176 static const int32_t mapCharToLevel[] = {
1177 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
1178 //
1179 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
1180 // ! " # $ % & ' ( ) * + , - . /
1181 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
1182 // 0 1 2 3 4 5 6 7 8 9 : ; < = > ?
1183 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, -1, -1, -1, -1, -1,
1184 // @ A B C D E F G H I J K L M N O
1185 -1, 40, -1, -1, 20, 30, 30, 0, 50, -1, -1, 50, 20, 20, -1, 0,
1186 // P Q R S T U V W X Y Z [ \ ] ^ _
1187 -1, 20, -1, 80, -1, 10, 0, 30, 0, 10, 0, -1, -1, -1, -1, -1,
1188 // ` a b c d e f g h i j k l m n o
1189 -1, 40, -1, 30, 30, 30, -1, 0, 50, -1, -1, 50, 0, 60, -1, -1,
1190 // p q r s t u v w x y z { | } ~
1191 -1, 20, 10, 70, -1, 10, 0, 20, 0, 10, 0, -1, -1, -1, -1, -1
1192 };
1193
1194 return ch < UPRV_LENGTHOF(mapCharToLevel) ? mapCharToLevel[ch] : -1;
1195 }
1196
1197 UBool SimpleDateFormat::isSyntaxChar(UChar ch) {
1198 static const UBool mapCharToIsSyntax[] = {
1199 //
1200 FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1201 //
1202 FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1203 //
1204 FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1205 //
1206 FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1207 // ! " # $ % & '
1208 FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1209 // ( ) * + , - . /
1210 FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1211 // 0 1 2 3 4 5 6 7
1212 FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,
1213 // 8 9 : ; < = > ?
1214 FALSE, FALSE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE,
1215 // @ A B C D E F G
1216 FALSE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
1217 // H I J K L M N O
1218 TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
1219 // P Q R S T U V W
1220 TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
1221 // X Y Z [ \ ] ^ _
1222 TRUE, TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE,
1223 // ` a b c d e f g
1224 FALSE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
1225 // h i j k l m n o
1226 TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
1227 // p q r s t u v w
1228 TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
1229 // x y z { | } ~
1230 TRUE, TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE
1231 };
1232
1233 return ch < UPRV_LENGTHOF(mapCharToIsSyntax) ? mapCharToIsSyntax[ch] : FALSE;
1234 }
1235
1236 // Map index into pattern character string to Calendar field number.
1237 const UCalendarDateFields
1238 SimpleDateFormat::fgPatternIndexToCalendarField[] =
1239 {
1240 /*GyM*/ UCAL_ERA, UCAL_YEAR, UCAL_MONTH,
1241 /*dkH*/ UCAL_DATE, UCAL_HOUR_OF_DAY, UCAL_HOUR_OF_DAY,
1242 /*msS*/ UCAL_MINUTE, UCAL_SECOND, UCAL_MILLISECOND,
1243 /*EDF*/ UCAL_DAY_OF_WEEK, UCAL_DAY_OF_YEAR, UCAL_DAY_OF_WEEK_IN_MONTH,
1244 /*wWa*/ UCAL_WEEK_OF_YEAR, UCAL_WEEK_OF_MONTH, UCAL_AM_PM,
1245 /*hKz*/ UCAL_HOUR, UCAL_HOUR, UCAL_ZONE_OFFSET,
1246 /*Yeu*/ UCAL_YEAR_WOY, UCAL_DOW_LOCAL, UCAL_EXTENDED_YEAR,
1247 /*gAZ*/ UCAL_JULIAN_DAY, UCAL_MILLISECONDS_IN_DAY, UCAL_ZONE_OFFSET,
1248 /*v*/ UCAL_ZONE_OFFSET,
1249 /*c*/ UCAL_DOW_LOCAL,
1250 /*L*/ UCAL_MONTH,
1251 /*Q*/ UCAL_MONTH,
1252 /*q*/ UCAL_MONTH,
1253 /*V*/ UCAL_ZONE_OFFSET,
1254 /*U*/ UCAL_YEAR,
1255 /*O*/ UCAL_ZONE_OFFSET,
1256 /*Xx*/ UCAL_ZONE_OFFSET, UCAL_ZONE_OFFSET,
1257 /*r*/ UCAL_EXTENDED_YEAR,
1258 /*:*/ UCAL_FIELD_COUNT, /* => no useful mapping to any calendar field */
1259 };
1260
1261 // Map index into pattern character string to DateFormat field number
1262 const UDateFormatField
1263 SimpleDateFormat::fgPatternIndexToDateFormatField[] = {
1264 /*GyM*/ UDAT_ERA_FIELD, UDAT_YEAR_FIELD, UDAT_MONTH_FIELD,
1265 /*dkH*/ UDAT_DATE_FIELD, UDAT_HOUR_OF_DAY1_FIELD, UDAT_HOUR_OF_DAY0_FIELD,
1266 /*msS*/ UDAT_MINUTE_FIELD, UDAT_SECOND_FIELD, UDAT_FRACTIONAL_SECOND_FIELD,
1267 /*EDF*/ UDAT_DAY_OF_WEEK_FIELD, UDAT_DAY_OF_YEAR_FIELD, UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
1268 /*wWa*/ UDAT_WEEK_OF_YEAR_FIELD, UDAT_WEEK_OF_MONTH_FIELD, UDAT_AM_PM_FIELD,
1269 /*hKz*/ UDAT_HOUR1_FIELD, UDAT_HOUR0_FIELD, UDAT_TIMEZONE_FIELD,
1270 /*Yeu*/ UDAT_YEAR_WOY_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_EXTENDED_YEAR_FIELD,
1271 /*gAZ*/ UDAT_JULIAN_DAY_FIELD, UDAT_MILLISECONDS_IN_DAY_FIELD, UDAT_TIMEZONE_RFC_FIELD,
1272 /*v*/ UDAT_TIMEZONE_GENERIC_FIELD,
1273 /*c*/ UDAT_STANDALONE_DAY_FIELD,
1274 /*L*/ UDAT_STANDALONE_MONTH_FIELD,
1275 /*Q*/ UDAT_QUARTER_FIELD,
1276 /*q*/ UDAT_STANDALONE_QUARTER_FIELD,
1277 /*V*/ UDAT_TIMEZONE_SPECIAL_FIELD,
1278 /*U*/ UDAT_YEAR_NAME_FIELD,
1279 /*O*/ UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD,
1280 /*Xx*/ UDAT_TIMEZONE_ISO_FIELD, UDAT_TIMEZONE_ISO_LOCAL_FIELD,
1281 /*r*/ UDAT_RELATED_YEAR_FIELD,
1282 /*:*/ UDAT_TIME_SEPARATOR_FIELD,
1283 };
1284
1285 //----------------------------------------------------------------------
1286
1287 /**
1288 * Append symbols[value] to dst. Make sure the array index is not out
1289 * of bounds.
1290 */
1291 static inline void
1292 _appendSymbol(UnicodeString& dst,
1293 int32_t value,
1294 const UnicodeString* symbols,
1295 int32_t symbolsCount) {
1296 U_ASSERT(0 <= value && value < symbolsCount);
1297 if (0 <= value && value < symbolsCount) {
1298 dst += symbols[value];
1299 }
1300 }
1301
1302 static inline void
1303 _appendSymbolWithMonthPattern(UnicodeString& dst, int32_t value, const UnicodeString* symbols, int32_t symbolsCount,
1304 const UnicodeString* monthPattern, UErrorCode& status) {
1305 U_ASSERT(0 <= value && value < symbolsCount);
1306 if (0 <= value && value < symbolsCount) {
1307 if (monthPattern == NULL) {
1308 dst += symbols[value];
1309 } else {
1310 Formattable monthName((const UnicodeString&)(symbols[value]));
1311 MessageFormat::format(*monthPattern, &monthName, 1, dst, status);
1312 }
1313 }
1314 }
1315
1316 //----------------------------------------------------------------------
1317 void
1318 SimpleDateFormat::initNumberFormatters(const Locale &locale,UErrorCode &status) {
1319 if (U_FAILURE(status)) {
1320 return;
1321 }
1322 if ( fDateOverride.isBogus() && fTimeOverride.isBogus() ) {
1323 return;
1324 }
1325 umtx_lock(&LOCK);
1326 if (fSharedNumberFormatters == NULL) {
1327 fSharedNumberFormatters = allocSharedNumberFormatters();
1328 if (fSharedNumberFormatters == NULL) {
1329 status = U_MEMORY_ALLOCATION_ERROR;
1330 }
1331 }
1332 umtx_unlock(&LOCK);
1333
1334 if (U_FAILURE(status)) {
1335 return;
1336 }
1337
1338 processOverrideString(locale,fDateOverride,kOvrStrDate,status);
1339 processOverrideString(locale,fTimeOverride,kOvrStrTime,status);
1340 }
1341
1342 void
1343 SimpleDateFormat::processOverrideString(const Locale &locale, const UnicodeString &str, int8_t type, UErrorCode &status) {
1344 if (str.isBogus() || U_FAILURE(status)) {
1345 return;
1346 }
1347
1348 int32_t start = 0;
1349 int32_t len;
1350 UnicodeString nsName;
1351 UnicodeString ovrField;
1352 UBool moreToProcess = TRUE;
1353 NSOverride *overrideList = NULL;
1354
1355 while (moreToProcess) {
1356 int32_t delimiterPosition = str.indexOf((UChar)ULOC_KEYWORD_ITEM_SEPARATOR_UNICODE,start);
1357 if (delimiterPosition == -1) {
1358 moreToProcess = FALSE;
1359 len = str.length() - start;
1360 } else {
1361 len = delimiterPosition - start;
1362 }
1363 UnicodeString currentString(str,start,len);
1364 int32_t equalSignPosition = currentString.indexOf((UChar)ULOC_KEYWORD_ASSIGN_UNICODE,0);
1365 if (equalSignPosition == -1) { // Simple override string such as "hebrew"
1366 nsName.setTo(currentString);
1367 ovrField.setToBogus();
1368 } else { // Field specific override string such as "y=hebrew"
1369 nsName.setTo(currentString,equalSignPosition+1);
1370 ovrField.setTo(currentString,0,1); // We just need the first character.
1371 }
1372
1373 int32_t nsNameHash = nsName.hashCode();
1374 // See if the numbering system is in the override list, if not, then add it.
1375 NSOverride *cur = overrideList;
1376 const SharedNumberFormat *snf = NULL;
1377 UBool found = FALSE;
1378 while ( cur && !found ) {
1379 if ( cur->hash == nsNameHash ) {
1380 snf = cur->snf;
1381 found = TRUE;
1382 }
1383 cur = cur->next;
1384 }
1385
1386 if (!found) {
1387 LocalPointer<NSOverride> cur(new NSOverride);
1388 if (!cur.isNull()) {
1389 char kw[ULOC_KEYWORD_AND_VALUES_CAPACITY];
1390 uprv_strcpy(kw,"numbers=");
1391 nsName.extract(0,len,kw+8,ULOC_KEYWORD_AND_VALUES_CAPACITY-8,US_INV);
1392
1393 Locale ovrLoc(locale.getLanguage(),locale.getCountry(),locale.getVariant(),kw);
1394 cur->hash = nsNameHash;
1395 cur->next = overrideList;
1396 SharedObject::copyPtr(
1397 createSharedNumberFormat(ovrLoc, status), cur->snf);
1398 if (U_FAILURE(status)) {
1399 if (overrideList) {
1400 overrideList->free();
1401 }
1402 return;
1403 }
1404 snf = cur->snf;
1405 overrideList = cur.orphan();
1406 } else {
1407 status = U_MEMORY_ALLOCATION_ERROR;
1408 if (overrideList) {
1409 overrideList->free();
1410 }
1411 return;
1412 }
1413 }
1414
1415 // Now that we have an appropriate number formatter, fill in the appropriate spaces in the
1416 // number formatters table.
1417 if (ovrField.isBogus()) {
1418 switch (type) {
1419 case kOvrStrDate:
1420 case kOvrStrBoth: {
1421 for ( int8_t i=0 ; i<kDateFieldsCount; i++ ) {
1422 SharedObject::copyPtr(snf, fSharedNumberFormatters[kDateFields[i]]);
1423 }
1424 if (type==kOvrStrDate) {
1425 break;
1426 }
1427 }
1428 case kOvrStrTime : {
1429 for ( int8_t i=0 ; i<kTimeFieldsCount; i++ ) {
1430 SharedObject::copyPtr(snf, fSharedNumberFormatters[kTimeFields[i]]);
1431 }
1432 break;
1433 }
1434 }
1435 } else {
1436 // if the pattern character is unrecognized, signal an error and bail out
1437 UDateFormatField patternCharIndex =
1438 DateFormatSymbols::getPatternCharIndex(ovrField.charAt(0));
1439 if (patternCharIndex == UDAT_FIELD_COUNT) {
1440 status = U_INVALID_FORMAT_ERROR;
1441 if (overrideList) {
1442 overrideList->free();
1443 }
1444 return;
1445 }
1446 SharedObject::copyPtr(snf, fSharedNumberFormatters[patternCharIndex]);
1447 }
1448
1449 start = delimiterPosition + 1;
1450 }
1451 if (overrideList) {
1452 overrideList->free();
1453 }
1454 }
1455
1456 //---------------------------------------------------------------------
1457 void
1458 SimpleDateFormat::subFormat(UnicodeString &appendTo,
1459 UChar ch,
1460 int32_t count,
1461 UDisplayContext capitalizationContext,
1462 int32_t fieldNum,
1463 FieldPositionHandler& handler,
1464 Calendar& cal,
1465 SimpleDateFormatMutableNFs &mutableNFs,
1466 UErrorCode& status) const
1467 {
1468 if (U_FAILURE(status)) {
1469 return;
1470 }
1471
1472 // this function gets called by format() to produce the appropriate substitution
1473 // text for an individual pattern symbol (e.g., "HH" or "yyyy")
1474
1475 UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
1476 const int32_t maxIntCount = 10;
1477 int32_t beginOffset = appendTo.length();
1478 NumberFormat *currentNumberFormat;
1479 DateFormatSymbols::ECapitalizationContextUsageType capContextUsageType = DateFormatSymbols::kCapContextUsageOther;
1480
1481 UBool isHebrewCalendar = (uprv_strcmp(cal.getType(),"hebrew") == 0);
1482 UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0 || uprv_strcmp(cal.getType(),"dangi") == 0);
1483
1484 // if the pattern character is unrecognized, signal an error and dump out
1485 if (patternCharIndex == UDAT_FIELD_COUNT)
1486 {
1487 if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
1488 status = U_INVALID_FORMAT_ERROR;
1489 }
1490 return;
1491 }
1492
1493 UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex];
1494 int32_t value = 0;
1495 // Don't get value unless it is useful
1496 if (field < UCAL_FIELD_COUNT) {
1497 value = (patternCharIndex != UDAT_RELATED_YEAR_FIELD)? cal.get(field, status): cal.getRelatedYear(status);
1498 }
1499 if (U_FAILURE(status)) {
1500 return;
1501 }
1502
1503 currentNumberFormat = mutableNFs.get(getNumberFormatByIndex(patternCharIndex));
1504 if (currentNumberFormat == NULL) {
1505 status = U_MEMORY_ALLOCATION_ERROR;
1506 return;
1507 }
1508 UnicodeString hebr("hebr", 4, US_INV);
1509
1510 switch (patternCharIndex) {
1511
1512 // for any "G" symbol, write out the appropriate era string
1513 // "GGGG" is wide era name, "GGGGG" is narrow era name, anything else is abbreviated name
1514 case UDAT_ERA_FIELD:
1515 if (isChineseCalendar) {
1516 zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, 9); // as in ICU4J
1517 } else {
1518 if (count == 5) {
1519 _appendSymbol(appendTo, value, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount);
1520 capContextUsageType = DateFormatSymbols::kCapContextUsageEraNarrow;
1521 } else if (count == 4) {
1522 _appendSymbol(appendTo, value, fSymbols->fEraNames, fSymbols->fEraNamesCount);
1523 capContextUsageType = DateFormatSymbols::kCapContextUsageEraWide;
1524 } else {
1525 _appendSymbol(appendTo, value, fSymbols->fEras, fSymbols->fErasCount);
1526 capContextUsageType = DateFormatSymbols::kCapContextUsageEraAbbrev;
1527 }
1528 }
1529 break;
1530
1531 case UDAT_YEAR_NAME_FIELD:
1532 if (fSymbols->fShortYearNames != NULL && value <= fSymbols->fShortYearNamesCount) {
1533 // the Calendar YEAR field runs 1 through 60 for cyclic years
1534 _appendSymbol(appendTo, value - 1, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount);
1535 break;
1536 }
1537 // else fall through to numeric year handling, do not break here
1538
1539 // OLD: for "yyyy", write out the whole year; for "yy", write out the last 2 digits
1540 // NEW: UTS#35:
1541 //Year y yy yyy yyyy yyyyy
1542 //AD 1 1 01 001 0001 00001
1543 //AD 12 12 12 012 0012 00012
1544 //AD 123 123 23 123 0123 00123
1545 //AD 1234 1234 34 1234 1234 01234
1546 //AD 12345 12345 45 12345 12345 12345
1547 case UDAT_YEAR_FIELD:
1548 case UDAT_YEAR_WOY_FIELD:
1549 if (fDateOverride.compare(hebr)==0 && value>HEBREW_CAL_CUR_MILLENIUM_START_YEAR && value<HEBREW_CAL_CUR_MILLENIUM_END_YEAR) {
1550 value-=HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
1551 }
1552 if(count == 2)
1553 zeroPaddingNumber(currentNumberFormat, appendTo, value, 2, 2);
1554 else
1555 zeroPaddingNumber(currentNumberFormat, appendTo, value, count, maxIntCount);
1556 break;
1557
1558 // for "MMMM"/"LLLL", write out the whole month name, for "MMM"/"LLL", write out the month
1559 // abbreviation, for "M"/"L" or "MM"/"LL", write out the month as a number with the
1560 // appropriate number of digits
1561 // for "MMMMM"/"LLLLL", use the narrow form
1562 case UDAT_MONTH_FIELD:
1563 case UDAT_STANDALONE_MONTH_FIELD:
1564 if ( isHebrewCalendar ) {
1565 HebrewCalendar *hc = (HebrewCalendar*)&cal;
1566 if (hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value == 6 && count >= 3 )
1567 value = 13; // Show alternate form for Adar II in leap years in Hebrew calendar.
1568 if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6 && count < 3 )
1569 value--; // Adjust the month number down 1 in Hebrew non-leap years, i.e. Adar is 6, not 7.
1570 }
1571 {
1572 int32_t isLeapMonth = (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount)?
1573 cal.get(UCAL_IS_LEAP_MONTH, status): 0;
1574 // should consolidate the next section by using arrays of pointers & counts for the right symbols...
1575 if (count == 5) {
1576 if (patternCharIndex == UDAT_MONTH_FIELD) {
1577 _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fNarrowMonths, fSymbols->fNarrowMonthsCount,
1578 (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatNarrow]): NULL, status);
1579 } else {
1580 _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneNarrowMonths, fSymbols->fStandaloneNarrowMonthsCount,
1581 (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneNarrow]): NULL, status);
1582 }
1583 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthNarrow;
1584 } else if (count == 4) {
1585 if (patternCharIndex == UDAT_MONTH_FIELD) {
1586 _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fMonths, fSymbols->fMonthsCount,
1587 (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide]): NULL, status);
1588 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1589 } else {
1590 _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount,
1591 (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide]): NULL, status);
1592 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1593 }
1594 } else if (count == 3) {
1595 if (patternCharIndex == UDAT_MONTH_FIELD) {
1596 _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fShortMonths, fSymbols->fShortMonthsCount,
1597 (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev]): NULL, status);
1598 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1599 } else {
1600 _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount,
1601 (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev]): NULL, status);
1602 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1603 }
1604 } else {
1605 UnicodeString monthNumber;
1606 zeroPaddingNumber(currentNumberFormat,monthNumber, value + 1, count, maxIntCount);
1607 _appendSymbolWithMonthPattern(appendTo, 0, &monthNumber, 1,
1608 (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric]): NULL, status);
1609 }
1610 }
1611 break;
1612
1613 // for "k" and "kk", write out the hour, adjusting midnight to appear as "24"
1614 case UDAT_HOUR_OF_DAY1_FIELD:
1615 if (value == 0)
1616 zeroPaddingNumber(currentNumberFormat,appendTo, cal.getMaximum(UCAL_HOUR_OF_DAY) + 1, count, maxIntCount);
1617 else
1618 zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1619 break;
1620
1621 case UDAT_FRACTIONAL_SECOND_FIELD:
1622 // Fractional seconds left-justify
1623 {
1624 currentNumberFormat->setMinimumIntegerDigits((count > 3) ? 3 : count);
1625 currentNumberFormat->setMaximumIntegerDigits(maxIntCount);
1626 if (count == 1) {
1627 value /= 100;
1628 } else if (count == 2) {
1629 value /= 10;
1630 }
1631 FieldPosition p(0);
1632 currentNumberFormat->format(value, appendTo, p);
1633 if (count > 3) {
1634 currentNumberFormat->setMinimumIntegerDigits(count - 3);
1635 currentNumberFormat->format((int32_t)0, appendTo, p);
1636 }
1637 }
1638 break;
1639
1640 // for "ee" or "e", use local numeric day-of-the-week
1641 // for "EEEEEE" or "eeeeee", write out the short day-of-the-week name
1642 // for "EEEEE" or "eeeee", write out the narrow day-of-the-week name
1643 // for "EEEE" or "eeee", write out the wide day-of-the-week name
1644 // for "EEE" or "EE" or "E" or "eee", write out the abbreviated day-of-the-week name
1645 case UDAT_DOW_LOCAL_FIELD:
1646 if ( count < 3 ) {
1647 zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1648 break;
1649 }
1650 // fall through to EEEEE-EEE handling, but for that we don't want local day-of-week,
1651 // we want standard day-of-week, so first fix value to work for EEEEE-EEE.
1652 value = cal.get(UCAL_DAY_OF_WEEK, status);
1653 if (U_FAILURE(status)) {
1654 return;
1655 }
1656 // fall through, do not break here
1657 case UDAT_DAY_OF_WEEK_FIELD:
1658 if (count == 5) {
1659 _appendSymbol(appendTo, value, fSymbols->fNarrowWeekdays,
1660 fSymbols->fNarrowWeekdaysCount);
1661 capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1662 } else if (count == 4) {
1663 _appendSymbol(appendTo, value, fSymbols->fWeekdays,
1664 fSymbols->fWeekdaysCount);
1665 capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1666 } else if (count == 6) {
1667 _appendSymbol(appendTo, value, fSymbols->fShorterWeekdays,
1668 fSymbols->fShorterWeekdaysCount);
1669 capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1670 } else {
1671 _appendSymbol(appendTo, value, fSymbols->fShortWeekdays,
1672 fSymbols->fShortWeekdaysCount);
1673 capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1674 }
1675 break;
1676
1677 // for "ccc", write out the abbreviated day-of-the-week name
1678 // for "cccc", write out the wide day-of-the-week name
1679 // for "ccccc", use the narrow day-of-the-week name
1680 // for "ccccc", use the short day-of-the-week name
1681 case UDAT_STANDALONE_DAY_FIELD:
1682 if ( count < 3 ) {
1683 zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, maxIntCount);
1684 break;
1685 }
1686 // fall through to alpha DOW handling, but for that we don't want local day-of-week,
1687 // we want standard day-of-week, so first fix value.
1688 value = cal.get(UCAL_DAY_OF_WEEK, status);
1689 if (U_FAILURE(status)) {
1690 return;
1691 }
1692 if (count == 5) {
1693 _appendSymbol(appendTo, value, fSymbols->fStandaloneNarrowWeekdays,
1694 fSymbols->fStandaloneNarrowWeekdaysCount);
1695 capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1696 } else if (count == 4) {
1697 _appendSymbol(appendTo, value, fSymbols->fStandaloneWeekdays,
1698 fSymbols->fStandaloneWeekdaysCount);
1699 capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1700 } else if (count == 6) {
1701 _appendSymbol(appendTo, value, fSymbols->fStandaloneShorterWeekdays,
1702 fSymbols->fStandaloneShorterWeekdaysCount);
1703 capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1704 } else { // count == 3
1705 _appendSymbol(appendTo, value, fSymbols->fStandaloneShortWeekdays,
1706 fSymbols->fStandaloneShortWeekdaysCount);
1707 capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1708 }
1709 break;
1710
1711 // for "a" symbol, write out the whole AM/PM string
1712 case UDAT_AM_PM_FIELD:
1713 if (count < 5) {
1714 _appendSymbol(appendTo, value, fSymbols->fAmPms,
1715 fSymbols->fAmPmsCount);
1716 } else {
1717 _appendSymbol(appendTo, value, fSymbols->fNarrowAmPms,
1718 fSymbols->fNarrowAmPmsCount);
1719 }
1720 break;
1721
1722 // for ":", write out the time separator string
1723 case UDAT_TIME_SEPARATOR_FIELD:
1724 {
1725 UnicodeString separator;
1726 appendTo += fSymbols->getTimeSeparatorString(separator);
1727 }
1728 break;
1729
1730 // for "h" and "hh", write out the hour, adjusting noon and midnight to show up
1731 // as "12"
1732 case UDAT_HOUR1_FIELD:
1733 if (value == 0)
1734 zeroPaddingNumber(currentNumberFormat,appendTo, cal.getLeastMaximum(UCAL_HOUR) + 1, count, maxIntCount);
1735 else
1736 zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1737 break;
1738
1739 case UDAT_TIMEZONE_FIELD: // 'z'
1740 case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
1741 case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
1742 case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
1743 case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
1744 case UDAT_TIMEZONE_ISO_FIELD: // 'X'
1745 case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
1746 {
1747 UChar zsbuf[64];
1748 UnicodeString zoneString(zsbuf, 0, UPRV_LENGTHOF(zsbuf));
1749 const TimeZone& tz = cal.getTimeZone();
1750 UDate date = cal.getTime(status);
1751 if (U_SUCCESS(status)) {
1752 if (patternCharIndex == UDAT_TIMEZONE_FIELD) {
1753 if (count < 4) {
1754 // "z", "zz", "zzz"
1755 tzFormat()->format(UTZFMT_STYLE_SPECIFIC_SHORT, tz, date, zoneString);
1756 capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1757 } else {
1758 // "zzzz" or longer
1759 tzFormat()->format(UTZFMT_STYLE_SPECIFIC_LONG, tz, date, zoneString);
1760 capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1761 }
1762 }
1763 else if (patternCharIndex == UDAT_TIMEZONE_RFC_FIELD) {
1764 if (count < 4) {
1765 // "Z"
1766 tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
1767 } else if (count == 5) {
1768 // "ZZZZZ"
1769 tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
1770 } else {
1771 // "ZZ", "ZZZ", "ZZZZ"
1772 tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
1773 }
1774 }
1775 else if (patternCharIndex == UDAT_TIMEZONE_GENERIC_FIELD) {
1776 if (count == 1) {
1777 // "v"
1778 tzFormat()->format(UTZFMT_STYLE_GENERIC_SHORT, tz, date, zoneString);
1779 capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1780 } else if (count == 4) {
1781 // "vvvv"
1782 tzFormat()->format(UTZFMT_STYLE_GENERIC_LONG, tz, date, zoneString);
1783 capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1784 }
1785 }
1786 else if (patternCharIndex == UDAT_TIMEZONE_SPECIAL_FIELD) {
1787 if (count == 1) {
1788 // "V"
1789 tzFormat()->format(UTZFMT_STYLE_ZONE_ID_SHORT, tz, date, zoneString);
1790 } else if (count == 2) {
1791 // "VV"
1792 tzFormat()->format(UTZFMT_STYLE_ZONE_ID, tz, date, zoneString);
1793 } else if (count == 3) {
1794 // "VVV"
1795 tzFormat()->format(UTZFMT_STYLE_EXEMPLAR_LOCATION, tz, date, zoneString);
1796 } else if (count == 4) {
1797 // "VVVV"
1798 tzFormat()->format(UTZFMT_STYLE_GENERIC_LOCATION, tz, date, zoneString);
1799 capContextUsageType = DateFormatSymbols::kCapContextUsageZoneLong;
1800 }
1801 }
1802 else if (patternCharIndex == UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD) {
1803 if (count == 1) {
1804 // "O"
1805 tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT_SHORT, tz, date, zoneString);
1806 } else if (count == 4) {
1807 // "OOOO"
1808 tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
1809 }
1810 }
1811 else if (patternCharIndex == UDAT_TIMEZONE_ISO_FIELD) {
1812 if (count == 1) {
1813 // "X"
1814 tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_SHORT, tz, date, zoneString);
1815 } else if (count == 2) {
1816 // "XX"
1817 tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_FIXED, tz, date, zoneString);
1818 } else if (count == 3) {
1819 // "XXX"
1820 tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FIXED, tz, date, zoneString);
1821 } else if (count == 4) {
1822 // "XXXX"
1823 tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_FULL, tz, date, zoneString);
1824 } else if (count == 5) {
1825 // "XXXXX"
1826 tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
1827 }
1828 }
1829 else if (patternCharIndex == UDAT_TIMEZONE_ISO_LOCAL_FIELD) {
1830 if (count == 1) {
1831 // "x"
1832 tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT, tz, date, zoneString);
1833 } else if (count == 2) {
1834 // "xx"
1835 tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED, tz, date, zoneString);
1836 } else if (count == 3) {
1837 // "xxx"
1838 tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED, tz, date, zoneString);
1839 } else if (count == 4) {
1840 // "xxxx"
1841 tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
1842 } else if (count == 5) {
1843 // "xxxxx"
1844 tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL, tz, date, zoneString);
1845 }
1846 }
1847 else {
1848 U_ASSERT(FALSE);
1849 }
1850 }
1851 appendTo += zoneString;
1852 }
1853 break;
1854
1855 case UDAT_QUARTER_FIELD:
1856 if (count >= 4)
1857 _appendSymbol(appendTo, value/3, fSymbols->fQuarters,
1858 fSymbols->fQuartersCount);
1859 else if (count == 3)
1860 _appendSymbol(appendTo, value/3, fSymbols->fShortQuarters,
1861 fSymbols->fShortQuartersCount);
1862 else
1863 zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1864 break;
1865
1866 case UDAT_STANDALONE_QUARTER_FIELD:
1867 if (count >= 4)
1868 _appendSymbol(appendTo, value/3, fSymbols->fStandaloneQuarters,
1869 fSymbols->fStandaloneQuartersCount);
1870 else if (count == 3)
1871 _appendSymbol(appendTo, value/3, fSymbols->fStandaloneShortQuarters,
1872 fSymbols->fStandaloneShortQuartersCount);
1873 else
1874 zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1875 break;
1876
1877
1878 // all of the other pattern symbols can be formatted as simple numbers with
1879 // appropriate zero padding
1880 default:
1881 zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1882 break;
1883 }
1884 #if !UCONFIG_NO_BREAK_ITERATION
1885 // if first field, check to see whether we need to and are able to titlecase it
1886 if (fieldNum == 0 && u_islower(appendTo.char32At(beginOffset)) && fCapitalizationBrkIter != NULL) {
1887 UBool titlecase = FALSE;
1888 switch (capitalizationContext) {
1889 case UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE:
1890 titlecase = TRUE;
1891 break;
1892 case UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU:
1893 titlecase = fSymbols->fCapitalization[capContextUsageType][0];
1894 break;
1895 case UDISPCTX_CAPITALIZATION_FOR_STANDALONE:
1896 titlecase = fSymbols->fCapitalization[capContextUsageType][1];
1897 break;
1898 default:
1899 // titlecase = FALSE;
1900 break;
1901 }
1902 if (titlecase) {
1903 UnicodeString firstField(appendTo, beginOffset);
1904 firstField.toTitle(fCapitalizationBrkIter, fLocale, U_TITLECASE_NO_LOWERCASE | U_TITLECASE_NO_BREAK_ADJUSTMENT);
1905 appendTo.replaceBetween(beginOffset, appendTo.length(), firstField);
1906 }
1907 }
1908 #endif
1909
1910 handler.addAttribute(fgPatternIndexToDateFormatField[patternCharIndex], beginOffset, appendTo.length());
1911 }
1912
1913 //----------------------------------------------------------------------
1914
1915 void SimpleDateFormat::adoptNumberFormat(NumberFormat *formatToAdopt) {
1916 fixNumberFormatForDates(*formatToAdopt);
1917 delete fNumberFormat;
1918 fNumberFormat = formatToAdopt;
1919
1920 // We successfully set the default number format. Now delete the overrides
1921 // (can't fail).
1922 if (fSharedNumberFormatters) {
1923 freeSharedNumberFormatters(fSharedNumberFormatters);
1924 fSharedNumberFormatters = NULL;
1925 }
1926 }
1927
1928 void SimpleDateFormat::adoptNumberFormat(const UnicodeString& fields, NumberFormat *formatToAdopt, UErrorCode &status){
1929 fixNumberFormatForDates(*formatToAdopt);
1930 LocalPointer<NumberFormat> fmt(formatToAdopt);
1931 if (U_FAILURE(status)) {
1932 return;
1933 }
1934
1935 // We must ensure fSharedNumberFormatters is allocated.
1936 if (fSharedNumberFormatters == NULL) {
1937 fSharedNumberFormatters = allocSharedNumberFormatters();
1938 if (fSharedNumberFormatters == NULL) {
1939 status = U_MEMORY_ALLOCATION_ERROR;
1940 return;
1941 }
1942 }
1943 const SharedNumberFormat *newFormat = createSharedNumberFormat(fmt.orphan());
1944 if (newFormat == NULL) {
1945 status = U_MEMORY_ALLOCATION_ERROR;
1946 return;
1947 }
1948 for (int i=0; i<fields.length(); i++) {
1949 UChar field = fields.charAt(i);
1950 // if the pattern character is unrecognized, signal an error and bail out
1951 UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(field);
1952 if (patternCharIndex == UDAT_FIELD_COUNT) {
1953 status = U_INVALID_FORMAT_ERROR;
1954 newFormat->deleteIfZeroRefCount();
1955 return;
1956 }
1957
1958 // Set the number formatter in the table
1959 SharedObject::copyPtr(
1960 newFormat, fSharedNumberFormatters[patternCharIndex]);
1961 }
1962 newFormat->deleteIfZeroRefCount();
1963 }
1964
1965 const NumberFormat *
1966 SimpleDateFormat::getNumberFormatForField(UChar field) const {
1967 UDateFormatField index = DateFormatSymbols::getPatternCharIndex(field);
1968 if (index == UDAT_FIELD_COUNT) {
1969 return NULL;
1970 }
1971 return getNumberFormatByIndex(index);
1972 }
1973
1974 //----------------------------------------------------------------------
1975 void
1976 SimpleDateFormat::zeroPaddingNumber(
1977 NumberFormat *currentNumberFormat,
1978 UnicodeString &appendTo,
1979 int32_t value, int32_t minDigits, int32_t maxDigits) const
1980 {
1981 if (currentNumberFormat!=NULL) {
1982 FieldPosition pos(0);
1983
1984 currentNumberFormat->setMinimumIntegerDigits(minDigits);
1985 currentNumberFormat->setMaximumIntegerDigits(maxDigits);
1986 currentNumberFormat->format(value, appendTo, pos); // 3rd arg is there to speed up processing
1987 }
1988 }
1989
1990 //----------------------------------------------------------------------
1991
1992 /**
1993 * Return true if the given format character, occuring count
1994 * times, represents a numeric field.
1995 */
1996 UBool SimpleDateFormat::isNumeric(UChar formatChar, int32_t count) {
1997 return DateFormatSymbols::isNumericPatternChar(formatChar, count);
1998 }
1999
2000 UBool
2001 SimpleDateFormat::isAtNumericField(const UnicodeString &pattern, int32_t patternOffset) {
2002 if (patternOffset >= pattern.length()) {
2003 // not at any field
2004 return FALSE;
2005 }
2006 UChar ch = pattern.charAt(patternOffset);
2007 UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
2008 if (f == UDAT_FIELD_COUNT) {
2009 // not at any field
2010 return FALSE;
2011 }
2012 int32_t i = patternOffset;
2013 while (pattern.charAt(++i) == ch) {}
2014 return DateFormatSymbols::isNumericField(f, i - patternOffset);
2015 }
2016
2017 UBool
2018 SimpleDateFormat::isAfterNonNumericField(const UnicodeString &pattern, int32_t patternOffset) {
2019 if (patternOffset <= 0) {
2020 // not after any field
2021 return FALSE;
2022 }
2023 UChar ch = pattern.charAt(--patternOffset);
2024 UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
2025 if (f == UDAT_FIELD_COUNT) {
2026 // not after any field
2027 return FALSE;
2028 }
2029 int32_t i = patternOffset;
2030 while (pattern.charAt(--i) == ch) {}
2031 return !DateFormatSymbols::isNumericField(f, patternOffset - i);
2032 }
2033
2034 void
2035 SimpleDateFormat::parse(const UnicodeString& text, Calendar& cal, ParsePosition& parsePos) const
2036 {
2037 UErrorCode status = U_ZERO_ERROR;
2038 int32_t pos = parsePos.getIndex();
2039 if(parsePos.getIndex() < 0) {
2040 parsePos.setErrorIndex(0);
2041 return;
2042 }
2043 int32_t start = pos;
2044
2045
2046 UBool ambiguousYear[] = { FALSE };
2047 int32_t saveHebrewMonth = -1;
2048 int32_t count = 0;
2049 UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2050 SimpleDateFormatMutableNFs mutableNFs;
2051
2052 // For parsing abutting numeric fields. 'abutPat' is the
2053 // offset into 'pattern' of the first of 2 or more abutting
2054 // numeric fields. 'abutStart' is the offset into 'text'
2055 // where parsing the fields begins. 'abutPass' starts off as 0
2056 // and increments each time we try to parse the fields.
2057 int32_t abutPat = -1; // If >=0, we are in a run of abutting numeric fields
2058 int32_t abutStart = 0;
2059 int32_t abutPass = 0;
2060 UBool inQuote = FALSE;
2061
2062 MessageFormat * numericLeapMonthFormatter = NULL;
2063
2064 Calendar* calClone = NULL;
2065 Calendar *workCal = &cal;
2066 if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
2067 // Different calendar type
2068 // We use the time/zone from the input calendar, but
2069 // do not use the input calendar for field calculation.
2070 calClone = fCalendar->clone();
2071 if (calClone != NULL) {
2072 calClone->setTime(cal.getTime(status),status);
2073 if (U_FAILURE(status)) {
2074 goto ExitParse;
2075 }
2076 calClone->setTimeZone(cal.getTimeZone());
2077 workCal = calClone;
2078 } else {
2079 status = U_MEMORY_ALLOCATION_ERROR;
2080 goto ExitParse;
2081 }
2082 }
2083
2084 if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
2085 numericLeapMonthFormatter = new MessageFormat(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric], fLocale, status);
2086 if (numericLeapMonthFormatter == NULL) {
2087 status = U_MEMORY_ALLOCATION_ERROR;
2088 goto ExitParse;
2089 } else if (U_FAILURE(status)) {
2090 goto ExitParse; // this will delete numericLeapMonthFormatter
2091 }
2092 }
2093
2094 for (int32_t i=0; i<fPattern.length(); ++i) {
2095 UChar ch = fPattern.charAt(i);
2096
2097 // Handle alphabetic field characters.
2098 if (!inQuote && isSyntaxChar(ch)) {
2099 int32_t fieldPat = i;
2100
2101 // Count the length of this field specifier
2102 count = 1;
2103 while ((i+1)<fPattern.length() &&
2104 fPattern.charAt(i+1) == ch) {
2105 ++count;
2106 ++i;
2107 }
2108
2109 if (isNumeric(ch, count)) {
2110 if (abutPat < 0) {
2111 // Determine if there is an abutting numeric field.
2112 // Record the start of a set of abutting numeric fields.
2113 if (isAtNumericField(fPattern, i + 1)) {
2114 abutPat = fieldPat;
2115 abutStart = pos;
2116 abutPass = 0;
2117 }
2118 }
2119 } else {
2120 abutPat = -1; // End of any abutting fields
2121 }
2122
2123 // Handle fields within a run of abutting numeric fields. Take
2124 // the pattern "HHmmss" as an example. We will try to parse
2125 // 2/2/2 characters of the input text, then if that fails,
2126 // 1/2/2. We only adjust the width of the leftmost field; the
2127 // others remain fixed. This allows "123456" => 12:34:56, but
2128 // "12345" => 1:23:45. Likewise, for the pattern "yyyyMMdd" we
2129 // try 4/2/2, 3/2/2, 2/2/2, and finally 1/2/2.
2130 if (abutPat >= 0) {
2131 // If we are at the start of a run of abutting fields, then
2132 // shorten this field in each pass. If we can't shorten
2133 // this field any more, then the parse of this set of
2134 // abutting numeric fields has failed.
2135 if (fieldPat == abutPat) {
2136 count -= abutPass++;
2137 if (count == 0) {
2138 status = U_PARSE_ERROR;
2139 goto ExitParse;
2140 }
2141 }
2142
2143 pos = subParse(text, pos, ch, count,
2144 TRUE, FALSE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter, &tzTimeType, mutableNFs);
2145
2146 // If the parse fails anywhere in the run, back up to the
2147 // start of the run and retry.
2148 if (pos < 0) {
2149 i = abutPat - 1;
2150 pos = abutStart;
2151 continue;
2152 }
2153 }
2154
2155 // Handle non-numeric fields and non-abutting numeric
2156 // fields.
2157 else if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
2158 int32_t s = subParse(text, pos, ch, count,
2159 FALSE, TRUE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter, &tzTimeType, mutableNFs);
2160
2161 if (s == -pos-1) {
2162 // era not present, in special cases allow this to continue
2163 // from the position where the era was expected
2164 s = pos;
2165
2166 if (i+1 < fPattern.length()) {
2167 // move to next pattern character
2168 UChar ch = fPattern.charAt(i+1);
2169
2170 // check for whitespace
2171 if (PatternProps::isWhiteSpace(ch)) {
2172 i++;
2173 // Advance over run in pattern
2174 while ((i+1)<fPattern.length() &&
2175 PatternProps::isWhiteSpace(fPattern.charAt(i+1))) {
2176 ++i;
2177 }
2178 }
2179 }
2180 }
2181 else if (s <= 0) {
2182 status = U_PARSE_ERROR;
2183 goto ExitParse;
2184 }
2185 pos = s;
2186 }
2187 }
2188
2189 // Handle literal pattern characters. These are any
2190 // quoted characters and non-alphabetic unquoted
2191 // characters.
2192 else {
2193
2194 abutPat = -1; // End of any abutting fields
2195
2196 if (! matchLiterals(fPattern, i, text, pos, getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status), getBooleanAttribute(UDAT_PARSE_PARTIAL_MATCH, status), isLenient())) {
2197 status = U_PARSE_ERROR;
2198 goto ExitParse;
2199 }
2200 }
2201 }
2202
2203 // Special hack for trailing "." after non-numeric field.
2204 if (text.charAt(pos) == 0x2e && getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) {
2205 // only do if the last field is not numeric
2206 if (isAfterNonNumericField(fPattern, fPattern.length())) {
2207 pos++; // skip the extra "."
2208 }
2209 }
2210
2211 // At this point the fields of Calendar have been set. Calendar
2212 // will fill in default values for missing fields when the time
2213 // is computed.
2214
2215 parsePos.setIndex(pos);
2216
2217 // This part is a problem: When we call parsedDate.after, we compute the time.
2218 // Take the date April 3 2004 at 2:30 am. When this is first set up, the year
2219 // will be wrong if we're parsing a 2-digit year pattern. It will be 1904.
2220 // April 3 1904 is a Sunday (unlike 2004) so it is the DST onset day. 2:30 am
2221 // is therefore an "impossible" time, since the time goes from 1:59 to 3:00 am
2222 // on that day. It is therefore parsed out to fields as 3:30 am. Then we
2223 // add 100 years, and get April 3 2004 at 3:30 am. Note that April 3 2004 is
2224 // a Saturday, so it can have a 2:30 am -- and it should. [LIU]
2225 /*
2226 UDate parsedDate = calendar.getTime();
2227 if( ambiguousYear[0] && !parsedDate.after(fDefaultCenturyStart) ) {
2228 calendar.add(Calendar.YEAR, 100);
2229 parsedDate = calendar.getTime();
2230 }
2231 */
2232 // Because of the above condition, save off the fields in case we need to readjust.
2233 // The procedure we use here is not particularly efficient, but there is no other
2234 // way to do this given the API restrictions present in Calendar. We minimize
2235 // inefficiency by only performing this computation when it might apply, that is,
2236 // when the two-digit year is equal to the start year, and thus might fall at the
2237 // front or the back of the default century. This only works because we adjust
2238 // the year correctly to start with in other cases -- see subParse().
2239 if (ambiguousYear[0] || tzTimeType != UTZFMT_TIME_TYPE_UNKNOWN) // If this is true then the two-digit year == the default start year
2240 {
2241 // We need a copy of the fields, and we need to avoid triggering a call to
2242 // complete(), which will recalculate the fields. Since we can't access
2243 // the fields[] array in Calendar, we clone the entire object. This will
2244 // stop working if Calendar.clone() is ever rewritten to call complete().
2245 Calendar *copy;
2246 if (ambiguousYear[0]) {
2247 copy = cal.clone();
2248 // Check for failed cloning.
2249 if (copy == NULL) {
2250 status = U_MEMORY_ALLOCATION_ERROR;
2251 goto ExitParse;
2252 }
2253 UDate parsedDate = copy->getTime(status);
2254 // {sfb} check internalGetDefaultCenturyStart
2255 if (fHaveDefaultCentury && (parsedDate < fDefaultCenturyStart)) {
2256 // We can't use add here because that does a complete() first.
2257 cal.set(UCAL_YEAR, fDefaultCenturyStartYear + 100);
2258 }
2259 delete copy;
2260 }
2261
2262 if (tzTimeType != UTZFMT_TIME_TYPE_UNKNOWN) {
2263 copy = cal.clone();
2264 // Check for failed cloning.
2265 if (copy == NULL) {
2266 status = U_MEMORY_ALLOCATION_ERROR;
2267 goto ExitParse;
2268 }
2269 const TimeZone & tz = cal.getTimeZone();
2270 BasicTimeZone *btz = NULL;
2271
2272 if (dynamic_cast<const OlsonTimeZone *>(&tz) != NULL
2273 || dynamic_cast<const SimpleTimeZone *>(&tz) != NULL
2274 || dynamic_cast<const RuleBasedTimeZone *>(&tz) != NULL
2275 || dynamic_cast<const VTimeZone *>(&tz) != NULL) {
2276 btz = (BasicTimeZone*)&tz;
2277 }
2278
2279 // Get local millis
2280 copy->set(UCAL_ZONE_OFFSET, 0);
2281 copy->set(UCAL_DST_OFFSET, 0);
2282 UDate localMillis = copy->getTime(status);
2283
2284 // Make sure parsed time zone type (Standard or Daylight)
2285 // matches the rule used by the parsed time zone.
2286 int32_t raw, dst;
2287 if (btz != NULL) {
2288 if (tzTimeType == UTZFMT_TIME_TYPE_STANDARD) {
2289 btz->getOffsetFromLocal(localMillis,
2290 BasicTimeZone::kStandard, BasicTimeZone::kStandard, raw, dst, status);
2291 } else {
2292 btz->getOffsetFromLocal(localMillis,
2293 BasicTimeZone::kDaylight, BasicTimeZone::kDaylight, raw, dst, status);
2294 }
2295 } else {
2296 // No good way to resolve ambiguous time at transition,
2297 // but following code work in most case.
2298 tz.getOffset(localMillis, TRUE, raw, dst, status);
2299 }
2300
2301 // Now, compare the results with parsed type, either standard or daylight saving time
2302 int32_t resolvedSavings = dst;
2303 if (tzTimeType == UTZFMT_TIME_TYPE_STANDARD) {
2304 if (dst != 0) {
2305 // Override DST_OFFSET = 0 in the result calendar
2306 resolvedSavings = 0;
2307 }
2308 } else { // tztype == TZTYPE_DST
2309 if (dst == 0) {
2310 if (btz != NULL) {
2311 UDate time = localMillis + raw;
2312 // We use the nearest daylight saving time rule.
2313 TimeZoneTransition beforeTrs, afterTrs;
2314 UDate beforeT = time, afterT = time;
2315 int32_t beforeSav = 0, afterSav = 0;
2316 UBool beforeTrsAvail, afterTrsAvail;
2317
2318 // Search for DST rule before or on the time
2319 while (TRUE) {
2320 beforeTrsAvail = btz->getPreviousTransition(beforeT, TRUE, beforeTrs);
2321 if (!beforeTrsAvail) {
2322 break;
2323 }
2324 beforeT = beforeTrs.getTime() - 1;
2325 beforeSav = beforeTrs.getFrom()->getDSTSavings();
2326 if (beforeSav != 0) {
2327 break;
2328 }
2329 }
2330
2331 // Search for DST rule after the time
2332 while (TRUE) {
2333 afterTrsAvail = btz->getNextTransition(afterT, FALSE, afterTrs);
2334 if (!afterTrsAvail) {
2335 break;
2336 }
2337 afterT = afterTrs.getTime();
2338 afterSav = afterTrs.getTo()->getDSTSavings();
2339 if (afterSav != 0) {
2340 break;
2341 }
2342 }
2343
2344 if (beforeTrsAvail && afterTrsAvail) {
2345 if (time - beforeT > afterT - time) {
2346 resolvedSavings = afterSav;
2347 } else {
2348 resolvedSavings = beforeSav;
2349 }
2350 } else if (beforeTrsAvail && beforeSav != 0) {
2351 resolvedSavings = beforeSav;
2352 } else if (afterTrsAvail && afterSav != 0) {
2353 resolvedSavings = afterSav;
2354 } else {
2355 resolvedSavings = btz->getDSTSavings();
2356 }
2357 } else {
2358 resolvedSavings = tz.getDSTSavings();
2359 }
2360 if (resolvedSavings == 0) {
2361 // final fallback
2362 resolvedSavings = U_MILLIS_PER_HOUR;
2363 }
2364 }
2365 }
2366 cal.set(UCAL_ZONE_OFFSET, raw);
2367 cal.set(UCAL_DST_OFFSET, resolvedSavings);
2368 delete copy;
2369 }
2370 }
2371 ExitParse:
2372 // Set the parsed result if local calendar is used
2373 // instead of the input calendar
2374 if (U_SUCCESS(status) && workCal != &cal) {
2375 cal.setTimeZone(workCal->getTimeZone());
2376 cal.setTime(workCal->getTime(status), status);
2377 }
2378
2379 if (numericLeapMonthFormatter != NULL) {
2380 delete numericLeapMonthFormatter;
2381 }
2382 if (calClone != NULL) {
2383 delete calClone;
2384 }
2385
2386 // If any Calendar calls failed, we pretend that we
2387 // couldn't parse the string, when in reality this isn't quite accurate--
2388 // we did parse it; the Calendar calls just failed.
2389 if (U_FAILURE(status)) {
2390 parsePos.setErrorIndex(pos);
2391 parsePos.setIndex(start);
2392 }
2393 }
2394
2395 //----------------------------------------------------------------------
2396
2397 static int32_t
2398 matchStringWithOptionalDot(const UnicodeString &text,
2399 int32_t index,
2400 const UnicodeString &data);
2401
2402 int32_t SimpleDateFormat::matchQuarterString(const UnicodeString& text,
2403 int32_t start,
2404 UCalendarDateFields field,
2405 const UnicodeString* data,
2406 int32_t dataCount,
2407 Calendar& cal) const
2408 {
2409 int32_t i = 0;
2410 int32_t count = dataCount;
2411
2412 // There may be multiple strings in the data[] array which begin with
2413 // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2414 // We keep track of the longest match, and return that. Note that this
2415 // unfortunately requires us to test all array elements.
2416 int32_t bestMatchLength = 0, bestMatch = -1;
2417 UnicodeString bestMatchName;
2418
2419 for (; i < count; ++i) {
2420 int32_t matchLength = 0;
2421 if ((matchLength = matchStringWithOptionalDot(text, start, data[i])) > bestMatchLength) {
2422 bestMatchLength = matchLength;
2423 bestMatch = i;
2424 }
2425 }
2426
2427 if (bestMatch >= 0) {
2428 cal.set(field, bestMatch * 3);
2429 return start + bestMatchLength;
2430 }
2431
2432 return -start;
2433 }
2434
2435 //----------------------------------------------------------------------
2436 #define IS_BIDI_MARK(c) (c==0x200E || c==0x200F || c==0x061C)
2437
2438 UBool SimpleDateFormat::matchLiterals(const UnicodeString &pattern,
2439 int32_t &patternOffset,
2440 const UnicodeString &text,
2441 int32_t &textOffset,
2442 UBool whitespaceLenient,
2443 UBool partialMatchLenient,
2444 UBool oldLeniency)
2445 {
2446 UBool inQuote = FALSE;
2447 UnicodeString literal;
2448 int32_t i = patternOffset;
2449
2450 // scan pattern looking for contiguous literal characters
2451 for ( ; i < pattern.length(); i += 1) {
2452 UChar ch = pattern.charAt(i);
2453
2454 if (!inQuote && isSyntaxChar(ch)) {
2455 break;
2456 }
2457
2458 if (ch == QUOTE) {
2459 // Match a quote literal ('') inside OR outside of quotes
2460 if ((i + 1) < pattern.length() && pattern.charAt(i + 1) == QUOTE) {
2461 i += 1;
2462 } else {
2463 inQuote = !inQuote;
2464 continue;
2465 }
2466 }
2467
2468 if (!IS_BIDI_MARK(ch)) {
2469 literal += ch;
2470 }
2471 }
2472
2473 // at this point, literal contains the pattern literal text (without bidi marks)
2474 // and i is the index of the next non-literal pattern character.
2475 int32_t p;
2476 int32_t t = textOffset;
2477
2478 if (whitespaceLenient) {
2479 // trim leading, trailing whitespace from the pattern literal
2480 literal.trim();
2481
2482 // ignore any leading whitespace (or bidi marks) in the text
2483 while (t < text.length()) {
2484 UChar ch = text.charAt(t);
2485 if (!u_isWhitespace(ch) && !IS_BIDI_MARK(ch)) {
2486 break;
2487 }
2488 t += 1;
2489 }
2490 }
2491
2492 // Get ignorables, move up here
2493 const UnicodeSet *ignorables = NULL;
2494 UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(pattern.charAt(i));
2495 if (patternCharIndex != UDAT_FIELD_COUNT) {
2496 ignorables = SimpleDateFormatStaticSets::getIgnorables(patternCharIndex);
2497 }
2498
2499 for (p = 0; p < literal.length() && t < text.length();) {
2500 UBool needWhitespace = FALSE;
2501
2502 // Skip any whitespace at current position in pattern,
2503 // but remember whether we found whitespace in the pattern
2504 // (we already deleted any bidi marks in the pattern).
2505 while (p < literal.length() && PatternProps::isWhiteSpace(literal.charAt(p))) {
2506 needWhitespace = TRUE;
2507 p += 1;
2508 }
2509
2510 // If the pattern has whitespace at this point, skip it in text as well
2511 // (if the text does not have any, that may be an error for strict parsing)
2512 if (needWhitespace) {
2513 UBool whitespaceInText = FALSE;
2514
2515 // Skip any whitespace (or bidi marks) at current position in text,
2516 // but remember whether we found whitespace in the text at this point.
2517 while (t < text.length()) {
2518 UChar tch = text.charAt(t);
2519 if (u_isUWhiteSpace(tch) || PatternProps::isWhiteSpace(tch)) {
2520 whitespaceInText = TRUE;
2521 } else if (!IS_BIDI_MARK(tch)) {
2522 break;
2523 }
2524
2525 t += 1;
2526 }
2527
2528 // TODO: should we require internal spaces
2529 // in lenient mode? (There won't be any
2530 // leading or trailing spaces)
2531 if (!whitespaceLenient && !whitespaceInText) {
2532 // didn't find matching whitespace:
2533 // an error in strict mode
2534 return FALSE;
2535 }
2536
2537 // In strict mode, this run of whitespace
2538 // may have been at the end.
2539 if (p >= literal.length()) {
2540 break;
2541 }
2542 } else {
2543 // Still need to skip any bidi marks in the text
2544 while (t < text.length() && IS_BIDI_MARK(text.charAt(t))) {
2545 ++t;
2546 }
2547 }
2548 if (t >= text.length() || literal.charAt(p) != text.charAt(t)) {
2549 // Ran out of text, or found a non-matching character:
2550 // OK in lenient mode, an error in strict mode.
2551 if (whitespaceLenient) {
2552 if (t == textOffset && text.charAt(t) == 0x2e &&
2553 isAfterNonNumericField(pattern, patternOffset)) {
2554 // Lenient mode and the literal input text begins with a "." and
2555 // we are after a non-numeric field: We skip the "."
2556 ++t;
2557 continue; // Do not update p.
2558 }
2559 // if it is actual whitespace and we're whitespace lenient it's OK
2560
2561 UChar wsc = text.charAt(t);
2562 if(PatternProps::isWhiteSpace(wsc)) {
2563 // Lenient mode and it's just whitespace we skip it
2564 ++t;
2565 continue; // Do not update p.
2566 }
2567 }
2568 // hack around oldleniency being a bit of a catch-all bucket and we're just adding support specifically for paritial matches
2569 // This fix is for http://bugs.icu-project.org/trac/ticket/10855 and adds "&& oldLeniency"
2570 //if(partialMatchLenient && oldLeniency) {
2571 // However this causes problems for Apple, see <rdar://problem/20692829> regressions in Chinese date parsing
2572 // We don't want to go back to just "if(partialMatchLenient)" as in ICU 53, that is too lenient for strict mode.
2573 // So if the pattern character is in the separator set, we allow the text character to be in that set or be an alpha char.
2574 if( partialMatchLenient && ( oldLeniency ||
2575 ( ignorables != NULL && ignorables->contains(literal.charAt(p)) && (ignorables->contains(text.charAt(t)) || u_isalpha(text.charAt(t))) ) )
2576 ) {
2577 break;
2578 }
2579
2580 return FALSE;
2581 }
2582 ++p;
2583 ++t;
2584 }
2585
2586 // At this point if we're in strict mode we have a complete match.
2587 // If we're in lenient mode we may have a partial match, or no
2588 // match at all.
2589 if (p <= 0) {
2590 // no match. Pretend it matched a run of whitespace
2591 // and ignorables in the text.
2592
2593 for (t = textOffset; t < text.length(); t += 1) {
2594 UChar ch = text.charAt(t);
2595
2596 if (!IS_BIDI_MARK(ch) && (ignorables == NULL || !ignorables->contains(ch))) {
2597 break;
2598 }
2599 }
2600 }
2601
2602 // if we get here, we've got a complete match.
2603 patternOffset = i - 1;
2604 textOffset = t;
2605
2606 return TRUE;
2607 }
2608
2609 //----------------------------------------------------------------------
2610
2611 int32_t SimpleDateFormat::matchString(const UnicodeString& text,
2612 int32_t start,
2613 UCalendarDateFields field,
2614 const UnicodeString* data,
2615 int32_t dataCount,
2616 const UnicodeString* monthPattern,
2617 Calendar& cal) const
2618 {
2619 int32_t i = 0;
2620 int32_t count = dataCount;
2621
2622 if (field == UCAL_DAY_OF_WEEK) i = 1;
2623
2624 // There may be multiple strings in the data[] array which begin with
2625 // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2626 // We keep track of the longest match, and return that. Note that this
2627 // unfortunately requires us to test all array elements.
2628 int32_t bestMatchLength = 0, bestMatch = -1;
2629 UnicodeString bestMatchName;
2630 int32_t isLeapMonth = 0;
2631
2632 for (; i < count; ++i) {
2633 int32_t matchLen = 0;
2634 if ((matchLen = matchStringWithOptionalDot(text, start, data[i])) > bestMatchLength) {
2635 bestMatch = i;
2636 bestMatchLength = matchLen;
2637 }
2638
2639 if (monthPattern != NULL) {
2640 UErrorCode status = U_ZERO_ERROR;
2641 UnicodeString leapMonthName;
2642 Formattable monthName((const UnicodeString&)(data[i]));
2643 MessageFormat::format(*monthPattern, &monthName, 1, leapMonthName, status);
2644 if (U_SUCCESS(status)) {
2645 if ((matchLen = matchStringWithOptionalDot(text, start, leapMonthName)) > bestMatchLength) {
2646 bestMatch = i;
2647 bestMatchLength = matchLen;
2648 isLeapMonth = 1;
2649 }
2650 }
2651 }
2652 }
2653
2654 if (bestMatch >= 0) {
2655 if (field < UCAL_FIELD_COUNT) {
2656 // Adjustment for Hebrew Calendar month Adar II
2657 if (!strcmp(cal.getType(),"hebrew") && field==UCAL_MONTH && bestMatch==13) {
2658 cal.set(field,6);
2659 } else {
2660 if (field == UCAL_YEAR) {
2661 bestMatch++; // only get here for cyclic year names, which match 1-based years 1-60
2662 }
2663 cal.set(field, bestMatch);
2664 }
2665 if (monthPattern != NULL) {
2666 cal.set(UCAL_IS_LEAP_MONTH, isLeapMonth);
2667 }
2668 }
2669
2670 return start + bestMatchLength;
2671 }
2672
2673 return -start;
2674 }
2675
2676 static int32_t
2677 matchStringWithOptionalDot(const UnicodeString &text,
2678 int32_t index,
2679 const UnicodeString &data) {
2680 UErrorCode sts = U_ZERO_ERROR;
2681 int32_t matchLenText = 0;
2682 int32_t matchLenData = 0;
2683
2684 u_caseInsensitivePrefixMatch(text.getBuffer() + index, text.length() - index,
2685 data.getBuffer(), data.length(),
2686 0 /* default case option */,
2687 &matchLenText, &matchLenData,
2688 &sts);
2689 U_ASSERT (U_SUCCESS(sts));
2690
2691 if (matchLenData == data.length() /* normal match */
2692 || (data.charAt(data.length() - 1) == 0x2e
2693 && matchLenData == data.length() - 1 /* match without trailing dot */)) {
2694 return matchLenText;
2695 }
2696
2697 return 0;
2698 }
2699
2700 //----------------------------------------------------------------------
2701
2702 void
2703 SimpleDateFormat::set2DigitYearStart(UDate d, UErrorCode& status)
2704 {
2705 parseAmbiguousDatesAsAfter(d, status);
2706 }
2707
2708 /**
2709 * Private member function that converts the parsed date strings into
2710 * timeFields. Returns -start (for ParsePosition) if failed.
2711 */
2712 int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UChar ch, int32_t count,
2713 UBool obeyCount, UBool allowNegative, UBool ambiguousYear[], int32_t& saveHebrewMonth, Calendar& cal,
2714 int32_t patLoc, MessageFormat * numericLeapMonthFormatter, UTimeZoneFormatTimeType *tzTimeType, SimpleDateFormatMutableNFs &mutableNFs) const
2715 {
2716 Formattable number;
2717 int32_t value = 0;
2718 int32_t i;
2719 int32_t ps = 0;
2720 UErrorCode status = U_ZERO_ERROR;
2721 ParsePosition pos(0);
2722 UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
2723 NumberFormat *currentNumberFormat;
2724 UnicodeString temp;
2725 int32_t tzParseOptions = (isLenient())? UTZFMT_PARSE_OPTION_ALL_STYLES: UTZFMT_PARSE_OPTION_NONE;
2726 UBool gotNumber = FALSE;
2727
2728 #if defined (U_DEBUG_CAL)
2729 //fprintf(stderr, "%s:%d - [%c] st=%d \n", __FILE__, __LINE__, (char) ch, start);
2730 #endif
2731
2732 if (patternCharIndex == UDAT_FIELD_COUNT) {
2733 return -start;
2734 }
2735
2736 currentNumberFormat = mutableNFs.get(getNumberFormatByIndex(patternCharIndex));
2737 if (currentNumberFormat == NULL) {
2738 return -start;
2739 }
2740 UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex]; // UCAL_FIELD_COUNT if irrelevant
2741 UnicodeString hebr("hebr", 4, US_INV);
2742
2743 if (numericLeapMonthFormatter != NULL) {
2744 numericLeapMonthFormatter->setFormats((const Format **)&currentNumberFormat, 1);
2745 }
2746 UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0 || uprv_strcmp(cal.getType(),"dangi") == 0);
2747
2748 // If there are any spaces here, skip over them. If we hit the end
2749 // of the string, then fail.
2750 for (;;) {
2751 if (start >= text.length()) {
2752 return -start;
2753 }
2754 UChar32 c = text.char32At(start);
2755 if (!u_isUWhiteSpace(c) /*||*/ && !PatternProps::isWhiteSpace(c)) {
2756 break;
2757 }
2758 start += U16_LENGTH(c);
2759 }
2760 pos.setIndex(start);
2761
2762 // We handle a few special cases here where we need to parse
2763 // a number value. We handle further, more generic cases below. We need
2764 // to handle some of them here because some fields require extra processing on
2765 // the parsed value.
2766 if (patternCharIndex == UDAT_HOUR_OF_DAY1_FIELD || // k
2767 patternCharIndex == UDAT_HOUR_OF_DAY0_FIELD || // H
2768 patternCharIndex == UDAT_HOUR1_FIELD || // h
2769 patternCharIndex == UDAT_HOUR0_FIELD || // K
2770 (patternCharIndex == UDAT_DOW_LOCAL_FIELD && count <= 2) || // e
2771 (patternCharIndex == UDAT_STANDALONE_DAY_FIELD && count <= 2) || // c
2772 (patternCharIndex == UDAT_MONTH_FIELD && count <= 2) || // M
2773 (patternCharIndex == UDAT_STANDALONE_MONTH_FIELD && count <= 2) || // L
2774 (patternCharIndex == UDAT_QUARTER_FIELD && count <= 2) || // Q
2775 (patternCharIndex == UDAT_STANDALONE_QUARTER_FIELD && count <= 2) || // q
2776 patternCharIndex == UDAT_YEAR_FIELD || // y
2777 patternCharIndex == UDAT_YEAR_WOY_FIELD || // Y
2778 patternCharIndex == UDAT_YEAR_NAME_FIELD || // U (falls back to numeric)
2779 (patternCharIndex == UDAT_ERA_FIELD && isChineseCalendar) || // G
2780 patternCharIndex == UDAT_FRACTIONAL_SECOND_FIELD) // S
2781 {
2782 int32_t parseStart = pos.getIndex();
2783 // It would be good to unify this with the obeyCount logic below,
2784 // but that's going to be difficult.
2785 const UnicodeString* src;
2786
2787 UBool parsedNumericLeapMonth = FALSE;
2788 if (numericLeapMonthFormatter != NULL && (patternCharIndex == UDAT_MONTH_FIELD || patternCharIndex == UDAT_STANDALONE_MONTH_FIELD)) {
2789 int32_t argCount;
2790 Formattable * args = numericLeapMonthFormatter->parse(text, pos, argCount);
2791 if (args != NULL && argCount == 1 && pos.getIndex() > parseStart && args[0].isNumeric()) {
2792 parsedNumericLeapMonth = TRUE;
2793 number.setLong(args[0].getLong());
2794 cal.set(UCAL_IS_LEAP_MONTH, 1);
2795 delete[] args;
2796 } else {
2797 pos.setIndex(parseStart);
2798 cal.set(UCAL_IS_LEAP_MONTH, 0);
2799 }
2800 }
2801
2802 if (!parsedNumericLeapMonth) {
2803 if (obeyCount) {
2804 if ((start+count) > text.length()) {
2805 return -start;
2806 }
2807
2808 text.extractBetween(0, start + count, temp);
2809 src = &temp;
2810 } else {
2811 src = &text;
2812 }
2813
2814 parseInt(*src, number, pos, allowNegative,currentNumberFormat);
2815 }
2816
2817 int32_t txtLoc = pos.getIndex();
2818
2819 if (txtLoc > parseStart) {
2820 value = number.getLong();
2821 gotNumber = TRUE;
2822
2823 // suffix processing
2824 if (value < 0 ) {
2825 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, TRUE);
2826 if (txtLoc != pos.getIndex()) {
2827 value *= -1;
2828 }
2829 }
2830 else {
2831 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, FALSE);
2832 }
2833
2834 // Check the range of the value
2835 if (!getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) {
2836 int32_t bias = gFieldRangeBias[patternCharIndex];
2837 if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
2838 return -start;
2839 }
2840 } else {
2841 int32_t bias = gFieldRangeBiasLenient[patternCharIndex];
2842 if (bias >= 0 && (value > cal.getMaximum(field) + bias)) {
2843 return -start;
2844 }
2845 }
2846
2847 pos.setIndex(txtLoc);
2848 }
2849 }
2850
2851 // Make sure that we got a number if
2852 // we want one, and didn't get one
2853 // if we don't want one.
2854 switch (patternCharIndex) {
2855 case UDAT_HOUR_OF_DAY1_FIELD:
2856 case UDAT_HOUR_OF_DAY0_FIELD:
2857 case UDAT_HOUR1_FIELD:
2858 case UDAT_HOUR0_FIELD:
2859 // special range check for hours:
2860 if (value < 0 || value > 24) {
2861 return -start;
2862 }
2863
2864 // fall through to gotNumber check
2865
2866 case UDAT_YEAR_FIELD:
2867 case UDAT_YEAR_WOY_FIELD:
2868 case UDAT_FRACTIONAL_SECOND_FIELD:
2869 // these must be a number
2870 if (! gotNumber) {
2871 return -start;
2872 }
2873
2874 break;
2875
2876 default:
2877 // we check the rest of the fields below.
2878 break;
2879 }
2880
2881 switch (patternCharIndex) {
2882 case UDAT_ERA_FIELD:
2883 if (isChineseCalendar) {
2884 if (!gotNumber) {
2885 return -start;
2886 }
2887 cal.set(UCAL_ERA, value);
2888 return pos.getIndex();
2889 }
2890 if (count == 5) {
2891 ps = matchString(text, start, UCAL_ERA, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount, NULL, cal);
2892 } else if (count == 4) {
2893 ps = matchString(text, start, UCAL_ERA, fSymbols->fEraNames, fSymbols->fEraNamesCount, NULL, cal);
2894 } else {
2895 ps = matchString(text, start, UCAL_ERA, fSymbols->fEras, fSymbols->fErasCount, NULL, cal);
2896 }
2897
2898 // check return position, if it equals -start, then matchString error
2899 // special case the return code so we don't necessarily fail out until we
2900 // verify no year information also
2901 if (ps == -start)
2902 ps--;
2903
2904 return ps;
2905
2906 case UDAT_YEAR_FIELD:
2907 // If there are 3 or more YEAR pattern characters, this indicates
2908 // that the year value is to be treated literally, without any
2909 // two-digit year adjustments (e.g., from "01" to 2001). Otherwise
2910 // we made adjustments to place the 2-digit year in the proper
2911 // century, for parsed strings from "00" to "99". Any other string
2912 // is treated literally: "2250", "-1", "1", "002".
2913 if (fDateOverride.compare(hebr)==0 && value < 1000) {
2914 value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2915 } else if ((pos.getIndex() - start) == 2 && !isChineseCalendar
2916 && u_isdigit(text.charAt(start))
2917 && u_isdigit(text.charAt(start+1)))
2918 {
2919 // only adjust year for patterns less than 3.
2920 if(count < 3) {
2921 // Assume for example that the defaultCenturyStart is 6/18/1903.
2922 // This means that two-digit years will be forced into the range
2923 // 6/18/1903 to 6/17/2003. As a result, years 00, 01, and 02
2924 // correspond to 2000, 2001, and 2002. Years 04, 05, etc. correspond
2925 // to 1904, 1905, etc. If the year is 03, then it is 2003 if the
2926 // other fields specify a date before 6/18, or 1903 if they specify a
2927 // date afterwards. As a result, 03 is an ambiguous year. All other
2928 // two-digit years are unambiguous.
2929 if(fHaveDefaultCentury) { // check if this formatter even has a pivot year
2930 int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2931 ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2932 value += (fDefaultCenturyStartYear/100)*100 +
2933 (value < ambiguousTwoDigitYear ? 100 : 0);
2934 }
2935 }
2936 }
2937 cal.set(UCAL_YEAR, value);
2938
2939 // Delayed checking for adjustment of Hebrew month numbers in non-leap years.
2940 if (saveHebrewMonth >= 0) {
2941 HebrewCalendar *hc = (HebrewCalendar*)&cal;
2942 if (!hc->isLeapYear(value) && saveHebrewMonth >= 6) {
2943 cal.set(UCAL_MONTH,saveHebrewMonth);
2944 } else {
2945 cal.set(UCAL_MONTH,saveHebrewMonth-1);
2946 }
2947 saveHebrewMonth = -1;
2948 }
2949 return pos.getIndex();
2950
2951 case UDAT_YEAR_WOY_FIELD:
2952 // Comment is the same as for UDAT_Year_FIELDs - look above
2953 if (fDateOverride.compare(hebr)==0 && value < 1000) {
2954 value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2955 } else if ((pos.getIndex() - start) == 2
2956 && u_isdigit(text.charAt(start))
2957 && u_isdigit(text.charAt(start+1))
2958 && fHaveDefaultCentury )
2959 {
2960 int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2961 ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2962 value += (fDefaultCenturyStartYear/100)*100 +
2963 (value < ambiguousTwoDigitYear ? 100 : 0);
2964 }
2965 cal.set(UCAL_YEAR_WOY, value);
2966 return pos.getIndex();
2967
2968 case UDAT_YEAR_NAME_FIELD:
2969 if (fSymbols->fShortYearNames != NULL) {
2970 int32_t newStart = matchString(text, start, UCAL_YEAR, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount, NULL, cal);
2971 if (newStart > 0) {
2972 return newStart;
2973 }
2974 }
2975 if (gotNumber && (getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC,status) || value > fSymbols->fShortYearNamesCount)) {
2976 cal.set(UCAL_YEAR, value);
2977 return pos.getIndex();
2978 }
2979 return -start;
2980
2981 case UDAT_MONTH_FIELD:
2982 case UDAT_STANDALONE_MONTH_FIELD:
2983 if (gotNumber) // i.e., M or MM.
2984 {
2985 // When parsing month numbers from the Hebrew Calendar, we might need to adjust the month depending on whether
2986 // 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
2987 // the year is parsed.
2988 if (!strcmp(cal.getType(),"hebrew")) {
2989 HebrewCalendar *hc = (HebrewCalendar*)&cal;
2990 if (cal.isSet(UCAL_YEAR)) {
2991 UErrorCode status = U_ZERO_ERROR;
2992 if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
2993 cal.set(UCAL_MONTH, value);
2994 } else {
2995 cal.set(UCAL_MONTH, value - 1);
2996 }
2997 } else {
2998 saveHebrewMonth = value;
2999 }
3000 } else {
3001 // Don't want to parse the month if it is a string
3002 // while pattern uses numeric style: M/MM, L/LL
3003 // [We computed 'value' above.]
3004 cal.set(UCAL_MONTH, value - 1);
3005 }
3006 return pos.getIndex();
3007 } else {
3008 // count >= 3 // i.e., MMM/MMMM, LLL/LLLL
3009 // Want to be able to parse both short and long forms.
3010 // Try count == 4 first:
3011 UnicodeString * wideMonthPat = NULL;
3012 UnicodeString * shortMonthPat = NULL;
3013 if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
3014 if (patternCharIndex==UDAT_MONTH_FIELD) {
3015 wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide];
3016 shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev];
3017 } else {
3018 wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide];
3019 shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev];
3020 }
3021 }
3022 int32_t newStart = 0;
3023 if (patternCharIndex==UDAT_MONTH_FIELD) {
3024 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3025 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fMonths, fSymbols->fMonthsCount, wideMonthPat, cal); // try MMMM
3026 if (newStart > 0) {
3027 return newStart;
3028 }
3029 }
3030 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3031 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fShortMonths, fSymbols->fShortMonthsCount, shortMonthPat, cal); // try MMM
3032 }
3033 } else {
3034 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3035 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount, wideMonthPat, cal); // try LLLL
3036 if (newStart > 0) {
3037 return newStart;
3038 }
3039 }
3040 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3041 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount, shortMonthPat, cal); // try LLL
3042 }
3043 }
3044 if (newStart > 0 || !getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status)) // currently we do not try to parse MMMMM/LLLLL: #8860
3045 return newStart;
3046 // else we allowing parsing as number, below
3047 }
3048 break;
3049
3050 case UDAT_HOUR_OF_DAY1_FIELD:
3051 // [We computed 'value' above.]
3052 if (value == cal.getMaximum(UCAL_HOUR_OF_DAY) + 1)
3053 value = 0;
3054
3055 // fall through to set field
3056
3057 case UDAT_HOUR_OF_DAY0_FIELD:
3058 cal.set(UCAL_HOUR_OF_DAY, value);
3059 return pos.getIndex();
3060
3061 case UDAT_FRACTIONAL_SECOND_FIELD:
3062 // Fractional seconds left-justify
3063 i = pos.getIndex() - start;
3064 if (i < 3) {
3065 while (i < 3) {
3066 value *= 10;
3067 i++;
3068 }
3069 } else {
3070 int32_t a = 1;
3071 while (i > 3) {
3072 a *= 10;
3073 i--;
3074 }
3075 value /= a;
3076 }
3077 cal.set(UCAL_MILLISECOND, value);
3078 return pos.getIndex();
3079
3080 case UDAT_DOW_LOCAL_FIELD:
3081 if (gotNumber) // i.e., e or ee
3082 {
3083 // [We computed 'value' above.]
3084 cal.set(UCAL_DOW_LOCAL, value);
3085 return pos.getIndex();
3086 }
3087 // else for eee-eeeee fall through to handling of EEE-EEEEE
3088 // fall through, do not break here
3089 case UDAT_DAY_OF_WEEK_FIELD:
3090 {
3091 // Want to be able to parse both short and long forms.
3092 // Try count == 4 (EEEE) wide first:
3093 int32_t newStart = 0;
3094 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3095 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3096 fSymbols->fWeekdays, fSymbols->fWeekdaysCount, NULL, cal)) > 0)
3097 return newStart;
3098 }
3099 // EEEE wide failed, now try EEE abbreviated
3100 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3101 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3102 fSymbols->fShortWeekdays, fSymbols->fShortWeekdaysCount, NULL, cal)) > 0)
3103 return newStart;
3104 }
3105 // EEE abbreviated failed, now try EEEEEE short
3106 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) {
3107 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3108 fSymbols->fShorterWeekdays, fSymbols->fShorterWeekdaysCount, NULL, cal)) > 0)
3109 return newStart;
3110 }
3111 // EEEEEE short failed, now try EEEEE narrow
3112 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 5) {
3113 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3114 fSymbols->fNarrowWeekdays, fSymbols->fNarrowWeekdaysCount, NULL, cal)) > 0)
3115 return newStart;
3116 }
3117 if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status) || patternCharIndex == UDAT_DAY_OF_WEEK_FIELD)
3118 return newStart;
3119 // else we allowing parsing as number, below
3120 }
3121 break;
3122
3123 case UDAT_STANDALONE_DAY_FIELD:
3124 {
3125 if (gotNumber) // c or cc
3126 {
3127 // [We computed 'value' above.]
3128 cal.set(UCAL_DOW_LOCAL, value);
3129 return pos.getIndex();
3130 }
3131 // Want to be able to parse both short and long forms.
3132 // Try count == 4 (cccc) first:
3133 int32_t newStart = 0;
3134 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3135 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3136 fSymbols->fStandaloneWeekdays, fSymbols->fStandaloneWeekdaysCount, NULL, cal)) > 0)
3137 return newStart;
3138 }
3139 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3140 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3141 fSymbols->fStandaloneShortWeekdays, fSymbols->fStandaloneShortWeekdaysCount, NULL, cal)) > 0)
3142 return newStart;
3143 }
3144 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) {
3145 if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
3146 fSymbols->fStandaloneShorterWeekdays, fSymbols->fStandaloneShorterWeekdaysCount, NULL, cal)) > 0)
3147 return newStart;
3148 }
3149 if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
3150 return newStart;
3151 // else we allowing parsing as number, below
3152 }
3153 break;
3154
3155 case UDAT_AM_PM_FIELD:
3156 {
3157 // optionally try both wide/abbrev and narrow forms
3158 int32_t newStart = 0;
3159 // try wide/abbrev
3160 if( getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count < 5 ) {
3161 if ((newStart = matchString(text, start, UCAL_AM_PM, fSymbols->fAmPms, fSymbols->fAmPmsCount, NULL, cal)) > 0) {
3162 return newStart;
3163 }
3164 }
3165 // try narrow
3166 if( getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count >= 5 ) {
3167 if ((newStart = matchString(text, start, UCAL_AM_PM, fSymbols->fNarrowAmPms, fSymbols->fNarrowAmPmsCount, NULL, cal)) > 0) {
3168 return newStart;
3169 }
3170 }
3171 // no matches for given options
3172 return -start;
3173 }
3174
3175 case UDAT_HOUR1_FIELD:
3176 // [We computed 'value' above.]
3177 if (value == cal.getLeastMaximum(UCAL_HOUR)+1)
3178 value = 0;
3179
3180 // fall through to set field
3181
3182 case UDAT_HOUR0_FIELD:
3183 cal.set(UCAL_HOUR, value);
3184 return pos.getIndex();
3185
3186 case UDAT_QUARTER_FIELD:
3187 if (gotNumber) // i.e., Q or QQ.
3188 {
3189 // Don't want to parse the month if it is a string
3190 // while pattern uses numeric style: Q or QQ.
3191 // [We computed 'value' above.]
3192 cal.set(UCAL_MONTH, (value - 1) * 3);
3193 return pos.getIndex();
3194 } else {
3195 // count >= 3 // i.e., QQQ or QQQQ
3196 // Want to be able to parse both short and long forms.
3197 // Try count == 4 first:
3198 int32_t newStart = 0;
3199
3200 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3201 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3202 fSymbols->fQuarters, fSymbols->fQuartersCount, cal)) > 0)
3203 return newStart;
3204 }
3205 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3206 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3207 fSymbols->fShortQuarters, fSymbols->fShortQuartersCount, cal)) > 0)
3208 return newStart;
3209 }
3210 if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
3211 return newStart;
3212 // else we allowing parsing as number, below
3213 if(!getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status))
3214 return -start;
3215 }
3216 break;
3217
3218 case UDAT_STANDALONE_QUARTER_FIELD:
3219 if (gotNumber) // i.e., q or qq.
3220 {
3221 // Don't want to parse the month if it is a string
3222 // while pattern uses numeric style: q or q.
3223 // [We computed 'value' above.]
3224 cal.set(UCAL_MONTH, (value - 1) * 3);
3225 return pos.getIndex();
3226 } else {
3227 // count >= 3 // i.e., qqq or qqqq
3228 // Want to be able to parse both short and long forms.
3229 // Try count == 4 first:
3230 int32_t newStart = 0;
3231
3232 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
3233 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3234 fSymbols->fStandaloneQuarters, fSymbols->fStandaloneQuartersCount, cal)) > 0)
3235 return newStart;
3236 }
3237 if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
3238 if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
3239 fSymbols->fStandaloneShortQuarters, fSymbols->fStandaloneShortQuartersCount, cal)) > 0)
3240 return newStart;
3241 }
3242 if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
3243 return newStart;
3244 // else we allowing parsing as number, below
3245 if(!getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status))
3246 return -start;
3247 }
3248 break;
3249
3250 case UDAT_TIMEZONE_FIELD: // 'z'
3251 {
3252 UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_SPECIFIC_SHORT : UTZFMT_STYLE_SPECIFIC_LONG;
3253 TimeZone *tz = tzFormat()->parse(style, text, pos, tzParseOptions, tzTimeType);
3254 if (tz != NULL) {
3255 cal.adoptTimeZone(tz);
3256 return pos.getIndex();
3257 }
3258 }
3259 break;
3260 case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
3261 {
3262 UTimeZoneFormatStyle style = (count < 4) ?
3263 UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL : ((count == 5) ? UTZFMT_STYLE_ISO_EXTENDED_FULL: UTZFMT_STYLE_LOCALIZED_GMT);
3264 TimeZone *tz = tzFormat()->parse(style, text, pos, tzTimeType);
3265 if (tz != NULL) {
3266 cal.adoptTimeZone(tz);
3267 return pos.getIndex();
3268 }
3269 return -start;
3270 }
3271 case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
3272 {
3273 UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_GENERIC_SHORT : UTZFMT_STYLE_GENERIC_LONG;
3274 TimeZone *tz = tzFormat()->parse(style, text, pos, tzParseOptions, tzTimeType);
3275 if (tz != NULL) {
3276 cal.adoptTimeZone(tz);
3277 return pos.getIndex();
3278 }
3279 return -start;
3280 }
3281 case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
3282 {
3283 UTimeZoneFormatStyle style;
3284 switch (count) {
3285 case 1:
3286 style = UTZFMT_STYLE_ZONE_ID_SHORT;
3287 break;
3288 case 2:
3289 style = UTZFMT_STYLE_ZONE_ID;
3290 break;
3291 case 3:
3292 style = UTZFMT_STYLE_EXEMPLAR_LOCATION;
3293 break;
3294 default:
3295 style = UTZFMT_STYLE_GENERIC_LOCATION;
3296 break;
3297 }
3298 TimeZone *tz = tzFormat()->parse(style, text, pos, tzTimeType);
3299 if (tz != NULL) {
3300 cal.adoptTimeZone(tz);
3301 return pos.getIndex();
3302 }
3303 return -start;
3304 }
3305 case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
3306 {
3307 UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_LOCALIZED_GMT_SHORT : UTZFMT_STYLE_LOCALIZED_GMT;
3308 TimeZone *tz = tzFormat()->parse(style, text, pos, tzTimeType);
3309 if (tz != NULL) {
3310 cal.adoptTimeZone(tz);
3311 return pos.getIndex();
3312 }
3313 return -start;
3314 }
3315 case UDAT_TIMEZONE_ISO_FIELD: // 'X'
3316 {
3317 UTimeZoneFormatStyle style;
3318 switch (count) {
3319 case 1:
3320 style = UTZFMT_STYLE_ISO_BASIC_SHORT;
3321 break;
3322 case 2:
3323 style = UTZFMT_STYLE_ISO_BASIC_FIXED;
3324 break;
3325 case 3:
3326 style = UTZFMT_STYLE_ISO_EXTENDED_FIXED;
3327 break;
3328 case 4:
3329 style = UTZFMT_STYLE_ISO_BASIC_FULL;
3330 break;
3331 default:
3332 style = UTZFMT_STYLE_ISO_EXTENDED_FULL;
3333 break;
3334 }
3335 TimeZone *tz = tzFormat()->parse(style, text, pos, tzTimeType);
3336 if (tz != NULL) {
3337 cal.adoptTimeZone(tz);
3338 return pos.getIndex();
3339 }
3340 return -start;
3341 }
3342 case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
3343 {
3344 UTimeZoneFormatStyle style;
3345 switch (count) {
3346 case 1:
3347 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT;
3348 break;
3349 case 2:
3350 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED;
3351 break;
3352 case 3:
3353 style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED;
3354 break;
3355 case 4:
3356 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL;
3357 break;
3358 default:
3359 style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL;
3360 break;
3361 }
3362 TimeZone *tz = tzFormat()->parse(style, text, pos, tzTimeType);
3363 if (tz != NULL) {
3364 cal.adoptTimeZone(tz);
3365 return pos.getIndex();
3366 }
3367 return -start;
3368 }
3369 case UDAT_TIME_SEPARATOR_FIELD: // ':'
3370 {
3371 static const UChar def_sep = DateFormatSymbols::DEFAULT_TIME_SEPARATOR;
3372 static const UChar alt_sep = DateFormatSymbols::ALTERNATE_TIME_SEPARATOR;
3373
3374 // Try matching a time separator.
3375 int32_t count = 1;
3376 UnicodeString data[3];
3377 fSymbols->getTimeSeparatorString(data[0]);
3378
3379 // Add the default, if different from the locale.
3380 if (data[0].compare(&def_sep, 1) != 0) {
3381 data[count++].setTo(def_sep);
3382 }
3383
3384 // If lenient, add also the alternate, if different from the locale.
3385 if (isLenient() && data[0].compare(&alt_sep, 1) != 0) {
3386 data[count++].setTo(alt_sep);
3387 }
3388
3389 return matchString(text, start, UCAL_FIELD_COUNT /* => nothing to set */, data, count, NULL, cal);
3390 }
3391
3392 default:
3393 // Handle "generic" fields
3394 // this is now handled below, outside the switch block
3395 break;
3396 }
3397 // Handle "generic" fields:
3398 // switch default case now handled here (outside switch block) to allow
3399 // parsing of some string fields as digits for lenient case
3400
3401 int32_t parseStart = pos.getIndex();
3402 const UnicodeString* src;
3403 if (obeyCount) {
3404 if ((start+count) > text.length()) {
3405 return -start;
3406 }
3407 text.extractBetween(0, start + count, temp);
3408 src = &temp;
3409 } else {
3410 src = &text;
3411 }
3412 parseInt(*src, number, pos, allowNegative,currentNumberFormat);
3413 if (pos.getIndex() != parseStart) {
3414 int32_t value = number.getLong();
3415
3416 // Don't need suffix processing here (as in number processing at the beginning of the function);
3417 // the new fields being handled as numeric values (month, weekdays, quarters) should not have suffixes.
3418
3419 // Check the range of the value
3420 if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status)) {
3421 int32_t bias = gFieldRangeBias[patternCharIndex];
3422 if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
3423 return -start;
3424 }
3425 } else {
3426 int32_t bias = gFieldRangeBiasLenient[patternCharIndex];
3427 if (bias >= 0 && (value > cal.getMaximum(field) + bias)) {
3428 return -start;
3429 }
3430 }
3431
3432 // For the following, need to repeat some of the "if (gotNumber)" code above:
3433 // UDAT_[STANDALONE_]MONTH_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_STANDALONE_DAY_FIELD,
3434 // UDAT_[STANDALONE_]QUARTER_FIELD
3435 switch (patternCharIndex) {
3436 case UDAT_MONTH_FIELD:
3437 // See notes under UDAT_MONTH_FIELD case above
3438 if (!strcmp(cal.getType(),"hebrew")) {
3439 HebrewCalendar *hc = (HebrewCalendar*)&cal;
3440 if (cal.isSet(UCAL_YEAR)) {
3441 UErrorCode status = U_ZERO_ERROR;
3442 if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
3443 cal.set(UCAL_MONTH, value);
3444 } else {
3445 cal.set(UCAL_MONTH, value - 1);
3446 }
3447 } else {
3448 saveHebrewMonth = value;
3449 }
3450 } else {
3451 cal.set(UCAL_MONTH, value - 1);
3452 }
3453 break;
3454 case UDAT_STANDALONE_MONTH_FIELD:
3455 cal.set(UCAL_MONTH, value - 1);
3456 break;
3457 case UDAT_DOW_LOCAL_FIELD:
3458 case UDAT_STANDALONE_DAY_FIELD:
3459 cal.set(UCAL_DOW_LOCAL, value);
3460 break;
3461 case UDAT_QUARTER_FIELD:
3462 case UDAT_STANDALONE_QUARTER_FIELD:
3463 cal.set(UCAL_MONTH, (value - 1) * 3);
3464 break;
3465 case UDAT_RELATED_YEAR_FIELD:
3466 cal.setRelatedYear(value);
3467 break;
3468 default:
3469 cal.set(field, value);
3470 break;
3471 }
3472 return pos.getIndex();
3473 }
3474 return -start;
3475 }
3476
3477 /**
3478 * Parse an integer using fNumberFormat. This method is semantically
3479 * const, but actually may modify fNumberFormat.
3480 */
3481 void SimpleDateFormat::parseInt(const UnicodeString& text,
3482 Formattable& number,
3483 ParsePosition& pos,
3484 UBool allowNegative,
3485 NumberFormat *fmt) const {
3486 parseInt(text, number, -1, pos, allowNegative,fmt);
3487 }
3488
3489 /**
3490 * Parse an integer using fNumberFormat up to maxDigits.
3491 */
3492 void SimpleDateFormat::parseInt(const UnicodeString& text,
3493 Formattable& number,
3494 int32_t maxDigits,
3495 ParsePosition& pos,
3496 UBool allowNegative,
3497 NumberFormat *fmt) const {
3498 UnicodeString oldPrefix;
3499 DecimalFormat* df = NULL;
3500 if (!allowNegative && (df = dynamic_cast<DecimalFormat*>(fmt)) != NULL) {
3501 df->getNegativePrefix(oldPrefix);
3502 df->setNegativePrefix(UnicodeString(TRUE, SUPPRESS_NEGATIVE_PREFIX, -1));
3503 }
3504 int32_t oldPos = pos.getIndex();
3505 fmt->parse(text, number, pos);
3506 if (df != NULL) {
3507 df->setNegativePrefix(oldPrefix);
3508 }
3509
3510 if (maxDigits > 0) {
3511 // adjust the result to fit into
3512 // the maxDigits and move the position back
3513 int32_t nDigits = pos.getIndex() - oldPos;
3514 if (nDigits > maxDigits) {
3515 int32_t val = number.getLong();
3516 nDigits -= maxDigits;
3517 while (nDigits > 0) {
3518 val /= 10;
3519 nDigits--;
3520 }
3521 pos.setIndex(oldPos + maxDigits);
3522 number.setLong(val);
3523 }
3524 }
3525 }
3526
3527 //----------------------------------------------------------------------
3528
3529 void SimpleDateFormat::translatePattern(const UnicodeString& originalPattern,
3530 UnicodeString& translatedPattern,
3531 const UnicodeString& from,
3532 const UnicodeString& to,
3533 UErrorCode& status)
3534 {
3535 // run through the pattern and convert any pattern symbols from the version
3536 // in "from" to the corresponding character ion "to". This code takes
3537 // quoted strings into account (it doesn't try to translate them), and it signals
3538 // an error if a particular "pattern character" doesn't appear in "from".
3539 // Depending on the values of "from" and "to" this can convert from generic
3540 // to localized patterns or localized to generic.
3541 if (U_FAILURE(status)) {
3542 return;
3543 }
3544
3545 translatedPattern.remove();
3546 UBool inQuote = FALSE;
3547 for (int32_t i = 0; i < originalPattern.length(); ++i) {
3548 UChar c = originalPattern[i];
3549 if (inQuote) {
3550 if (c == QUOTE) {
3551 inQuote = FALSE;
3552 }
3553 } else {
3554 if (c == QUOTE) {
3555 inQuote = TRUE;
3556 } else if (isSyntaxChar(c)) {
3557 int32_t ci = from.indexOf(c);
3558 if (ci == -1) {
3559 status = U_INVALID_FORMAT_ERROR;
3560 return;
3561 }
3562 c = to[ci];
3563 }
3564 }
3565 translatedPattern += c;
3566 }
3567 if (inQuote) {
3568 status = U_INVALID_FORMAT_ERROR;
3569 return;
3570 }
3571 }
3572
3573 //----------------------------------------------------------------------
3574
3575 UnicodeString&
3576 SimpleDateFormat::toPattern(UnicodeString& result) const
3577 {
3578 result = fPattern;
3579 return result;
3580 }
3581
3582 //----------------------------------------------------------------------
3583
3584 UnicodeString&
3585 SimpleDateFormat::toLocalizedPattern(UnicodeString& result,
3586 UErrorCode& status) const
3587 {
3588 translatePattern(fPattern, result,
3589 UnicodeString(DateFormatSymbols::getPatternUChars()),
3590 fSymbols->fLocalPatternChars, status);
3591 return result;
3592 }
3593
3594 //----------------------------------------------------------------------
3595
3596 void
3597 SimpleDateFormat::applyPattern(const UnicodeString& pattern)
3598 {
3599 fPattern = pattern;
3600 updateTimeSepFromPattern(fPattern, fSymbols);
3601 }
3602
3603 //----------------------------------------------------------------------
3604
3605 void
3606 SimpleDateFormat::applyLocalizedPattern(const UnicodeString& pattern,
3607 UErrorCode &status)
3608 {
3609 translatePattern(pattern, fPattern,
3610 fSymbols->fLocalPatternChars,
3611 UnicodeString(DateFormatSymbols::getPatternUChars()), status);
3612 updateTimeSepFromPattern(fPattern, fSymbols);
3613 }
3614
3615 //----------------------------------------------------------------------
3616
3617 const DateFormatSymbols*
3618 SimpleDateFormat::getDateFormatSymbols() const
3619 {
3620 return fSymbols;
3621 }
3622
3623 //----------------------------------------------------------------------
3624
3625 void
3626 SimpleDateFormat::adoptDateFormatSymbols(DateFormatSymbols* newFormatSymbols)
3627 {
3628 delete fSymbols;
3629 fSymbols = newFormatSymbols;
3630 }
3631
3632 //----------------------------------------------------------------------
3633 void
3634 SimpleDateFormat::setDateFormatSymbols(const DateFormatSymbols& newFormatSymbols)
3635 {
3636 delete fSymbols;
3637 fSymbols = new DateFormatSymbols(newFormatSymbols);
3638 }
3639
3640 //----------------------------------------------------------------------
3641 const TimeZoneFormat*
3642 SimpleDateFormat::getTimeZoneFormat(void) const {
3643 return (const TimeZoneFormat*)tzFormat();
3644 }
3645
3646 //----------------------------------------------------------------------
3647 void
3648 SimpleDateFormat::adoptTimeZoneFormat(TimeZoneFormat* timeZoneFormatToAdopt)
3649 {
3650 delete fTimeZoneFormat;
3651 fTimeZoneFormat = timeZoneFormatToAdopt;
3652 }
3653
3654 //----------------------------------------------------------------------
3655 void
3656 SimpleDateFormat::setTimeZoneFormat(const TimeZoneFormat& newTimeZoneFormat)
3657 {
3658 delete fTimeZoneFormat;
3659 fTimeZoneFormat = new TimeZoneFormat(newTimeZoneFormat);
3660 }
3661
3662 //----------------------------------------------------------------------
3663
3664
3665 void SimpleDateFormat::adoptCalendar(Calendar* calendarToAdopt)
3666 {
3667 UErrorCode status = U_ZERO_ERROR;
3668 Locale calLocale(fLocale);
3669 calLocale.setKeywordValue("calendar", calendarToAdopt->getType(), status);
3670 DateFormatSymbols *newSymbols =
3671 DateFormatSymbols::createForLocale(calLocale, status);
3672 if (U_FAILURE(status)) {
3673 return;
3674 }
3675 DateFormat::adoptCalendar(calendarToAdopt);
3676 delete fSymbols;
3677 fSymbols = newSymbols;
3678 initializeDefaultCentury(); // we need a new century (possibly)
3679 }
3680
3681
3682 //----------------------------------------------------------------------
3683
3684
3685 // override the DateFormat implementation in order to
3686 // lazily initialize fCapitalizationBrkIter
3687 void
3688 SimpleDateFormat::setContext(UDisplayContext value, UErrorCode& status)
3689 {
3690 DateFormat::setContext(value, status);
3691 #if !UCONFIG_NO_BREAK_ITERATION
3692 if (U_SUCCESS(status)) {
3693 if ( fCapitalizationBrkIter == NULL && (value==UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE ||
3694 value==UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU || value==UDISPCTX_CAPITALIZATION_FOR_STANDALONE) ) {
3695 UErrorCode status = U_ZERO_ERROR;
3696 fCapitalizationBrkIter = BreakIterator::createSentenceInstance(fLocale, status);
3697 if (U_FAILURE(status)) {
3698 delete fCapitalizationBrkIter;
3699 fCapitalizationBrkIter = NULL;
3700 }
3701 }
3702 }
3703 #endif
3704 }
3705
3706
3707 //----------------------------------------------------------------------
3708
3709
3710 UBool
3711 SimpleDateFormat::isFieldUnitIgnored(UCalendarDateFields field) const {
3712 return isFieldUnitIgnored(fPattern, field);
3713 }
3714
3715
3716 UBool
3717 SimpleDateFormat::isFieldUnitIgnored(const UnicodeString& pattern,
3718 UCalendarDateFields field) {
3719 int32_t fieldLevel = fgCalendarFieldToLevel[field];
3720 int32_t level;
3721 UChar ch;
3722 UBool inQuote = FALSE;
3723 UChar prevCh = 0;
3724 int32_t count = 0;
3725
3726 for (int32_t i = 0; i < pattern.length(); ++i) {
3727 ch = pattern[i];
3728 if (ch != prevCh && count > 0) {
3729 level = getLevelFromChar(prevCh);
3730 // the larger the level, the smaller the field unit.
3731 if (fieldLevel <= level) {
3732 return FALSE;
3733 }
3734 count = 0;
3735 }
3736 if (ch == QUOTE) {
3737 if ((i+1) < pattern.length() && pattern[i+1] == QUOTE) {
3738 ++i;
3739 } else {
3740 inQuote = ! inQuote;
3741 }
3742 }
3743 else if (!inQuote && isSyntaxChar(ch)) {
3744 prevCh = ch;
3745 ++count;
3746 }
3747 }
3748 if (count > 0) {
3749 // last item
3750 level = getLevelFromChar(prevCh);
3751 if (fieldLevel <= level) {
3752 return FALSE;
3753 }
3754 }
3755 return TRUE;
3756 }
3757
3758 //----------------------------------------------------------------------
3759
3760 const Locale&
3761 SimpleDateFormat::getSmpFmtLocale(void) const {
3762 return fLocale;
3763 }
3764
3765 //----------------------------------------------------------------------
3766
3767 int32_t
3768 SimpleDateFormat::checkIntSuffix(const UnicodeString& text, int32_t start,
3769 int32_t patLoc, UBool isNegative) const {
3770 // local variables
3771 UnicodeString suf;
3772 int32_t patternMatch;
3773 int32_t textPreMatch;
3774 int32_t textPostMatch;
3775
3776 // check that we are still in range
3777 if ( (start > text.length()) ||
3778 (start < 0) ||
3779 (patLoc < 0) ||
3780 (patLoc > fPattern.length())) {
3781 // out of range, don't advance location in text
3782 return start;
3783 }
3784
3785 // get the suffix
3786 DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(fNumberFormat);
3787 if (decfmt != NULL) {
3788 if (isNegative) {
3789 suf = decfmt->getNegativeSuffix(suf);
3790 }
3791 else {
3792 suf = decfmt->getPositiveSuffix(suf);
3793 }
3794 }
3795
3796 // check for suffix
3797 if (suf.length() <= 0) {
3798 return start;
3799 }
3800
3801 // check suffix will be encountered in the pattern
3802 patternMatch = compareSimpleAffix(suf,fPattern,patLoc);
3803
3804 // check if a suffix will be encountered in the text
3805 textPreMatch = compareSimpleAffix(suf,text,start);
3806
3807 // check if a suffix was encountered in the text
3808 textPostMatch = compareSimpleAffix(suf,text,start-suf.length());
3809
3810 // check for suffix match
3811 if ((textPreMatch >= 0) && (patternMatch >= 0) && (textPreMatch == patternMatch)) {
3812 return start;
3813 }
3814 else if ((textPostMatch >= 0) && (patternMatch >= 0) && (textPostMatch == patternMatch)) {
3815 return start - suf.length();
3816 }
3817
3818 // should not get here
3819 return start;
3820 }
3821
3822 //----------------------------------------------------------------------
3823
3824 int32_t
3825 SimpleDateFormat::compareSimpleAffix(const UnicodeString& affix,
3826 const UnicodeString& input,
3827 int32_t pos) const {
3828 int32_t start = pos;
3829 for (int32_t i=0; i<affix.length(); ) {
3830 UChar32 c = affix.char32At(i);
3831 int32_t len = U16_LENGTH(c);
3832 if (PatternProps::isWhiteSpace(c)) {
3833 // We may have a pattern like: \u200F \u0020
3834 // and input text like: \u200F \u0020
3835 // Note that U+200F and U+0020 are Pattern_White_Space but only
3836 // U+0020 is UWhiteSpace. So we have to first do a direct
3837 // match of the run of Pattern_White_Space in the pattern,
3838 // then match any extra characters.
3839 UBool literalMatch = FALSE;
3840 while (pos < input.length() &&
3841 input.char32At(pos) == c) {
3842 literalMatch = TRUE;
3843 i += len;
3844 pos += len;
3845 if (i == affix.length()) {
3846 break;
3847 }
3848 c = affix.char32At(i);
3849 len = U16_LENGTH(c);
3850 if (!PatternProps::isWhiteSpace(c)) {
3851 break;
3852 }
3853 }
3854
3855 // Advance over run in pattern
3856 i = skipPatternWhiteSpace(affix, i);
3857
3858 // Advance over run in input text
3859 // Must see at least one white space char in input,
3860 // unless we've already matched some characters literally.
3861 int32_t s = pos;
3862 pos = skipUWhiteSpace(input, pos);
3863 if (pos == s && !literalMatch) {
3864 return -1;
3865 }
3866
3867 // If we skip UWhiteSpace in the input text, we need to skip it in the pattern.
3868 // Otherwise, the previous lines may have skipped over text (such as U+00A0) that
3869 // is also in the affix.
3870 i = skipUWhiteSpace(affix, i);
3871 } else {
3872 if (pos < input.length() &&
3873 input.char32At(pos) == c) {
3874 i += len;
3875 pos += len;
3876 } else {
3877 return -1;
3878 }
3879 }
3880 }
3881 return pos - start;
3882 }
3883
3884 //----------------------------------------------------------------------
3885
3886 int32_t
3887 SimpleDateFormat::skipPatternWhiteSpace(const UnicodeString& text, int32_t pos) const {
3888 const UChar* s = text.getBuffer();
3889 return (int32_t)(PatternProps::skipWhiteSpace(s + pos, text.length() - pos) - s);
3890 }
3891
3892 //----------------------------------------------------------------------
3893
3894 int32_t
3895 SimpleDateFormat::skipUWhiteSpace(const UnicodeString& text, int32_t pos) const {
3896 while (pos < text.length()) {
3897 UChar32 c = text.char32At(pos);
3898 if (!u_isUWhiteSpace(c)) {
3899 break;
3900 }
3901 pos += U16_LENGTH(c);
3902 }
3903 return pos;
3904 }
3905
3906 //----------------------------------------------------------------------
3907
3908 // Lazy TimeZoneFormat instantiation, semantically const.
3909 TimeZoneFormat *
3910 SimpleDateFormat::tzFormat() const {
3911 if (fTimeZoneFormat == NULL) {
3912 umtx_lock(&LOCK);
3913 {
3914 if (fTimeZoneFormat == NULL) {
3915 UErrorCode status = U_ZERO_ERROR;
3916 TimeZoneFormat *tzfmt = TimeZoneFormat::createInstance(fLocale, status);
3917 if (U_FAILURE(status)) {
3918 return NULL;
3919 }
3920
3921 const_cast<SimpleDateFormat *>(this)->fTimeZoneFormat = tzfmt;
3922 }
3923 }
3924 umtx_unlock(&LOCK);
3925 }
3926 return fTimeZoneFormat;
3927 }
3928
3929 U_NAMESPACE_END
3930
3931 #endif /* #if !UCONFIG_NO_FORMATTING */
3932
3933 //eof
mirror of ICU