• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2 *******************************************************************************
3 * Copyright (C) 1997-2013, International Business Machines Corporation and    *
4 * others. All Rights Reserved.                                                *
5 *******************************************************************************
6 *
7 * File SMPDTFMT.CPP
8 *
9 * Modification History:
10 *
11 *   Date        Name        Description
12 *   02/19/97    aliu        Converted from java.
13 *   03/31/97    aliu        Modified extensively to work with 50 locales.
14 *   04/01/97    aliu        Added support for centuries.
15 *   07/09/97    helena      Made ParsePosition into a class.
16 *   07/21/98    stephen     Added initializeDefaultCentury.
17 *                             Removed getZoneIndex (added in DateFormatSymbols)
18 *                             Removed subParseLong
19 *                             Removed chk
20 *  02/22/99     stephen     Removed character literals for EBCDIC safety
21 *   10/14/99    aliu        Updated 2-digit year parsing so that only "00" thru
22 *                           "99" are recognized. {j28 4182066}
23 *   11/15/99    weiv        Added support for week of year/day of week format
24 ********************************************************************************
25 */
26 
27 #define ZID_KEY_MAX 128
28 
29 #include "unicode/utypes.h"
30 
31 #if !UCONFIG_NO_FORMATTING
32 
33 #include "unicode/smpdtfmt.h"
34 #include "unicode/dtfmtsym.h"
35 #include "unicode/ures.h"
36 #include "unicode/msgfmt.h"
37 #include "unicode/calendar.h"
38 #include "unicode/gregocal.h"
39 #include "unicode/timezone.h"
40 #include "unicode/decimfmt.h"
41 #include "unicode/dcfmtsym.h"
42 #include "unicode/uchar.h"
43 #include "unicode/uniset.h"
44 #include "unicode/ustring.h"
45 #include "unicode/basictz.h"
46 #include "unicode/simpletz.h"
47 #include "unicode/rbtz.h"
48 #include "unicode/tzfmt.h"
49 #include "unicode/utf16.h"
50 #include "unicode/vtzone.h"
51 #include "unicode/udisplaycontext.h"
52 #include "olsontz.h"
53 #include "patternprops.h"
54 #include "fphdlimp.h"
55 #include "gregoimp.h"
56 #include "hebrwcal.h"
57 #include "cstring.h"
58 #include "uassert.h"
59 #include "cmemory.h"
60 #include "umutex.h"
61 #include <float.h>
62 #include "smpdtfst.h"
63 
64 #if defined( U_DEBUG_CALSVC ) || defined (U_DEBUG_CAL)
65 #include <stdio.h>
66 #endif
67 
68 // *****************************************************************************
69 // class SimpleDateFormat
70 // *****************************************************************************
71 
72 U_NAMESPACE_BEGIN
73 
74 static const UChar PATTERN_CHAR_BASE = 0x40;
75 
76 /**
77  * Last-resort string to use for "GMT" when constructing time zone strings.
78  */
79 // For time zones that have no names, use strings GMT+minutes and
80 // GMT-minutes. For instance, in France the time zone is GMT+60.
81 // Also accepted are GMT+H:MM or GMT-H:MM.
82 // Currently not being used
83 //static const UChar gGmt[]      = {0x0047, 0x004D, 0x0054, 0x0000};         // "GMT"
84 //static const UChar gGmtPlus[]  = {0x0047, 0x004D, 0x0054, 0x002B, 0x0000}; // "GMT+"
85 //static const UChar gGmtMinus[] = {0x0047, 0x004D, 0x0054, 0x002D, 0x0000}; // "GMT-"
86 //static const UChar gDefGmtPat[]       = {0x0047, 0x004D, 0x0054, 0x007B, 0x0030, 0x007D, 0x0000}; /* GMT{0} */
87 //static const UChar gDefGmtNegHmsPat[] = {0x002D, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x003A, 0x0073, 0x0073, 0x0000}; /* -HH:mm:ss */
88 //static const UChar gDefGmtNegHmPat[]  = {0x002D, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x0000}; /* -HH:mm */
89 //static const UChar gDefGmtPosHmsPat[] = {0x002B, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x003A, 0x0073, 0x0073, 0x0000}; /* +HH:mm:ss */
90 //static const UChar gDefGmtPosHmPat[]  = {0x002B, 0x0048, 0x0048, 0x003A, 0x006D, 0x006D, 0x0000}; /* +HH:mm */
91 //static const UChar gUt[]       = {0x0055, 0x0054, 0x0000};  // "UT"
92 //static const UChar gUtc[]      = {0x0055, 0x0054, 0x0043, 0x0000};  // "UT"
93 
94 typedef enum GmtPatSize {
95     kGmtLen = 3,
96     kGmtPatLen = 6,
97     kNegHmsLen = 9,
98     kNegHmLen = 6,
99     kPosHmsLen = 9,
100     kPosHmLen = 6,
101     kUtLen = 2,
102     kUtcLen = 3
103 } GmtPatSize;
104 
105 // Stuff needed for numbering system overrides
106 
107 typedef enum OvrStrType {
108     kOvrStrDate = 0,
109     kOvrStrTime = 1,
110     kOvrStrBoth = 2
111 } OvrStrType;
112 
113 static const UDateFormatField kDateFields[] = {
114     UDAT_YEAR_FIELD,
115     UDAT_MONTH_FIELD,
116     UDAT_DATE_FIELD,
117     UDAT_DAY_OF_YEAR_FIELD,
118     UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
119     UDAT_WEEK_OF_YEAR_FIELD,
120     UDAT_WEEK_OF_MONTH_FIELD,
121     UDAT_YEAR_WOY_FIELD,
122     UDAT_EXTENDED_YEAR_FIELD,
123     UDAT_JULIAN_DAY_FIELD,
124     UDAT_STANDALONE_DAY_FIELD,
125     UDAT_STANDALONE_MONTH_FIELD,
126     UDAT_QUARTER_FIELD,
127     UDAT_STANDALONE_QUARTER_FIELD,
128     UDAT_YEAR_NAME_FIELD };
129 static const int8_t kDateFieldsCount = 15;
130 
131 static const UDateFormatField kTimeFields[] = {
132     UDAT_HOUR_OF_DAY1_FIELD,
133     UDAT_HOUR_OF_DAY0_FIELD,
134     UDAT_MINUTE_FIELD,
135     UDAT_SECOND_FIELD,
136     UDAT_FRACTIONAL_SECOND_FIELD,
137     UDAT_HOUR1_FIELD,
138     UDAT_HOUR0_FIELD,
139     UDAT_MILLISECONDS_IN_DAY_FIELD,
140     UDAT_TIMEZONE_RFC_FIELD,
141     UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD };
142 static const int8_t kTimeFieldsCount = 10;
143 
144 
145 // This is a pattern-of-last-resort used when we can't load a usable pattern out
146 // of a resource.
147 static const UChar gDefaultPattern[] =
148 {
149     0x79, 0x79, 0x79, 0x79, 0x4D, 0x4D, 0x64, 0x64, 0x20, 0x68, 0x68, 0x3A, 0x6D, 0x6D, 0x20, 0x61, 0
150 };  /* "yyyyMMdd hh:mm a" */
151 
152 // This prefix is designed to NEVER MATCH real text, in order to
153 // suppress the parsing of negative numbers.  Adjust as needed (if
154 // this becomes valid Unicode).
155 static const UChar SUPPRESS_NEGATIVE_PREFIX[] = {0xAB00, 0};
156 
157 /**
158  * These are the tags we expect to see in normal resource bundle files associated
159  * with a locale.
160  */
161 static const char gDateTimePatternsTag[]="DateTimePatterns";
162 
163 //static const UChar gEtcUTC[] = {0x45, 0x74, 0x63, 0x2F, 0x55, 0x54, 0x43, 0x00}; // "Etc/UTC"
164 static const UChar QUOTE = 0x27; // Single quote
165 
166 /*
167  * The field range check bias for each UDateFormatField.
168  * The bias is added to the minimum and maximum values
169  * before they are compared to the parsed number.
170  * For example, the calendar stores zero-based month numbers
171  * but the parsed month numbers start at 1, so the bias is 1.
172  *
173  * A value of -1 means that the value is not checked.
174  */
175 static const int32_t gFieldRangeBias[] = {
176     -1,  // 'G' - UDAT_ERA_FIELD
177     -1,  // 'y' - UDAT_YEAR_FIELD
178      1,  // 'M' - UDAT_MONTH_FIELD
179      0,  // 'd' - UDAT_DATE_FIELD
180     -1,  // 'k' - UDAT_HOUR_OF_DAY1_FIELD
181     -1,  // 'H' - UDAT_HOUR_OF_DAY0_FIELD
182      0,  // 'm' - UDAT_MINUTE_FIELD
183      0,  // 's' - UDAT_SEOND_FIELD
184     -1,  // 'S' - UDAT_FRACTIONAL_SECOND_FIELD (0-999?)
185     -1,  // 'E' - UDAT_DAY_OF_WEEK_FIELD (1-7?)
186     -1,  // 'D' - UDAT_DAY_OF_YEAR_FIELD (1 - 366?)
187     -1,  // 'F' - UDAT_DAY_OF_WEEK_IN_MONTH_FIELD (1-5?)
188     -1,  // 'w' - UDAT_WEEK_OF_YEAR_FIELD (1-52?)
189     -1,  // 'W' - UDAT_WEEK_OF_MONTH_FIELD (1-5?)
190     -1,  // 'a' - UDAT_AM_PM_FIELD
191     -1,  // 'h' - UDAT_HOUR1_FIELD
192     -1,  // 'K' - UDAT_HOUR0_FIELD
193     -1,  // 'z' - UDAT_TIMEZONE_FIELD
194     -1,  // 'Y' - UDAT_YEAR_WOY_FIELD
195     -1,  // 'e' - UDAT_DOW_LOCAL_FIELD
196     -1,  // 'u' - UDAT_EXTENDED_YEAR_FIELD
197     -1,  // 'g' - UDAT_JULIAN_DAY_FIELD
198     -1,  // 'A' - UDAT_MILLISECONDS_IN_DAY_FIELD
199     -1,  // 'Z' - UDAT_TIMEZONE_RFC_FIELD
200     -1,  // 'v' - UDAT_TIMEZONE_GENERIC_FIELD
201      0,  // 'c' - UDAT_STANDALONE_DAY_FIELD
202      1,  // 'L' - UDAT_STANDALONE_MONTH_FIELD
203     -1,  // 'Q' - UDAT_QUARTER_FIELD (1-4?)
204     -1,  // 'q' - UDAT_STANDALONE_QUARTER_FIELD
205     -1   // 'V' - UDAT_TIMEZONE_SPECIAL_FIELD
206     -1,  // 'U' - UDAT_YEAR_NAME_FIELD
207     -1,  // 'O' - UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD
208     -1,  // 'X' - UDAT_TIMEZONE_ISO_FIELD
209     -1,  // 'x' - UDAT_TIMEZONE_ISO_LOCAL_FIELD
210 };
211 
212 // When calendar uses hebr numbering (i.e. he@calendar=hebrew),
213 // offset the years within the current millenium down to 1-999
214 static const int32_t HEBREW_CAL_CUR_MILLENIUM_START_YEAR = 5000;
215 static const int32_t HEBREW_CAL_CUR_MILLENIUM_END_YEAR = 6000;
216 
217 static UMutex LOCK = U_MUTEX_INITIALIZER;
218 
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)219 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleDateFormat)
220 
221 //----------------------------------------------------------------------
222 
223 SimpleDateFormat::~SimpleDateFormat()
224 {
225     delete fSymbols;
226     if (fNumberFormatters) {
227         uprv_free(fNumberFormatters);
228     }
229     if (fTimeZoneFormat) {
230         delete fTimeZoneFormat;
231     }
232 
233     while (fOverrideList) {
234         NSOverride *cur = fOverrideList;
235         fOverrideList = cur->next;
236         delete cur->nf;
237         uprv_free(cur);
238     }
239 }
240 
241 //----------------------------------------------------------------------
242 
SimpleDateFormat(UErrorCode & status)243 SimpleDateFormat::SimpleDateFormat(UErrorCode& status)
244   :   fLocale(Locale::getDefault()),
245       fSymbols(NULL),
246       fTimeZoneFormat(NULL),
247       fNumberFormatters(NULL),
248       fOverrideList(NULL),
249       fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
250 {
251     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
252     construct(kShort, (EStyle) (kShort + kDateOffset), fLocale, status);
253     initializeDefaultCentury();
254 }
255 
256 //----------------------------------------------------------------------
257 
SimpleDateFormat(const UnicodeString & pattern,UErrorCode & status)258 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
259                                    UErrorCode &status)
260 :   fPattern(pattern),
261     fLocale(Locale::getDefault()),
262     fSymbols(NULL),
263     fTimeZoneFormat(NULL),
264     fNumberFormatters(NULL),
265     fOverrideList(NULL),
266     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
267 {
268     fDateOverride.setToBogus();
269     fTimeOverride.setToBogus();
270     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
271     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
272     initialize(fLocale, status);
273     initializeDefaultCentury();
274 
275 }
276 //----------------------------------------------------------------------
277 
SimpleDateFormat(const UnicodeString & pattern,const UnicodeString & override,UErrorCode & status)278 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
279                                    const UnicodeString& override,
280                                    UErrorCode &status)
281 :   fPattern(pattern),
282     fLocale(Locale::getDefault()),
283     fSymbols(NULL),
284     fTimeZoneFormat(NULL),
285     fNumberFormatters(NULL),
286     fOverrideList(NULL),
287     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
288 {
289     fDateOverride.setTo(override);
290     fTimeOverride.setToBogus();
291     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
292     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
293     initialize(fLocale, status);
294     initializeDefaultCentury();
295 
296     processOverrideString(fLocale,override,kOvrStrBoth,status);
297 
298 }
299 
300 //----------------------------------------------------------------------
301 
SimpleDateFormat(const UnicodeString & pattern,const Locale & locale,UErrorCode & status)302 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
303                                    const Locale& locale,
304                                    UErrorCode& status)
305 :   fPattern(pattern),
306     fLocale(locale),
307     fTimeZoneFormat(NULL),
308     fNumberFormatters(NULL),
309     fOverrideList(NULL),
310     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
311 {
312 
313     fDateOverride.setToBogus();
314     fTimeOverride.setToBogus();
315     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
316 
317     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
318     initialize(fLocale, status);
319     initializeDefaultCentury();
320 }
321 
322 //----------------------------------------------------------------------
323 
SimpleDateFormat(const UnicodeString & pattern,const UnicodeString & override,const Locale & locale,UErrorCode & status)324 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
325                                    const UnicodeString& override,
326                                    const Locale& locale,
327                                    UErrorCode& status)
328 :   fPattern(pattern),
329     fLocale(locale),
330     fTimeZoneFormat(NULL),
331     fNumberFormatters(NULL),
332     fOverrideList(NULL),
333     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
334 {
335 
336     fDateOverride.setTo(override);
337     fTimeOverride.setToBogus();
338     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
339 
340     initializeSymbols(fLocale, initializeCalendar(NULL,fLocale,status), status);
341     initialize(fLocale, status);
342     initializeDefaultCentury();
343 
344     processOverrideString(locale,override,kOvrStrBoth,status);
345 
346 }
347 
348 //----------------------------------------------------------------------
349 
SimpleDateFormat(const UnicodeString & pattern,DateFormatSymbols * symbolsToAdopt,UErrorCode & status)350 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
351                                    DateFormatSymbols* symbolsToAdopt,
352                                    UErrorCode& status)
353 :   fPattern(pattern),
354     fLocale(Locale::getDefault()),
355     fSymbols(symbolsToAdopt),
356     fTimeZoneFormat(NULL),
357     fNumberFormatters(NULL),
358     fOverrideList(NULL),
359     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
360 {
361 
362     fDateOverride.setToBogus();
363     fTimeOverride.setToBogus();
364     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
365 
366     initializeCalendar(NULL,fLocale,status);
367     initialize(fLocale, status);
368     initializeDefaultCentury();
369 }
370 
371 //----------------------------------------------------------------------
372 
SimpleDateFormat(const UnicodeString & pattern,const DateFormatSymbols & symbols,UErrorCode & status)373 SimpleDateFormat::SimpleDateFormat(const UnicodeString& pattern,
374                                    const DateFormatSymbols& symbols,
375                                    UErrorCode& status)
376 :   fPattern(pattern),
377     fLocale(Locale::getDefault()),
378     fSymbols(new DateFormatSymbols(symbols)),
379     fTimeZoneFormat(NULL),
380     fNumberFormatters(NULL),
381     fOverrideList(NULL),
382     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
383 {
384 
385     fDateOverride.setToBogus();
386     fTimeOverride.setToBogus();
387     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
388 
389     initializeCalendar(NULL, fLocale, status);
390     initialize(fLocale, status);
391     initializeDefaultCentury();
392 }
393 
394 //----------------------------------------------------------------------
395 
396 // Not for public consumption; used by DateFormat
SimpleDateFormat(EStyle timeStyle,EStyle dateStyle,const Locale & locale,UErrorCode & status)397 SimpleDateFormat::SimpleDateFormat(EStyle timeStyle,
398                                    EStyle dateStyle,
399                                    const Locale& locale,
400                                    UErrorCode& status)
401 :   fLocale(locale),
402     fSymbols(NULL),
403     fTimeZoneFormat(NULL),
404     fNumberFormatters(NULL),
405     fOverrideList(NULL),
406     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
407 {
408     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
409     construct(timeStyle, dateStyle, fLocale, status);
410     if(U_SUCCESS(status)) {
411       initializeDefaultCentury();
412     }
413 }
414 
415 //----------------------------------------------------------------------
416 
417 /**
418  * Not for public consumption; used by DateFormat.  This constructor
419  * never fails.  If the resource data is not available, it uses the
420  * the last resort symbols.
421  */
SimpleDateFormat(const Locale & locale,UErrorCode & status)422 SimpleDateFormat::SimpleDateFormat(const Locale& locale,
423                                    UErrorCode& status)
424 :   fPattern(gDefaultPattern),
425     fLocale(locale),
426     fSymbols(NULL),
427     fTimeZoneFormat(NULL),
428     fNumberFormatters(NULL),
429     fOverrideList(NULL),
430     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
431 {
432     if (U_FAILURE(status)) return;
433     initializeSymbols(fLocale, initializeCalendar(NULL, fLocale, status),status);
434     if (U_FAILURE(status))
435     {
436         status = U_ZERO_ERROR;
437         delete fSymbols;
438         // This constructor doesn't fail; it uses last resort data
439         fSymbols = new DateFormatSymbols(status);
440         /* test for NULL */
441         if (fSymbols == 0) {
442             status = U_MEMORY_ALLOCATION_ERROR;
443             return;
444         }
445     }
446 
447     fDateOverride.setToBogus();
448     fTimeOverride.setToBogus();
449     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
450 
451     initialize(fLocale, status);
452     if(U_SUCCESS(status)) {
453       initializeDefaultCentury();
454     }
455 }
456 
457 //----------------------------------------------------------------------
458 
SimpleDateFormat(const SimpleDateFormat & other)459 SimpleDateFormat::SimpleDateFormat(const SimpleDateFormat& other)
460 :   DateFormat(other),
461     fLocale(other.fLocale),
462     fSymbols(NULL),
463     fTimeZoneFormat(NULL),
464     fNumberFormatters(NULL),
465     fOverrideList(NULL),
466     fCapitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
467 {
468     UErrorCode status = U_ZERO_ERROR;
469     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status).setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
470     *this = other;
471 }
472 
473 //----------------------------------------------------------------------
474 
operator =(const SimpleDateFormat & other)475 SimpleDateFormat& SimpleDateFormat::operator=(const SimpleDateFormat& other)
476 {
477     if (this == &other) {
478         return *this;
479     }
480     DateFormat::operator=(other);
481 
482     delete fSymbols;
483     fSymbols = NULL;
484 
485     if (other.fSymbols)
486         fSymbols = new DateFormatSymbols(*other.fSymbols);
487 
488     fDefaultCenturyStart         = other.fDefaultCenturyStart;
489     fDefaultCenturyStartYear     = other.fDefaultCenturyStartYear;
490     fHaveDefaultCentury          = other.fHaveDefaultCentury;
491 
492     fPattern = other.fPattern;
493 
494     // TimeZoneFormat in ICU4C only depends on a locale for now
495     if (fLocale != other.fLocale) {
496         delete fTimeZoneFormat;
497         fTimeZoneFormat = NULL; // forces lazy instantiation with the other locale
498         fLocale = other.fLocale;
499     }
500 
501     fCapitalizationContext = other.fCapitalizationContext;
502 
503     return *this;
504 }
505 
506 //----------------------------------------------------------------------
507 
508 Format*
clone() const509 SimpleDateFormat::clone() const
510 {
511     return new SimpleDateFormat(*this);
512 }
513 
514 //----------------------------------------------------------------------
515 
516 UBool
operator ==(const Format & other) const517 SimpleDateFormat::operator==(const Format& other) const
518 {
519     if (DateFormat::operator==(other)) {
520         // DateFormat::operator== guarantees following cast is safe
521         SimpleDateFormat* that = (SimpleDateFormat*)&other;
522         return (fPattern             == that->fPattern &&
523                 fSymbols             != NULL && // Check for pathological object
524                 that->fSymbols       != NULL && // Check for pathological object
525                 *fSymbols            == *that->fSymbols &&
526                 fHaveDefaultCentury  == that->fHaveDefaultCentury &&
527                 fDefaultCenturyStart == that->fDefaultCenturyStart &&
528                 fCapitalizationContext == that->fCapitalizationContext);
529     }
530     return FALSE;
531 }
532 
533 //----------------------------------------------------------------------
534 
construct(EStyle timeStyle,EStyle dateStyle,const Locale & locale,UErrorCode & status)535 void SimpleDateFormat::construct(EStyle timeStyle,
536                                  EStyle dateStyle,
537                                  const Locale& locale,
538                                  UErrorCode& status)
539 {
540     // called by several constructors to load pattern data from the resources
541     if (U_FAILURE(status)) return;
542 
543     // We will need the calendar to know what type of symbols to load.
544     initializeCalendar(NULL, locale, status);
545     if (U_FAILURE(status)) return;
546 
547     CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
548     UResourceBundle *dateTimePatterns = calData.getByKey(gDateTimePatternsTag, status);
549     UResourceBundle *currentBundle;
550 
551     if (U_FAILURE(status)) return;
552 
553     if (ures_getSize(dateTimePatterns) <= kDateTime)
554     {
555         status = U_INVALID_FORMAT_ERROR;
556         return;
557     }
558 
559     setLocaleIDs(ures_getLocaleByType(dateTimePatterns, ULOC_VALID_LOCALE, &status),
560                  ures_getLocaleByType(dateTimePatterns, ULOC_ACTUAL_LOCALE, &status));
561 
562     // create a symbols object from the locale
563     initializeSymbols(locale,fCalendar, status);
564     if (U_FAILURE(status)) return;
565     /* test for NULL */
566     if (fSymbols == 0) {
567         status = U_MEMORY_ALLOCATION_ERROR;
568         return;
569     }
570 
571     const UChar *resStr,*ovrStr;
572     int32_t resStrLen,ovrStrLen = 0;
573     fDateOverride.setToBogus();
574     fTimeOverride.setToBogus();
575 
576     // if the pattern should include both date and time information, use the date/time
577     // pattern string as a guide to tell use how to glue together the appropriate date
578     // and time pattern strings.  The actual gluing-together is handled by a convenience
579     // method on MessageFormat.
580     if ((timeStyle != kNone) && (dateStyle != kNone))
581     {
582         Formattable timeDateArray[2];
583 
584         // use Formattable::adoptString() so that we can use fastCopyFrom()
585         // instead of Formattable::setString()'s unaware, safe, deep string clone
586         // see Jitterbug 2296
587 
588         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
589         if (U_FAILURE(status)) {
590            status = U_INVALID_FORMAT_ERROR;
591            return;
592         }
593         switch (ures_getType(currentBundle)) {
594             case URES_STRING: {
595                resStr = ures_getString(currentBundle, &resStrLen, &status);
596                break;
597             }
598             case URES_ARRAY: {
599                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
600                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
601                fTimeOverride.setTo(TRUE, ovrStr, ovrStrLen);
602                break;
603             }
604             default: {
605                status = U_INVALID_FORMAT_ERROR;
606                ures_close(currentBundle);
607                return;
608             }
609         }
610         ures_close(currentBundle);
611 
612         UnicodeString *tempus1 = new UnicodeString(TRUE, resStr, resStrLen);
613         // NULL pointer check
614         if (tempus1 == NULL) {
615             status = U_MEMORY_ALLOCATION_ERROR;
616             return;
617         }
618         timeDateArray[0].adoptString(tempus1);
619 
620         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)dateStyle, NULL, &status);
621         if (U_FAILURE(status)) {
622            status = U_INVALID_FORMAT_ERROR;
623            return;
624         }
625         switch (ures_getType(currentBundle)) {
626             case URES_STRING: {
627                resStr = ures_getString(currentBundle, &resStrLen, &status);
628                break;
629             }
630             case URES_ARRAY: {
631                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
632                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
633                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
634                break;
635             }
636             default: {
637                status = U_INVALID_FORMAT_ERROR;
638                ures_close(currentBundle);
639                return;
640             }
641         }
642         ures_close(currentBundle);
643 
644         UnicodeString *tempus2 = new UnicodeString(TRUE, resStr, resStrLen);
645         // Null pointer check
646         if (tempus2 == NULL) {
647             status = U_MEMORY_ALLOCATION_ERROR;
648             return;
649         }
650         timeDateArray[1].adoptString(tempus2);
651 
652         int32_t glueIndex = kDateTime;
653         int32_t patternsSize = ures_getSize(dateTimePatterns);
654         if (patternsSize >= (kDateTimeOffset + kShort + 1)) {
655             // Get proper date time format
656             glueIndex = (int32_t)(kDateTimeOffset + (dateStyle - kDateOffset));
657         }
658 
659         resStr = ures_getStringByIndex(dateTimePatterns, glueIndex, &resStrLen, &status);
660         MessageFormat::format(UnicodeString(TRUE, resStr, resStrLen), timeDateArray, 2, fPattern, status);
661     }
662     // if the pattern includes just time data or just date date, load the appropriate
663     // pattern string from the resources
664     // setTo() - see DateFormatSymbols::assignArray comments
665     else if (timeStyle != kNone) {
666         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)timeStyle, NULL, &status);
667         if (U_FAILURE(status)) {
668            status = U_INVALID_FORMAT_ERROR;
669            return;
670         }
671         switch (ures_getType(currentBundle)) {
672             case URES_STRING: {
673                resStr = ures_getString(currentBundle, &resStrLen, &status);
674                break;
675             }
676             case URES_ARRAY: {
677                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
678                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
679                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
680                break;
681             }
682             default: {
683                status = U_INVALID_FORMAT_ERROR;
684                 ures_close(currentBundle);
685                return;
686             }
687         }
688         fPattern.setTo(TRUE, resStr, resStrLen);
689         ures_close(currentBundle);
690     }
691     else if (dateStyle != kNone) {
692         currentBundle = ures_getByIndex(dateTimePatterns, (int32_t)dateStyle, NULL, &status);
693         if (U_FAILURE(status)) {
694            status = U_INVALID_FORMAT_ERROR;
695            return;
696         }
697         switch (ures_getType(currentBundle)) {
698             case URES_STRING: {
699                resStr = ures_getString(currentBundle, &resStrLen, &status);
700                break;
701             }
702             case URES_ARRAY: {
703                resStr = ures_getStringByIndex(currentBundle, 0, &resStrLen, &status);
704                ovrStr = ures_getStringByIndex(currentBundle, 1, &ovrStrLen, &status);
705                fDateOverride.setTo(TRUE, ovrStr, ovrStrLen);
706                break;
707             }
708             default: {
709                status = U_INVALID_FORMAT_ERROR;
710                ures_close(currentBundle);
711                return;
712             }
713         }
714         fPattern.setTo(TRUE, resStr, resStrLen);
715         ures_close(currentBundle);
716     }
717 
718     // and if it includes _neither_, that's an error
719     else
720         status = U_INVALID_FORMAT_ERROR;
721 
722     // finally, finish initializing by creating a Calendar and a NumberFormat
723     initialize(locale, status);
724 }
725 
726 //----------------------------------------------------------------------
727 
728 Calendar*
initializeCalendar(TimeZone * adoptZone,const Locale & locale,UErrorCode & status)729 SimpleDateFormat::initializeCalendar(TimeZone* adoptZone, const Locale& locale, UErrorCode& status)
730 {
731     if(!U_FAILURE(status)) {
732         fCalendar = Calendar::createInstance(adoptZone?adoptZone:TimeZone::createDefault(), locale, status);
733     }
734     if (U_SUCCESS(status) && fCalendar == NULL) {
735         status = U_MEMORY_ALLOCATION_ERROR;
736     }
737     return fCalendar;
738 }
739 
740 void
initializeSymbols(const Locale & locale,Calendar * calendar,UErrorCode & status)741 SimpleDateFormat::initializeSymbols(const Locale& locale, Calendar* calendar, UErrorCode& status)
742 {
743   if(U_FAILURE(status)) {
744     fSymbols = NULL;
745   } else {
746     // pass in calendar type - use NULL (default) if no calendar set (or err).
747     fSymbols = new DateFormatSymbols(locale, calendar?calendar->getType() :NULL , status);
748     // Null pointer check
749     if (fSymbols == NULL) {
750         status = U_MEMORY_ALLOCATION_ERROR;
751         return;
752     }
753   }
754 }
755 
756 void
initialize(const Locale & locale,UErrorCode & status)757 SimpleDateFormat::initialize(const Locale& locale,
758                              UErrorCode& status)
759 {
760     if (U_FAILURE(status)) return;
761 
762     // We don't need to check that the row count is >= 1, since all 2d arrays have at
763     // least one row
764     fNumberFormat = NumberFormat::createInstance(locale, status);
765     if (fNumberFormat != NULL && U_SUCCESS(status))
766     {
767         // no matter what the locale's default number format looked like, we want
768         // to modify it so that it doesn't use thousands separators, doesn't always
769         // show the decimal point, and recognizes integers only when parsing
770 
771         fNumberFormat->setGroupingUsed(FALSE);
772         DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(fNumberFormat);
773         if (decfmt != NULL) {
774             decfmt->setDecimalSeparatorAlwaysShown(FALSE);
775         }
776         fNumberFormat->setParseIntegerOnly(TRUE);
777         fNumberFormat->setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
778 
779         //fNumberFormat->setLenient(TRUE); // Java uses a custom DateNumberFormat to format/parse
780 
781         initNumberFormatters(locale,status);
782 
783     }
784     else if (U_SUCCESS(status))
785     {
786         status = U_MISSING_RESOURCE_ERROR;
787     }
788 }
789 
790 /* Initialize the fields we use to disambiguate ambiguous years. Separate
791  * so we can call it from readObject().
792  */
initializeDefaultCentury()793 void SimpleDateFormat::initializeDefaultCentury()
794 {
795   if(fCalendar) {
796     fHaveDefaultCentury = fCalendar->haveDefaultCentury();
797     if(fHaveDefaultCentury) {
798       fDefaultCenturyStart = fCalendar->defaultCenturyStart();
799       fDefaultCenturyStartYear = fCalendar->defaultCenturyStartYear();
800     } else {
801       fDefaultCenturyStart = DBL_MIN;
802       fDefaultCenturyStartYear = -1;
803     }
804   }
805 }
806 
807 /* Define one-century window into which to disambiguate dates using
808  * two-digit years. Make public in JDK 1.2.
809  */
parseAmbiguousDatesAsAfter(UDate startDate,UErrorCode & status)810 void SimpleDateFormat::parseAmbiguousDatesAsAfter(UDate startDate, UErrorCode& status)
811 {
812     if(U_FAILURE(status)) {
813         return;
814     }
815     if(!fCalendar) {
816       status = U_ILLEGAL_ARGUMENT_ERROR;
817       return;
818     }
819 
820     fCalendar->setTime(startDate, status);
821     if(U_SUCCESS(status)) {
822         fHaveDefaultCentury = TRUE;
823         fDefaultCenturyStart = startDate;
824         fDefaultCenturyStartYear = fCalendar->get(UCAL_YEAR, status);
825     }
826 }
827 
828 //----------------------------------------------------------------------
829 
830 UnicodeString&
format(Calendar & cal,UnicodeString & appendTo,FieldPosition & pos) const831 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo, FieldPosition& pos) const
832 {
833   UErrorCode status = U_ZERO_ERROR;
834   FieldPositionOnlyHandler handler(pos);
835   return _format(cal, appendTo, handler, status);
836 }
837 
838 //----------------------------------------------------------------------
839 
840 UnicodeString&
format(Calendar & cal,UnicodeString & appendTo,FieldPositionIterator * posIter,UErrorCode & status) const841 SimpleDateFormat::format(Calendar& cal, UnicodeString& appendTo,
842                          FieldPositionIterator* posIter, UErrorCode& status) const
843 {
844   FieldPositionIteratorHandler handler(posIter, status);
845   return _format(cal, appendTo, handler, status);
846 }
847 
848 //----------------------------------------------------------------------
849 
850 UnicodeString&
_format(Calendar & cal,UnicodeString & appendTo,FieldPositionHandler & handler,UErrorCode & status) const851 SimpleDateFormat::_format(Calendar& cal, UnicodeString& appendTo,
852                             FieldPositionHandler& handler, UErrorCode& status) const
853 {
854     if ( U_FAILURE(status) ) {
855        return appendTo;
856     }
857     Calendar* workCal = &cal;
858     Calendar* calClone = NULL;
859     if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
860         // Different calendar type
861         // We use the time and time zone from the input calendar, but
862         // do not use the input calendar for field calculation.
863         calClone = fCalendar->clone();
864         if (calClone != NULL) {
865             UDate t = cal.getTime(status);
866             calClone->setTime(t, status);
867             calClone->setTimeZone(cal.getTimeZone());
868             workCal = calClone;
869         } else {
870             status = U_MEMORY_ALLOCATION_ERROR;
871             return appendTo;
872         }
873     }
874 
875     UBool inQuote = FALSE;
876     UChar prevCh = 0;
877     int32_t count = 0;
878     int32_t fieldNum = 0;
879 
880     // loop through the pattern string character by character
881     for (int32_t i = 0; i < fPattern.length() && U_SUCCESS(status); ++i) {
882         UChar ch = fPattern[i];
883 
884         // Use subFormat() to format a repeated pattern character
885         // when a different pattern or non-pattern character is seen
886         if (ch != prevCh && count > 0) {
887             subFormat(appendTo, prevCh, count, fCapitalizationContext, fieldNum++, handler, *workCal, status);
888             count = 0;
889         }
890         if (ch == QUOTE) {
891             // Consecutive single quotes are a single quote literal,
892             // either outside of quotes or between quotes
893             if ((i+1) < fPattern.length() && fPattern[i+1] == QUOTE) {
894                 appendTo += (UChar)QUOTE;
895                 ++i;
896             } else {
897                 inQuote = ! inQuote;
898             }
899         }
900         else if ( ! inQuote && ((ch >= 0x0061 /*'a'*/ && ch <= 0x007A /*'z'*/)
901                     || (ch >= 0x0041 /*'A'*/ && ch <= 0x005A /*'Z'*/))) {
902             // ch is a date-time pattern character to be interpreted
903             // by subFormat(); count the number of times it is repeated
904             prevCh = ch;
905             ++count;
906         }
907         else {
908             // Append quoted characters and unquoted non-pattern characters
909             appendTo += ch;
910         }
911     }
912 
913     // Format the last item in the pattern, if any
914     if (count > 0) {
915         subFormat(appendTo, prevCh, count, fCapitalizationContext, fieldNum++, handler, *workCal, status);
916     }
917 
918     if (calClone != NULL) {
919         delete calClone;
920     }
921 
922     return appendTo;
923 }
924 
925 //----------------------------------------------------------------------
926 
927 /* Map calendar field into calendar field level.
928  * the larger the level, the smaller the field unit.
929  * For example, UCAL_ERA level is 0, UCAL_YEAR level is 10,
930  * UCAL_MONTH level is 20.
931  * NOTE: if new fields adds in, the table needs to update.
932  */
933 const int32_t
934 SimpleDateFormat::fgCalendarFieldToLevel[] =
935 {
936     /*GyM*/ 0, 10, 20,
937     /*wW*/ 20, 30,
938     /*dDEF*/ 30, 20, 30, 30,
939     /*ahHm*/ 40, 50, 50, 60,
940     /*sS..*/ 70, 80,
941     /*z?Y*/ 0, 0, 10,
942     /*eug*/ 30, 10, 0,
943     /*A*/ 40
944 };
945 
946 
947 /* Map calendar field LETTER into calendar field level.
948  * the larger the level, the smaller the field unit.
949  * NOTE: if new fields adds in, the table needs to update.
950  */
951 const int32_t
952 SimpleDateFormat::fgPatternCharToLevel[] = {
953     //       A   B   C   D   E   F   G   H   I   J   K   L   M   N   O
954         -1, 40, -1, -1, 20, 30, 30,  0, 50, -1, -1, 50, 20, 20, -1,  0,
955     //   P   Q   R   S   T   U   V   W   X   Y   Z
956         -1, 20, -1, 80, -1, 10,  0, 30,  0, 10,  0, -1, -1, -1, -1, -1,
957     //       a   b   c   d   e   f   g   h   i   j   k   l   m   n   o
958         -1, 40, -1, 30, 30, 30, -1,  0, 50, -1, -1, 50, -1, 60, -1, -1,
959     //   p   q   r   s   t   u   v   w   x   y   z
960         -1, 20, -1, 70, -1, 10,  0, 20,  0, 10,  0, -1, -1, -1, -1, -1
961 };
962 
963 
964 // Map index into pattern character string to Calendar field number.
965 const UCalendarDateFields
966 SimpleDateFormat::fgPatternIndexToCalendarField[] =
967 {
968     /*GyM*/ UCAL_ERA, UCAL_YEAR, UCAL_MONTH,
969     /*dkH*/ UCAL_DATE, UCAL_HOUR_OF_DAY, UCAL_HOUR_OF_DAY,
970     /*msS*/ UCAL_MINUTE, UCAL_SECOND, UCAL_MILLISECOND,
971     /*EDF*/ UCAL_DAY_OF_WEEK, UCAL_DAY_OF_YEAR, UCAL_DAY_OF_WEEK_IN_MONTH,
972     /*wWa*/ UCAL_WEEK_OF_YEAR, UCAL_WEEK_OF_MONTH, UCAL_AM_PM,
973     /*hKz*/ UCAL_HOUR, UCAL_HOUR, UCAL_ZONE_OFFSET,
974     /*Yeu*/ UCAL_YEAR_WOY, UCAL_DOW_LOCAL, UCAL_EXTENDED_YEAR,
975     /*gAZ*/ UCAL_JULIAN_DAY, UCAL_MILLISECONDS_IN_DAY, UCAL_ZONE_OFFSET,
976     /*v*/   UCAL_ZONE_OFFSET,
977     /*c*/   UCAL_DOW_LOCAL,
978     /*L*/   UCAL_MONTH,
979     /*Q*/   UCAL_MONTH,
980     /*q*/   UCAL_MONTH,
981     /*V*/   UCAL_ZONE_OFFSET,
982     /*U*/   UCAL_YEAR,
983     /*O*/   UCAL_ZONE_OFFSET,
984     /*Xx*/  UCAL_ZONE_OFFSET, UCAL_ZONE_OFFSET,
985 };
986 
987 // Map index into pattern character string to DateFormat field number
988 const UDateFormatField
989 SimpleDateFormat::fgPatternIndexToDateFormatField[] = {
990     /*GyM*/ UDAT_ERA_FIELD, UDAT_YEAR_FIELD, UDAT_MONTH_FIELD,
991     /*dkH*/ UDAT_DATE_FIELD, UDAT_HOUR_OF_DAY1_FIELD, UDAT_HOUR_OF_DAY0_FIELD,
992     /*msS*/ UDAT_MINUTE_FIELD, UDAT_SECOND_FIELD, UDAT_FRACTIONAL_SECOND_FIELD,
993     /*EDF*/ UDAT_DAY_OF_WEEK_FIELD, UDAT_DAY_OF_YEAR_FIELD, UDAT_DAY_OF_WEEK_IN_MONTH_FIELD,
994     /*wWa*/ UDAT_WEEK_OF_YEAR_FIELD, UDAT_WEEK_OF_MONTH_FIELD, UDAT_AM_PM_FIELD,
995     /*hKz*/ UDAT_HOUR1_FIELD, UDAT_HOUR0_FIELD, UDAT_TIMEZONE_FIELD,
996     /*Yeu*/ UDAT_YEAR_WOY_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_EXTENDED_YEAR_FIELD,
997     /*gAZ*/ UDAT_JULIAN_DAY_FIELD, UDAT_MILLISECONDS_IN_DAY_FIELD, UDAT_TIMEZONE_RFC_FIELD,
998     /*v*/   UDAT_TIMEZONE_GENERIC_FIELD,
999     /*c*/   UDAT_STANDALONE_DAY_FIELD,
1000     /*L*/   UDAT_STANDALONE_MONTH_FIELD,
1001     /*Q*/   UDAT_QUARTER_FIELD,
1002     /*q*/   UDAT_STANDALONE_QUARTER_FIELD,
1003     /*V*/   UDAT_TIMEZONE_SPECIAL_FIELD,
1004     /*U*/   UDAT_YEAR_NAME_FIELD,
1005     /*O*/   UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD,
1006     /*Xx*/  UDAT_TIMEZONE_ISO_FIELD, UDAT_TIMEZONE_ISO_LOCAL_FIELD,
1007 };
1008 
1009 //----------------------------------------------------------------------
1010 
1011 /**
1012  * Append symbols[value] to dst.  Make sure the array index is not out
1013  * of bounds.
1014  */
1015 static inline void
_appendSymbol(UnicodeString & dst,int32_t value,const UnicodeString * symbols,int32_t symbolsCount)1016 _appendSymbol(UnicodeString& dst,
1017               int32_t value,
1018               const UnicodeString* symbols,
1019               int32_t symbolsCount) {
1020     U_ASSERT(0 <= value && value < symbolsCount);
1021     if (0 <= value && value < symbolsCount) {
1022         dst += symbols[value];
1023     }
1024 }
1025 
1026 static inline void
_appendSymbolWithMonthPattern(UnicodeString & dst,int32_t value,const UnicodeString * symbols,int32_t symbolsCount,const UnicodeString * monthPattern,UErrorCode & status)1027 _appendSymbolWithMonthPattern(UnicodeString& dst, int32_t value, const UnicodeString* symbols, int32_t symbolsCount,
1028               const UnicodeString* monthPattern, UErrorCode& status) {
1029     U_ASSERT(0 <= value && value < symbolsCount);
1030     if (0 <= value && value < symbolsCount) {
1031         if (monthPattern == NULL) {
1032             dst += symbols[value];
1033         } else {
1034             Formattable monthName((const UnicodeString&)(symbols[value]));
1035             MessageFormat::format(*monthPattern, &monthName, 1, dst, status);
1036         }
1037     }
1038 }
1039 
1040 //----------------------------------------------------------------------
1041 void
initNumberFormatters(const Locale & locale,UErrorCode & status)1042 SimpleDateFormat::initNumberFormatters(const Locale &locale,UErrorCode &status) {
1043     if (U_FAILURE(status)) {
1044         return;
1045     }
1046     if ( fDateOverride.isBogus() && fTimeOverride.isBogus() ) {
1047         return;
1048     }
1049     umtx_lock(&LOCK);
1050     if (fNumberFormatters == NULL) {
1051         fNumberFormatters = (NumberFormat**)uprv_malloc(UDAT_FIELD_COUNT * sizeof(NumberFormat*));
1052         if (fNumberFormatters) {
1053             for (int32_t i = 0; i < UDAT_FIELD_COUNT; i++) {
1054                 fNumberFormatters[i] = fNumberFormat;
1055             }
1056         } else {
1057             status = U_MEMORY_ALLOCATION_ERROR;
1058         }
1059     }
1060     umtx_unlock(&LOCK);
1061 
1062     processOverrideString(locale,fDateOverride,kOvrStrDate,status);
1063     processOverrideString(locale,fTimeOverride,kOvrStrTime,status);
1064 
1065 }
1066 
1067 void
processOverrideString(const Locale & locale,const UnicodeString & str,int8_t type,UErrorCode & status)1068 SimpleDateFormat::processOverrideString(const Locale &locale, const UnicodeString &str, int8_t type, UErrorCode &status) {
1069     if (str.isBogus()) {
1070         return;
1071     }
1072     int32_t start = 0;
1073     int32_t len;
1074     UnicodeString nsName;
1075     UnicodeString ovrField;
1076     UBool moreToProcess = TRUE;
1077 
1078     while (moreToProcess) {
1079         int32_t delimiterPosition = str.indexOf((UChar)ULOC_KEYWORD_ITEM_SEPARATOR_UNICODE,start);
1080         if (delimiterPosition == -1) {
1081             moreToProcess = FALSE;
1082             len = str.length() - start;
1083         } else {
1084             len = delimiterPosition - start;
1085         }
1086         UnicodeString currentString(str,start,len);
1087         int32_t equalSignPosition = currentString.indexOf((UChar)ULOC_KEYWORD_ASSIGN_UNICODE,0);
1088         if (equalSignPosition == -1) { // Simple override string such as "hebrew"
1089             nsName.setTo(currentString);
1090             ovrField.setToBogus();
1091         } else { // Field specific override string such as "y=hebrew"
1092             nsName.setTo(currentString,equalSignPosition+1);
1093             ovrField.setTo(currentString,0,1); // We just need the first character.
1094         }
1095 
1096         int32_t nsNameHash = nsName.hashCode();
1097         // See if the numbering system is in the override list, if not, then add it.
1098         NSOverride *cur = fOverrideList;
1099         NumberFormat *nf = NULL;
1100         UBool found = FALSE;
1101         while ( cur && !found ) {
1102             if ( cur->hash == nsNameHash ) {
1103                 nf = cur->nf;
1104                 found = TRUE;
1105             }
1106             cur = cur->next;
1107         }
1108 
1109         if (!found) {
1110            cur = (NSOverride *)uprv_malloc(sizeof(NSOverride));
1111            if (cur) {
1112                char kw[ULOC_KEYWORD_AND_VALUES_CAPACITY];
1113                uprv_strcpy(kw,"numbers=");
1114                nsName.extract(0,len,kw+8,ULOC_KEYWORD_AND_VALUES_CAPACITY-8,US_INV);
1115 
1116                Locale ovrLoc(locale.getLanguage(),locale.getCountry(),locale.getVariant(),kw);
1117                nf = NumberFormat::createInstance(ovrLoc,status);
1118 
1119                // no matter what the locale's default number format looked like, we want
1120                // to modify it so that it doesn't use thousands separators, doesn't always
1121                // show the decimal point, and recognizes integers only when parsing
1122 
1123                if (U_SUCCESS(status)) {
1124                    nf->setGroupingUsed(FALSE);
1125                    DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(nf);
1126                    if (decfmt != NULL) {
1127                        decfmt->setDecimalSeparatorAlwaysShown(FALSE);
1128                    }
1129                    nf->setParseIntegerOnly(TRUE);
1130                    nf->setMinimumFractionDigits(0); // To prevent "Jan 1.00, 1997.00"
1131 
1132                    cur->nf = nf;
1133                    cur->hash = nsNameHash;
1134                    cur->next = fOverrideList;
1135                    fOverrideList = cur;
1136                }
1137                else {
1138                    // clean up before returning
1139                    if (cur != NULL) {
1140                        uprv_free(cur);
1141                    }
1142                   return;
1143                }
1144 
1145            } else {
1146                status = U_MEMORY_ALLOCATION_ERROR;
1147                return;
1148            }
1149         }
1150 
1151         // Now that we have an appropriate number formatter, fill in the appropriate spaces in the
1152         // number formatters table.
1153 
1154         if (ovrField.isBogus()) {
1155             switch (type) {
1156                 case kOvrStrDate:
1157                 case kOvrStrBoth: {
1158                     for ( int8_t i=0 ; i<kDateFieldsCount; i++ ) {
1159                         fNumberFormatters[kDateFields[i]] = nf;
1160                     }
1161                     if (type==kOvrStrDate) {
1162                         break;
1163                     }
1164                 }
1165                 case kOvrStrTime : {
1166                     for ( int8_t i=0 ; i<kTimeFieldsCount; i++ ) {
1167                         fNumberFormatters[kTimeFields[i]] = nf;
1168                     }
1169                     break;
1170                 }
1171             }
1172         } else {
1173            // if the pattern character is unrecognized, signal an error and bail out
1174            UDateFormatField patternCharIndex =
1175               DateFormatSymbols::getPatternCharIndex(ovrField.charAt(0));
1176            if (patternCharIndex == UDAT_FIELD_COUNT) {
1177                status = U_INVALID_FORMAT_ERROR;
1178                return;
1179            }
1180 
1181            // Set the number formatter in the table
1182            fNumberFormatters[patternCharIndex] = nf;
1183         }
1184 
1185         start = delimiterPosition + 1;
1186     }
1187 }
1188 
1189 //---------------------------------------------------------------------
1190 void
subFormat(UnicodeString & appendTo,UChar ch,int32_t count,UDisplayContext capitalizationContext,int32_t fieldNum,FieldPositionHandler & handler,Calendar & cal,UErrorCode & status) const1191 SimpleDateFormat::subFormat(UnicodeString &appendTo,
1192                             UChar ch,
1193                             int32_t count,
1194                             UDisplayContext capitalizationContext,
1195                             int32_t fieldNum,
1196                             FieldPositionHandler& handler,
1197                             Calendar& cal,
1198                             UErrorCode& status) const
1199 {
1200     if (U_FAILURE(status)) {
1201         return;
1202     }
1203 
1204     // this function gets called by format() to produce the appropriate substitution
1205     // text for an individual pattern symbol (e.g., "HH" or "yyyy")
1206 
1207     UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
1208     const int32_t maxIntCount = 10;
1209     int32_t beginOffset = appendTo.length();
1210     NumberFormat *currentNumberFormat;
1211     DateFormatSymbols::ECapitalizationContextUsageType capContextUsageType = DateFormatSymbols::kCapContextUsageOther;
1212 
1213     UBool isHebrewCalendar = (uprv_strcmp(cal.getType(),"hebrew") == 0);
1214     UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0 || uprv_strcmp(cal.getType(),"dangi") == 0);
1215 
1216     // if the pattern character is unrecognized, signal an error and dump out
1217     if (patternCharIndex == UDAT_FIELD_COUNT)
1218     {
1219         if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
1220             status = U_INVALID_FORMAT_ERROR;
1221         }
1222         return;
1223     }
1224 
1225     UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex];
1226     int32_t value = cal.get(field, status);
1227     if (U_FAILURE(status)) {
1228         return;
1229     }
1230 
1231     currentNumberFormat = getNumberFormatByIndex(patternCharIndex);
1232     UnicodeString hebr("hebr", 4, US_INV);
1233 
1234     switch (patternCharIndex) {
1235 
1236     // for any "G" symbol, write out the appropriate era string
1237     // "GGGG" is wide era name, "GGGGG" is narrow era name, anything else is abbreviated name
1238     case UDAT_ERA_FIELD:
1239         if (isChineseCalendar) {
1240             zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, 9); // as in ICU4J
1241         } else {
1242             if (count == 5) {
1243                 _appendSymbol(appendTo, value, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount);
1244                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraNarrow;
1245             } else if (count == 4) {
1246                 _appendSymbol(appendTo, value, fSymbols->fEraNames, fSymbols->fEraNamesCount);
1247                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraWide;
1248             } else {
1249                 _appendSymbol(appendTo, value, fSymbols->fEras, fSymbols->fErasCount);
1250                 capContextUsageType = DateFormatSymbols::kCapContextUsageEraAbbrev;
1251             }
1252         }
1253         break;
1254 
1255      case UDAT_YEAR_NAME_FIELD:
1256         if (fSymbols->fShortYearNames != NULL && value <= fSymbols->fShortYearNamesCount) {
1257             // the Calendar YEAR field runs 1 through 60 for cyclic years
1258             _appendSymbol(appendTo, value - 1, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount);
1259             break;
1260         }
1261         // else fall through to numeric year handling, do not break here
1262 
1263    // OLD: for "yyyy", write out the whole year; for "yy", write out the last 2 digits
1264     // NEW: UTS#35:
1265 //Year         y     yy     yyy     yyyy     yyyyy
1266 //AD 1         1     01     001     0001     00001
1267 //AD 12       12     12     012     0012     00012
1268 //AD 123     123     23     123     0123     00123
1269 //AD 1234   1234     34    1234     1234     01234
1270 //AD 12345 12345     45   12345    12345     12345
1271     case UDAT_YEAR_FIELD:
1272     case UDAT_YEAR_WOY_FIELD:
1273         if (fDateOverride.compare(hebr)==0 && value>HEBREW_CAL_CUR_MILLENIUM_START_YEAR && value<HEBREW_CAL_CUR_MILLENIUM_END_YEAR) {
1274             value-=HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
1275         }
1276         if(count == 2)
1277             zeroPaddingNumber(currentNumberFormat, appendTo, value, 2, 2);
1278         else
1279             zeroPaddingNumber(currentNumberFormat, appendTo, value, count, maxIntCount);
1280         break;
1281 
1282     // for "MMMM"/"LLLL", write out the whole month name, for "MMM"/"LLL", write out the month
1283     // abbreviation, for "M"/"L" or "MM"/"LL", write out the month as a number with the
1284     // appropriate number of digits
1285     // for "MMMMM"/"LLLLL", use the narrow form
1286     case UDAT_MONTH_FIELD:
1287     case UDAT_STANDALONE_MONTH_FIELD:
1288         if ( isHebrewCalendar ) {
1289            HebrewCalendar *hc = (HebrewCalendar*)&cal;
1290            if (hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value == 6 && count >= 3 )
1291                value = 13; // Show alternate form for Adar II in leap years in Hebrew calendar.
1292            if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6 && count < 3 )
1293                value--; // Adjust the month number down 1 in Hebrew non-leap years, i.e. Adar is 6, not 7.
1294         }
1295         {
1296             int32_t isLeapMonth = (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount)?
1297                         cal.get(UCAL_IS_LEAP_MONTH, status): 0;
1298             // should consolidate the next section by using arrays of pointers & counts for the right symbols...
1299             if (count == 5) {
1300                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1301                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fNarrowMonths, fSymbols->fNarrowMonthsCount,
1302                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatNarrow]): NULL, status);
1303                 } else {
1304                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneNarrowMonths, fSymbols->fStandaloneNarrowMonthsCount,
1305                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneNarrow]): NULL, status);
1306                 }
1307                 capContextUsageType = DateFormatSymbols::kCapContextUsageMonthNarrow;
1308             } else if (count == 4) {
1309                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1310                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fMonths, fSymbols->fMonthsCount,
1311                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide]): NULL, status);
1312                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1313                 } else {
1314                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount,
1315                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide]): NULL, status);
1316                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1317                 }
1318             } else if (count == 3) {
1319                 if (patternCharIndex == UDAT_MONTH_FIELD) {
1320                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fShortMonths, fSymbols->fShortMonthsCount,
1321                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev]): NULL, status);
1322                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthFormat;
1323                 } else {
1324                     _appendSymbolWithMonthPattern(appendTo, value, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount,
1325                             (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev]): NULL, status);
1326                     capContextUsageType = DateFormatSymbols::kCapContextUsageMonthStandalone;
1327                 }
1328             } else {
1329                 UnicodeString monthNumber;
1330                 zeroPaddingNumber(currentNumberFormat,monthNumber, value + 1, count, maxIntCount);
1331                 _appendSymbolWithMonthPattern(appendTo, 0, &monthNumber, 1,
1332                         (isLeapMonth!=0)? &(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric]): NULL, status);
1333             }
1334         }
1335         break;
1336 
1337     // for "k" and "kk", write out the hour, adjusting midnight to appear as "24"
1338     case UDAT_HOUR_OF_DAY1_FIELD:
1339         if (value == 0)
1340             zeroPaddingNumber(currentNumberFormat,appendTo, cal.getMaximum(UCAL_HOUR_OF_DAY) + 1, count, maxIntCount);
1341         else
1342             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1343         break;
1344 
1345     case UDAT_FRACTIONAL_SECOND_FIELD:
1346         // Fractional seconds left-justify
1347         {
1348             currentNumberFormat->setMinimumIntegerDigits((count > 3) ? 3 : count);
1349             currentNumberFormat->setMaximumIntegerDigits(maxIntCount);
1350             if (count == 1) {
1351                 value /= 100;
1352             } else if (count == 2) {
1353                 value /= 10;
1354             }
1355             FieldPosition p(0);
1356             currentNumberFormat->format(value, appendTo, p);
1357             if (count > 3) {
1358                 currentNumberFormat->setMinimumIntegerDigits(count - 3);
1359                 currentNumberFormat->format((int32_t)0, appendTo, p);
1360             }
1361         }
1362         break;
1363 
1364     // for "ee" or "e", use local numeric day-of-the-week
1365     // for "EEEEEE" or "eeeeee", write out the short day-of-the-week name
1366     // for "EEEEE" or "eeeee", write out the narrow day-of-the-week name
1367     // for "EEEE" or "eeee", write out the wide day-of-the-week name
1368     // for "EEE" or "EE" or "E" or "eee", write out the abbreviated day-of-the-week name
1369     case UDAT_DOW_LOCAL_FIELD:
1370         if ( count < 3 ) {
1371             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1372             break;
1373         }
1374         // fall through to EEEEE-EEE handling, but for that we don't want local day-of-week,
1375         // we want standard day-of-week, so first fix value to work for EEEEE-EEE.
1376         value = cal.get(UCAL_DAY_OF_WEEK, status);
1377         if (U_FAILURE(status)) {
1378             return;
1379         }
1380         // fall through, do not break here
1381     case UDAT_DAY_OF_WEEK_FIELD:
1382         if (count == 5) {
1383             _appendSymbol(appendTo, value, fSymbols->fNarrowWeekdays,
1384                           fSymbols->fNarrowWeekdaysCount);
1385             capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1386         } else if (count == 4) {
1387             _appendSymbol(appendTo, value, fSymbols->fWeekdays,
1388                           fSymbols->fWeekdaysCount);
1389             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1390         } else if (count == 6) {
1391             _appendSymbol(appendTo, value, fSymbols->fShorterWeekdays,
1392                           fSymbols->fShorterWeekdaysCount);
1393             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1394         } else {
1395             _appendSymbol(appendTo, value, fSymbols->fShortWeekdays,
1396                           fSymbols->fShortWeekdaysCount);
1397             capContextUsageType = DateFormatSymbols::kCapContextUsageDayFormat;
1398         }
1399         break;
1400 
1401     // for "ccc", write out the abbreviated day-of-the-week name
1402     // for "cccc", write out the wide day-of-the-week name
1403     // for "ccccc", use the narrow day-of-the-week name
1404     // for "ccccc", use the short day-of-the-week name
1405     case UDAT_STANDALONE_DAY_FIELD:
1406         if ( count < 3 ) {
1407             zeroPaddingNumber(currentNumberFormat,appendTo, value, 1, maxIntCount);
1408             break;
1409         }
1410         // fall through to alpha DOW handling, but for that we don't want local day-of-week,
1411         // we want standard day-of-week, so first fix value.
1412         value = cal.get(UCAL_DAY_OF_WEEK, status);
1413         if (U_FAILURE(status)) {
1414             return;
1415         }
1416         if (count == 5) {
1417             _appendSymbol(appendTo, value, fSymbols->fStandaloneNarrowWeekdays,
1418                           fSymbols->fStandaloneNarrowWeekdaysCount);
1419             capContextUsageType = DateFormatSymbols::kCapContextUsageDayNarrow;
1420         } else if (count == 4) {
1421             _appendSymbol(appendTo, value, fSymbols->fStandaloneWeekdays,
1422                           fSymbols->fStandaloneWeekdaysCount);
1423             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1424         } else if (count == 6) {
1425             _appendSymbol(appendTo, value, fSymbols->fStandaloneShorterWeekdays,
1426                           fSymbols->fStandaloneShorterWeekdaysCount);
1427             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1428         } else { // count == 3
1429             _appendSymbol(appendTo, value, fSymbols->fStandaloneShortWeekdays,
1430                           fSymbols->fStandaloneShortWeekdaysCount);
1431             capContextUsageType = DateFormatSymbols::kCapContextUsageDayStandalone;
1432         }
1433         break;
1434 
1435     // for and "a" symbol, write out the whole AM/PM string
1436     case UDAT_AM_PM_FIELD:
1437         _appendSymbol(appendTo, value, fSymbols->fAmPms,
1438                       fSymbols->fAmPmsCount);
1439         break;
1440 
1441     // for "h" and "hh", write out the hour, adjusting noon and midnight to show up
1442     // as "12"
1443     case UDAT_HOUR1_FIELD:
1444         if (value == 0)
1445             zeroPaddingNumber(currentNumberFormat,appendTo, cal.getLeastMaximum(UCAL_HOUR) + 1, count, maxIntCount);
1446         else
1447             zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1448         break;
1449 
1450     case UDAT_TIMEZONE_FIELD: // 'z'
1451     case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
1452     case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
1453     case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
1454     case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
1455     case UDAT_TIMEZONE_ISO_FIELD: // 'X'
1456     case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
1457         {
1458             UnicodeString zoneString;
1459             const TimeZone& tz = cal.getTimeZone();
1460             UDate date = cal.getTime(status);
1461             if (U_SUCCESS(status)) {
1462                 if (patternCharIndex == UDAT_TIMEZONE_FIELD) {
1463                     if (count < 4) {
1464                         // "z", "zz", "zzz"
1465                         tzFormat()->format(UTZFMT_STYLE_SPECIFIC_SHORT, tz, date, zoneString);
1466                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1467                     } else {
1468                         // "zzzz" or longer
1469                         tzFormat()->format(UTZFMT_STYLE_SPECIFIC_LONG, tz, date, zoneString);
1470                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1471                     }
1472                 }
1473                 else if (patternCharIndex == UDAT_TIMEZONE_RFC_FIELD) {
1474                     if (count < 4) {
1475                         // "Z"
1476                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
1477                     } else if (count == 5) {
1478                         // "ZZZZZ"
1479                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
1480                     } else {
1481                         // "ZZ", "ZZZ", "ZZZZ"
1482                         tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
1483                     }
1484                 }
1485                 else if (patternCharIndex == UDAT_TIMEZONE_GENERIC_FIELD) {
1486                     if (count == 1) {
1487                         // "v"
1488                         tzFormat()->format(UTZFMT_STYLE_GENERIC_SHORT, tz, date, zoneString);
1489                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneShort;
1490                     } else if (count == 4) {
1491                         // "vvvv"
1492                         tzFormat()->format(UTZFMT_STYLE_GENERIC_LONG, tz, date, zoneString);
1493                         capContextUsageType = DateFormatSymbols::kCapContextUsageMetazoneLong;
1494                     }
1495                 }
1496                 else if (patternCharIndex == UDAT_TIMEZONE_SPECIAL_FIELD) {
1497                     if (count == 1) {
1498                         // "V"
1499                         tzFormat()->format(UTZFMT_STYLE_ZONE_ID_SHORT, tz, date, zoneString);
1500                     } else if (count == 2) {
1501                         // "VV"
1502                         tzFormat()->format(UTZFMT_STYLE_ZONE_ID, tz, date, zoneString);
1503                     } else if (count == 3) {
1504                         // "VVV"
1505                         tzFormat()->format(UTZFMT_STYLE_EXEMPLAR_LOCATION, tz, date, zoneString);
1506                     } else if (count == 4) {
1507                         // "VVVV"
1508                         tzFormat()->format(UTZFMT_STYLE_GENERIC_LOCATION, tz, date, zoneString);
1509                         capContextUsageType = DateFormatSymbols::kCapContextUsageZoneLong;
1510                     }
1511                 }
1512                 else if (patternCharIndex == UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD) {
1513                     if (count == 1) {
1514                         // "O"
1515                         tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT_SHORT, tz, date, zoneString);
1516                     } else if (count == 4) {
1517                         // "OOOO"
1518                         tzFormat()->format(UTZFMT_STYLE_LOCALIZED_GMT, tz, date, zoneString);
1519                     }
1520                 }
1521                 else if (patternCharIndex == UDAT_TIMEZONE_ISO_FIELD) {
1522                     if (count == 1) {
1523                         // "X"
1524                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_SHORT, tz, date, zoneString);
1525                     } else if (count == 2) {
1526                         // "XX"
1527                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_FIXED, tz, date, zoneString);
1528                     } else if (count == 3) {
1529                         // "XXX"
1530                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FIXED, tz, date, zoneString);
1531                     } else if (count == 4) {
1532                         // "XXXX"
1533                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_FULL, tz, date, zoneString);
1534                     } else if (count == 5) {
1535                         // "XXXXX"
1536                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_FULL, tz, date, zoneString);
1537                     }
1538                 }
1539                 else if (patternCharIndex == UDAT_TIMEZONE_ISO_LOCAL_FIELD) {
1540                     if (count == 1) {
1541                         // "x"
1542                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT, tz, date, zoneString);
1543                     } else if (count == 2) {
1544                         // "xx"
1545                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED, tz, date, zoneString);
1546                     } else if (count == 3) {
1547                         // "xxx"
1548                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED, tz, date, zoneString);
1549                     } else if (count == 4) {
1550                         // "xxxx"
1551                         tzFormat()->format(UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL, tz, date, zoneString);
1552                     } else if (count == 5) {
1553                         // "xxxxx"
1554                         tzFormat()->format(UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL, tz, date, zoneString);
1555                     }
1556                 }
1557                 else {
1558                     U_ASSERT(FALSE);
1559                 }
1560             }
1561             appendTo += zoneString;
1562         }
1563         break;
1564 
1565     case UDAT_QUARTER_FIELD:
1566         if (count >= 4)
1567             _appendSymbol(appendTo, value/3, fSymbols->fQuarters,
1568                           fSymbols->fQuartersCount);
1569         else if (count == 3)
1570             _appendSymbol(appendTo, value/3, fSymbols->fShortQuarters,
1571                           fSymbols->fShortQuartersCount);
1572         else
1573             zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1574         break;
1575 
1576     case UDAT_STANDALONE_QUARTER_FIELD:
1577         if (count >= 4)
1578             _appendSymbol(appendTo, value/3, fSymbols->fStandaloneQuarters,
1579                           fSymbols->fStandaloneQuartersCount);
1580         else if (count == 3)
1581             _appendSymbol(appendTo, value/3, fSymbols->fStandaloneShortQuarters,
1582                           fSymbols->fStandaloneShortQuartersCount);
1583         else
1584             zeroPaddingNumber(currentNumberFormat,appendTo, (value/3) + 1, count, maxIntCount);
1585         break;
1586 
1587 
1588     // all of the other pattern symbols can be formatted as simple numbers with
1589     // appropriate zero padding
1590     default:
1591         zeroPaddingNumber(currentNumberFormat,appendTo, value, count, maxIntCount);
1592         break;
1593     }
1594 #if !UCONFIG_NO_BREAK_ITERATION
1595     if (fieldNum == 0) {
1596         // first field, check to see whether we need to titlecase it
1597         UBool titlecase = FALSE;
1598         switch (capitalizationContext) {
1599             case UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE:
1600                 titlecase = TRUE;
1601                 break;
1602             case UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU:
1603                 titlecase = fSymbols->fCapitalization[capContextUsageType][0];
1604                 break;
1605             case UDISPCTX_CAPITALIZATION_FOR_STANDALONE:
1606                 titlecase = fSymbols->fCapitalization[capContextUsageType][1];
1607                 break;
1608             default:
1609                 // titlecase = FALSE;
1610                 break;
1611         }
1612         if (titlecase) {
1613             UnicodeString firstField(appendTo, beginOffset);
1614             firstField.toTitle(NULL, fLocale, U_TITLECASE_NO_LOWERCASE | U_TITLECASE_NO_BREAK_ADJUSTMENT);
1615             appendTo.replaceBetween(beginOffset, appendTo.length(), firstField);
1616         }
1617     }
1618 #endif
1619 
1620     handler.addAttribute(fgPatternIndexToDateFormatField[patternCharIndex], beginOffset, appendTo.length());
1621 }
1622 
1623 //----------------------------------------------------------------------
1624 
1625 NumberFormat *
getNumberFormatByIndex(UDateFormatField index) const1626 SimpleDateFormat::getNumberFormatByIndex(UDateFormatField index) const {
1627     if (fNumberFormatters != NULL) {
1628         return fNumberFormatters[index];
1629     } else {
1630         return fNumberFormat;
1631     }
1632 }
1633 
1634 //----------------------------------------------------------------------
1635 void
zeroPaddingNumber(NumberFormat * currentNumberFormat,UnicodeString & appendTo,int32_t value,int32_t minDigits,int32_t maxDigits) const1636 SimpleDateFormat::zeroPaddingNumber(NumberFormat *currentNumberFormat,UnicodeString &appendTo,
1637                                     int32_t value, int32_t minDigits, int32_t maxDigits) const
1638 {
1639     if (currentNumberFormat!=NULL) {
1640         FieldPosition pos(0);
1641 
1642         currentNumberFormat->setMinimumIntegerDigits(minDigits);
1643         currentNumberFormat->setMaximumIntegerDigits(maxDigits);
1644         currentNumberFormat->format(value, appendTo, pos);  // 3rd arg is there to speed up processing
1645     }
1646 }
1647 
1648 //----------------------------------------------------------------------
1649 
1650 /**
1651  * Return true if the given format character, occuring count
1652  * times, represents a numeric field.
1653  */
isNumeric(UChar formatChar,int32_t count)1654 UBool SimpleDateFormat::isNumeric(UChar formatChar, int32_t count) {
1655     return DateFormatSymbols::isNumericPatternChar(formatChar, count);
1656 }
1657 
1658 UBool
isAtNumericField(const UnicodeString & pattern,int32_t patternOffset)1659 SimpleDateFormat::isAtNumericField(const UnicodeString &pattern, int32_t patternOffset) {
1660     if (patternOffset >= pattern.length()) {
1661         // not at any field
1662         return FALSE;
1663     }
1664     UChar ch = pattern.charAt(patternOffset);
1665     UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
1666     if (f == UDAT_FIELD_COUNT) {
1667         // not at any field
1668         return FALSE;
1669     }
1670     int32_t i = patternOffset;
1671     while (pattern.charAt(++i) == ch) {}
1672     return DateFormatSymbols::isNumericField(f, i - patternOffset);
1673 }
1674 
1675 UBool
isAfterNonNumericField(const UnicodeString & pattern,int32_t patternOffset)1676 SimpleDateFormat::isAfterNonNumericField(const UnicodeString &pattern, int32_t patternOffset) {
1677     if (patternOffset <= 0) {
1678         // not after any field
1679         return FALSE;
1680     }
1681     UChar ch = pattern.charAt(--patternOffset);
1682     UDateFormatField f = DateFormatSymbols::getPatternCharIndex(ch);
1683     if (f == UDAT_FIELD_COUNT) {
1684         // not after any field
1685         return FALSE;
1686     }
1687     int32_t i = patternOffset;
1688     while (pattern.charAt(--i) == ch) {}
1689     return !DateFormatSymbols::isNumericField(f, patternOffset - i);
1690 }
1691 
1692 void
parse(const UnicodeString & text,Calendar & cal,ParsePosition & parsePos) const1693 SimpleDateFormat::parse(const UnicodeString& text, Calendar& cal, ParsePosition& parsePos) const
1694 {
1695     UErrorCode status = U_ZERO_ERROR;
1696     int32_t pos = parsePos.getIndex();
1697     int32_t start = pos;
1698 
1699     UBool ambiguousYear[] = { FALSE };
1700     int32_t saveHebrewMonth = -1;
1701     int32_t count = 0;
1702 
1703     // hack, reset tztype, cast away const
1704     ((SimpleDateFormat*)this)->tztype = UTZFMT_TIME_TYPE_UNKNOWN;
1705 
1706     // For parsing abutting numeric fields. 'abutPat' is the
1707     // offset into 'pattern' of the first of 2 or more abutting
1708     // numeric fields.  'abutStart' is the offset into 'text'
1709     // where parsing the fields begins. 'abutPass' starts off as 0
1710     // and increments each time we try to parse the fields.
1711     int32_t abutPat = -1; // If >=0, we are in a run of abutting numeric fields
1712     int32_t abutStart = 0;
1713     int32_t abutPass = 0;
1714     UBool inQuote = FALSE;
1715 
1716     MessageFormat * numericLeapMonthFormatter = NULL;
1717 
1718     Calendar* calClone = NULL;
1719     Calendar *workCal = &cal;
1720     if (&cal != fCalendar && uprv_strcmp(cal.getType(), fCalendar->getType()) != 0) {
1721         // Different calendar type
1722         // We use the time/zone from the input calendar, but
1723         // do not use the input calendar for field calculation.
1724         calClone = fCalendar->clone();
1725         if (calClone != NULL) {
1726             calClone->setTime(cal.getTime(status),status);
1727             if (U_FAILURE(status)) {
1728                 goto ExitParse;
1729             }
1730             calClone->setTimeZone(cal.getTimeZone());
1731             workCal = calClone;
1732         } else {
1733             status = U_MEMORY_ALLOCATION_ERROR;
1734             goto ExitParse;
1735         }
1736     }
1737 
1738     if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
1739         numericLeapMonthFormatter = new MessageFormat(fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternNumeric], fLocale, status);
1740         if (numericLeapMonthFormatter == NULL) {
1741              status = U_MEMORY_ALLOCATION_ERROR;
1742              goto ExitParse;
1743         } else if (U_FAILURE(status)) {
1744              goto ExitParse; // this will delete numericLeapMonthFormatter
1745         }
1746     }
1747 
1748     for (int32_t i=0; i<fPattern.length(); ++i) {
1749         UChar ch = fPattern.charAt(i);
1750 
1751         // Handle alphabetic field characters.
1752         if (!inQuote && ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { // [A-Za-z]
1753             int32_t fieldPat = i;
1754 
1755             // Count the length of this field specifier
1756             count = 1;
1757             while ((i+1)<fPattern.length() &&
1758                    fPattern.charAt(i+1) == ch) {
1759                 ++count;
1760                 ++i;
1761             }
1762 
1763             if (isNumeric(ch, count)) {
1764                 if (abutPat < 0) {
1765                     // Determine if there is an abutting numeric field.
1766                     // Record the start of a set of abutting numeric fields.
1767                     if (isAtNumericField(fPattern, i + 1)) {
1768                         abutPat = fieldPat;
1769                         abutStart = pos;
1770                         abutPass = 0;
1771                     }
1772                 }
1773             } else {
1774                 abutPat = -1; // End of any abutting fields
1775             }
1776 
1777             // Handle fields within a run of abutting numeric fields.  Take
1778             // the pattern "HHmmss" as an example. We will try to parse
1779             // 2/2/2 characters of the input text, then if that fails,
1780             // 1/2/2.  We only adjust the width of the leftmost field; the
1781             // others remain fixed.  This allows "123456" => 12:34:56, but
1782             // "12345" => 1:23:45.  Likewise, for the pattern "yyyyMMdd" we
1783             // try 4/2/2, 3/2/2, 2/2/2, and finally 1/2/2.
1784             if (abutPat >= 0) {
1785                 // If we are at the start of a run of abutting fields, then
1786                 // shorten this field in each pass.  If we can't shorten
1787                 // this field any more, then the parse of this set of
1788                 // abutting numeric fields has failed.
1789                 if (fieldPat == abutPat) {
1790                     count -= abutPass++;
1791                     if (count == 0) {
1792                         status = U_PARSE_ERROR;
1793                         goto ExitParse;
1794                     }
1795                 }
1796 
1797                 pos = subParse(text, pos, ch, count,
1798                                TRUE, FALSE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter);
1799 
1800                 // If the parse fails anywhere in the run, back up to the
1801                 // start of the run and retry.
1802                 if (pos < 0) {
1803                     i = abutPat - 1;
1804                     pos = abutStart;
1805                     continue;
1806                 }
1807             }
1808 
1809             // Handle non-numeric fields and non-abutting numeric
1810             // fields.
1811             else if (ch != 0x6C) { // pattern char 'l' (SMALL LETTER L) just gets ignored
1812                 int32_t s = subParse(text, pos, ch, count,
1813                                FALSE, TRUE, ambiguousYear, saveHebrewMonth, *workCal, i, numericLeapMonthFormatter);
1814 
1815                 if (s == -pos-1) {
1816                     // era not present, in special cases allow this to continue
1817                     // from the position where the era was expected
1818                     s = pos;
1819 
1820                     if (i+1 < fPattern.length()) {
1821                         // move to next pattern character
1822                         UChar ch = fPattern.charAt(i+1);
1823 
1824                         // check for whitespace
1825                         if (PatternProps::isWhiteSpace(ch)) {
1826                             i++;
1827                             // Advance over run in pattern
1828                             while ((i+1)<fPattern.length() &&
1829                                    PatternProps::isWhiteSpace(fPattern.charAt(i+1))) {
1830                                 ++i;
1831                             }
1832                         }
1833                     }
1834                 }
1835                 else if (s <= 0) {
1836                     status = U_PARSE_ERROR;
1837                     goto ExitParse;
1838                 }
1839                 pos = s;
1840             }
1841         }
1842 
1843         // Handle literal pattern characters.  These are any
1844         // quoted characters and non-alphabetic unquoted
1845         // characters.
1846         else {
1847 
1848             abutPat = -1; // End of any abutting fields
1849 
1850             if (! matchLiterals(fPattern, i, text, pos, getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status))) {
1851                 status = U_PARSE_ERROR;
1852                 goto ExitParse;
1853             }
1854         }
1855     }
1856 
1857     // Special hack for trailing "." after non-numeric field.
1858     if (text.charAt(pos) == 0x2e && getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) {
1859         // only do if the last field is not numeric
1860         if (isAfterNonNumericField(fPattern, fPattern.length())) {
1861             pos++; // skip the extra "."
1862         }
1863     }
1864 
1865     // At this point the fields of Calendar have been set.  Calendar
1866     // will fill in default values for missing fields when the time
1867     // is computed.
1868 
1869     parsePos.setIndex(pos);
1870 
1871     // This part is a problem:  When we call parsedDate.after, we compute the time.
1872     // Take the date April 3 2004 at 2:30 am.  When this is first set up, the year
1873     // will be wrong if we're parsing a 2-digit year pattern.  It will be 1904.
1874     // April 3 1904 is a Sunday (unlike 2004) so it is the DST onset day.  2:30 am
1875     // is therefore an "impossible" time, since the time goes from 1:59 to 3:00 am
1876     // on that day.  It is therefore parsed out to fields as 3:30 am.  Then we
1877     // add 100 years, and get April 3 2004 at 3:30 am.  Note that April 3 2004 is
1878     // a Saturday, so it can have a 2:30 am -- and it should. [LIU]
1879     /*
1880         UDate parsedDate = calendar.getTime();
1881         if( ambiguousYear[0] && !parsedDate.after(fDefaultCenturyStart) ) {
1882             calendar.add(Calendar.YEAR, 100);
1883             parsedDate = calendar.getTime();
1884         }
1885     */
1886     // Because of the above condition, save off the fields in case we need to readjust.
1887     // The procedure we use here is not particularly efficient, but there is no other
1888     // way to do this given the API restrictions present in Calendar.  We minimize
1889     // inefficiency by only performing this computation when it might apply, that is,
1890     // when the two-digit year is equal to the start year, and thus might fall at the
1891     // front or the back of the default century.  This only works because we adjust
1892     // the year correctly to start with in other cases -- see subParse().
1893     if (ambiguousYear[0] || tztype != UTZFMT_TIME_TYPE_UNKNOWN) // If this is true then the two-digit year == the default start year
1894     {
1895         // We need a copy of the fields, and we need to avoid triggering a call to
1896         // complete(), which will recalculate the fields.  Since we can't access
1897         // the fields[] array in Calendar, we clone the entire object.  This will
1898         // stop working if Calendar.clone() is ever rewritten to call complete().
1899         Calendar *copy;
1900         if (ambiguousYear[0]) {
1901             copy = cal.clone();
1902             // Check for failed cloning.
1903             if (copy == NULL) {
1904                 status = U_MEMORY_ALLOCATION_ERROR;
1905                 goto ExitParse;
1906             }
1907             UDate parsedDate = copy->getTime(status);
1908             // {sfb} check internalGetDefaultCenturyStart
1909             if (fHaveDefaultCentury && (parsedDate < fDefaultCenturyStart)) {
1910                 // We can't use add here because that does a complete() first.
1911                 cal.set(UCAL_YEAR, fDefaultCenturyStartYear + 100);
1912             }
1913             delete copy;
1914         }
1915 
1916         if (tztype != UTZFMT_TIME_TYPE_UNKNOWN) {
1917             copy = cal.clone();
1918             // Check for failed cloning.
1919             if (copy == NULL) {
1920                 status = U_MEMORY_ALLOCATION_ERROR;
1921                 goto ExitParse;
1922             }
1923             const TimeZone & tz = cal.getTimeZone();
1924             BasicTimeZone *btz = NULL;
1925 
1926             if (dynamic_cast<const OlsonTimeZone *>(&tz) != NULL
1927                 || dynamic_cast<const SimpleTimeZone *>(&tz) != NULL
1928                 || dynamic_cast<const RuleBasedTimeZone *>(&tz) != NULL
1929                 || dynamic_cast<const VTimeZone *>(&tz) != NULL) {
1930                 btz = (BasicTimeZone*)&tz;
1931             }
1932 
1933             // Get local millis
1934             copy->set(UCAL_ZONE_OFFSET, 0);
1935             copy->set(UCAL_DST_OFFSET, 0);
1936             UDate localMillis = copy->getTime(status);
1937 
1938             // Make sure parsed time zone type (Standard or Daylight)
1939             // matches the rule used by the parsed time zone.
1940             int32_t raw, dst;
1941             if (btz != NULL) {
1942                 if (tztype == UTZFMT_TIME_TYPE_STANDARD) {
1943                     btz->getOffsetFromLocal(localMillis,
1944                         BasicTimeZone::kStandard, BasicTimeZone::kStandard, raw, dst, status);
1945                 } else {
1946                     btz->getOffsetFromLocal(localMillis,
1947                         BasicTimeZone::kDaylight, BasicTimeZone::kDaylight, raw, dst, status);
1948                 }
1949             } else {
1950                 // No good way to resolve ambiguous time at transition,
1951                 // but following code work in most case.
1952                 tz.getOffset(localMillis, TRUE, raw, dst, status);
1953             }
1954 
1955             // Now, compare the results with parsed type, either standard or daylight saving time
1956             int32_t resolvedSavings = dst;
1957             if (tztype == UTZFMT_TIME_TYPE_STANDARD) {
1958                 if (dst != 0) {
1959                     // Override DST_OFFSET = 0 in the result calendar
1960                     resolvedSavings = 0;
1961                 }
1962             } else { // tztype == TZTYPE_DST
1963                 if (dst == 0) {
1964                     if (btz != NULL) {
1965                         UDate time = localMillis + raw;
1966                         // We use the nearest daylight saving time rule.
1967                         TimeZoneTransition beforeTrs, afterTrs;
1968                         UDate beforeT = time, afterT = time;
1969                         int32_t beforeSav = 0, afterSav = 0;
1970                         UBool beforeTrsAvail, afterTrsAvail;
1971 
1972                         // Search for DST rule before or on the time
1973                         while (TRUE) {
1974                             beforeTrsAvail = btz->getPreviousTransition(beforeT, TRUE, beforeTrs);
1975                             if (!beforeTrsAvail) {
1976                                 break;
1977                             }
1978                             beforeT = beforeTrs.getTime() - 1;
1979                             beforeSav = beforeTrs.getFrom()->getDSTSavings();
1980                             if (beforeSav != 0) {
1981                                 break;
1982                             }
1983                         }
1984 
1985                         // Search for DST rule after the time
1986                         while (TRUE) {
1987                             afterTrsAvail = btz->getNextTransition(afterT, FALSE, afterTrs);
1988                             if (!afterTrsAvail) {
1989                                 break;
1990                             }
1991                             afterT = afterTrs.getTime();
1992                             afterSav = afterTrs.getTo()->getDSTSavings();
1993                             if (afterSav != 0) {
1994                                 break;
1995                             }
1996                         }
1997 
1998                         if (beforeTrsAvail && afterTrsAvail) {
1999                             if (time - beforeT > afterT - time) {
2000                                 resolvedSavings = afterSav;
2001                             } else {
2002                                 resolvedSavings = beforeSav;
2003                             }
2004                         } else if (beforeTrsAvail && beforeSav != 0) {
2005                             resolvedSavings = beforeSav;
2006                         } else if (afterTrsAvail && afterSav != 0) {
2007                             resolvedSavings = afterSav;
2008                         } else {
2009                             resolvedSavings = btz->getDSTSavings();
2010                         }
2011                     } else {
2012                         resolvedSavings = tz.getDSTSavings();
2013                     }
2014                     if (resolvedSavings == 0) {
2015                         // final fallback
2016                         resolvedSavings = U_MILLIS_PER_HOUR;
2017                     }
2018                 }
2019             }
2020             cal.set(UCAL_ZONE_OFFSET, raw);
2021             cal.set(UCAL_DST_OFFSET, resolvedSavings);
2022             delete copy;
2023         }
2024     }
2025 ExitParse:
2026     // Set the parsed result if local calendar is used
2027     // instead of the input calendar
2028     if (U_SUCCESS(status) && workCal != &cal) {
2029         cal.setTimeZone(workCal->getTimeZone());
2030         cal.setTime(workCal->getTime(status), status);
2031     }
2032 
2033     if (numericLeapMonthFormatter != NULL) {
2034         delete numericLeapMonthFormatter;
2035     }
2036     if (calClone != NULL) {
2037         delete calClone;
2038     }
2039 
2040     // If any Calendar calls failed, we pretend that we
2041     // couldn't parse the string, when in reality this isn't quite accurate--
2042     // we did parse it; the Calendar calls just failed.
2043     if (U_FAILURE(status)) {
2044         parsePos.setErrorIndex(pos);
2045         parsePos.setIndex(start);
2046     }
2047 }
2048 
2049 //----------------------------------------------------------------------
2050 
2051 static UBool
2052 newBestMatchWithOptionalDot(const UnicodeString &lcaseText,
2053                             const UnicodeString &data,
2054                             UnicodeString &bestMatchName,
2055                             int32_t &bestMatchLength);
2056 
matchQuarterString(const UnicodeString & text,int32_t start,UCalendarDateFields field,const UnicodeString * data,int32_t dataCount,Calendar & cal) const2057 int32_t SimpleDateFormat::matchQuarterString(const UnicodeString& text,
2058                               int32_t start,
2059                               UCalendarDateFields field,
2060                               const UnicodeString* data,
2061                               int32_t dataCount,
2062                               Calendar& cal) const
2063 {
2064     int32_t i = 0;
2065     int32_t count = dataCount;
2066 
2067     // There may be multiple strings in the data[] array which begin with
2068     // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2069     // We keep track of the longest match, and return that.  Note that this
2070     // unfortunately requires us to test all array elements.
2071     int32_t bestMatchLength = 0, bestMatch = -1;
2072     UnicodeString bestMatchName;
2073 
2074     // {sfb} kludge to support case-insensitive comparison
2075     // {markus 2002oct11} do not just use caseCompareBetween because we do not know
2076     // the length of the match after case folding
2077     // {alan 20040607} don't case change the whole string, since the length
2078     // can change
2079     // TODO we need a case-insensitive startsWith function
2080     UnicodeString lcaseText;
2081     text.extract(start, INT32_MAX, lcaseText);
2082     lcaseText.foldCase();
2083 
2084     for (; i < count; ++i)
2085     {
2086         // Always compare if we have no match yet; otherwise only compare
2087         // against potentially better matches (longer strings).
2088 
2089         if (newBestMatchWithOptionalDot(lcaseText, data[i], bestMatchName, bestMatchLength)) {
2090             bestMatch = i;
2091         }
2092     }
2093     if (bestMatch >= 0)
2094     {
2095         cal.set(field, bestMatch * 3);
2096 
2097         // Once we have a match, we have to determine the length of the
2098         // original source string.  This will usually be == the length of
2099         // the case folded string, but it may differ (e.g. sharp s).
2100 
2101         // Most of the time, the length will be the same as the length
2102         // of the string from the locale data.  Sometimes it will be
2103         // different, in which case we will have to figure it out by
2104         // adding a character at a time, until we have a match.  We do
2105         // this all in one loop, where we try 'len' first (at index
2106         // i==0).
2107         int32_t len = bestMatchName.length(); // 99+% of the time
2108         int32_t n = text.length() - start;
2109         for (i=0; i<=n; ++i) {
2110             int32_t j=i;
2111             if (i == 0) {
2112                 j = len;
2113             } else if (i == len) {
2114                 continue; // already tried this when i was 0
2115             }
2116             text.extract(start, j, lcaseText);
2117             lcaseText.foldCase();
2118             if (bestMatchName == lcaseText) {
2119                 return start + j;
2120             }
2121         }
2122     }
2123 
2124     return -start;
2125 }
2126 
2127 //----------------------------------------------------------------------
matchLiterals(const UnicodeString & pattern,int32_t & patternOffset,const UnicodeString & text,int32_t & textOffset,UBool lenient)2128 UBool SimpleDateFormat::matchLiterals(const UnicodeString &pattern,
2129                                       int32_t &patternOffset,
2130                                       const UnicodeString &text,
2131                                       int32_t &textOffset,
2132                                       UBool lenient)
2133 {
2134     UBool inQuote = FALSE;
2135     UnicodeString literal;
2136     int32_t i = patternOffset;
2137 
2138     // scan pattern looking for contiguous literal characters
2139     for ( ; i < pattern.length(); i += 1) {
2140         UChar ch = pattern.charAt(i);
2141 
2142         if (!inQuote && ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { // unquoted [A-Za-z]
2143             break;
2144         }
2145 
2146         if (ch == QUOTE) {
2147             // Match a quote literal ('') inside OR outside of quotes
2148             if ((i + 1) < pattern.length() && pattern.charAt(i + 1) == QUOTE) {
2149                 i += 1;
2150             } else {
2151                 inQuote = !inQuote;
2152                 continue;
2153             }
2154         }
2155 
2156         literal += ch;
2157     }
2158 
2159     // at this point, literal contains the literal text
2160     // and i is the index of the next non-literal pattern character.
2161     int32_t p;
2162     int32_t t = textOffset;
2163 
2164     if (lenient) {
2165         // trim leading, trailing whitespace from
2166         // the literal text
2167         literal.trim();
2168 
2169         // ignore any leading whitespace in the text
2170         while (t < text.length() && u_isWhitespace(text.charAt(t))) {
2171             t += 1;
2172         }
2173     }
2174 
2175     for (p = 0; p < literal.length() && t < text.length();) {
2176         UBool needWhitespace = FALSE;
2177 
2178         while (p < literal.length() && PatternProps::isWhiteSpace(literal.charAt(p))) {
2179             needWhitespace = TRUE;
2180             p += 1;
2181         }
2182 
2183         if (needWhitespace) {
2184             int32_t tStart = t;
2185 
2186             while (t < text.length()) {
2187                 UChar tch = text.charAt(t);
2188 
2189                 if (!u_isUWhiteSpace(tch) && !PatternProps::isWhiteSpace(tch)) {
2190                     break;
2191                 }
2192 
2193                 t += 1;
2194             }
2195 
2196             // TODO: should we require internal spaces
2197             // in lenient mode? (There won't be any
2198             // leading or trailing spaces)
2199             if (!lenient && t == tStart) {
2200                 // didn't find matching whitespace:
2201                 // an error in strict mode
2202                 return FALSE;
2203             }
2204 
2205             // In strict mode, this run of whitespace
2206             // may have been at the end.
2207             if (p >= literal.length()) {
2208                 break;
2209             }
2210         }
2211 
2212         if (t >= text.length() || literal.charAt(p) != text.charAt(t)) {
2213             // Ran out of text, or found a non-matching character:
2214             // OK in lenient mode, an error in strict mode.
2215             if (lenient) {
2216                 if (t == textOffset && text.charAt(t) == 0x2e &&
2217                         isAfterNonNumericField(pattern, patternOffset)) {
2218                     // Lenient mode and the literal input text begins with a "." and
2219                     // we are after a non-numeric field: We skip the "."
2220                     ++t;
2221                     continue;  // Do not update p.
2222                 }
2223                 break;
2224             }
2225 
2226             return FALSE;
2227         }
2228         ++p;
2229         ++t;
2230     }
2231 
2232     // At this point if we're in strict mode we have a complete match.
2233     // If we're in lenient mode we may have a partial match, or no
2234     // match at all.
2235     if (p <= 0) {
2236         // no match. Pretend it matched a run of whitespace
2237         // and ignorables in the text.
2238         const  UnicodeSet *ignorables = NULL;
2239         UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(pattern.charAt(i));
2240         if (patternCharIndex != UDAT_FIELD_COUNT) {
2241             ignorables = SimpleDateFormatStaticSets::getIgnorables(patternCharIndex);
2242         }
2243 
2244         for (t = textOffset; t < text.length(); t += 1) {
2245             UChar ch = text.charAt(t);
2246 
2247             if (ignorables == NULL || !ignorables->contains(ch)) {
2248                 break;
2249             }
2250         }
2251     }
2252 
2253     // if we get here, we've got a complete match.
2254     patternOffset = i - 1;
2255     textOffset = t;
2256 
2257     return TRUE;
2258 }
2259 
2260 //----------------------------------------------------------------------
2261 
matchString(const UnicodeString & text,int32_t start,UCalendarDateFields field,const UnicodeString * data,int32_t dataCount,const UnicodeString * monthPattern,Calendar & cal) const2262 int32_t SimpleDateFormat::matchString(const UnicodeString& text,
2263                               int32_t start,
2264                               UCalendarDateFields field,
2265                               const UnicodeString* data,
2266                               int32_t dataCount,
2267                               const UnicodeString* monthPattern,
2268                               Calendar& cal) const
2269 {
2270     int32_t i = 0;
2271     int32_t count = dataCount;
2272 
2273     if (field == UCAL_DAY_OF_WEEK) i = 1;
2274 
2275     // There may be multiple strings in the data[] array which begin with
2276     // the same prefix (e.g., Cerven and Cervenec (June and July) in Czech).
2277     // We keep track of the longest match, and return that.  Note that this
2278     // unfortunately requires us to test all array elements.
2279     int32_t bestMatchLength = 0, bestMatch = -1;
2280     UnicodeString bestMatchName;
2281     int32_t isLeapMonth = 0;
2282 
2283     // {sfb} kludge to support case-insensitive comparison
2284     // {markus 2002oct11} do not just use caseCompareBetween because we do not know
2285     // the length of the match after case folding
2286     // {alan 20040607} don't case change the whole string, since the length
2287     // can change
2288     // TODO we need a case-insensitive startsWith function
2289     UnicodeString lcaseText;
2290     text.extract(start, INT32_MAX, lcaseText);
2291     lcaseText.foldCase();
2292 
2293     for (; i < count; ++i)
2294     {
2295         // Always compare if we have no match yet; otherwise only compare
2296         // against potentially better matches (longer strings).
2297 
2298         if (newBestMatchWithOptionalDot(lcaseText, data[i], bestMatchName, bestMatchLength)) {
2299             bestMatch = i;
2300             isLeapMonth = 0;
2301         }
2302 
2303         if (monthPattern != NULL) {
2304             UErrorCode status = U_ZERO_ERROR;
2305             UnicodeString leapMonthName;
2306             Formattable monthName((const UnicodeString&)(data[i]));
2307             MessageFormat::format(*monthPattern, &monthName, 1, leapMonthName, status);
2308             if (U_SUCCESS(status)) {
2309                 if (newBestMatchWithOptionalDot(lcaseText, leapMonthName, bestMatchName, bestMatchLength)) {
2310                     bestMatch = i;
2311                     isLeapMonth = 1;
2312                 }
2313             }
2314         }
2315     }
2316     if (bestMatch >= 0)
2317     {
2318         // Adjustment for Hebrew Calendar month Adar II
2319         if (!strcmp(cal.getType(),"hebrew") && field==UCAL_MONTH && bestMatch==13) {
2320             cal.set(field,6);
2321         }
2322         else {
2323             if (field == UCAL_YEAR) {
2324                 bestMatch++; // only get here for cyclic year names, which match 1-based years 1-60
2325             }
2326             cal.set(field, bestMatch);
2327         }
2328         if (monthPattern != NULL) {
2329             cal.set(UCAL_IS_LEAP_MONTH, isLeapMonth);
2330         }
2331 
2332         // Once we have a match, we have to determine the length of the
2333         // original source string.  This will usually be == the length of
2334         // the case folded string, but it may differ (e.g. sharp s).
2335 
2336         // Most of the time, the length will be the same as the length
2337         // of the string from the locale data.  Sometimes it will be
2338         // different, in which case we will have to figure it out by
2339         // adding a character at a time, until we have a match.  We do
2340         // this all in one loop, where we try 'len' first (at index
2341         // i==0).
2342         int32_t len = bestMatchName.length(); // 99+% of the time
2343         int32_t n = text.length() - start;
2344         for (i=0; i<=n; ++i) {
2345             int32_t j=i;
2346             if (i == 0) {
2347                 j = len;
2348             } else if (i == len) {
2349                 continue; // already tried this when i was 0
2350             }
2351             text.extract(start, j, lcaseText);
2352             lcaseText.foldCase();
2353             if (bestMatchName == lcaseText) {
2354                 return start + j;
2355             }
2356         }
2357     }
2358 
2359     return -start;
2360 }
2361 
2362 static UBool
newBestMatchWithOptionalDot(const UnicodeString & lcaseText,const UnicodeString & data,UnicodeString & bestMatchName,int32_t & bestMatchLength)2363 newBestMatchWithOptionalDot(const UnicodeString &lcaseText,
2364                             const UnicodeString &data,
2365                             UnicodeString &bestMatchName,
2366                             int32_t &bestMatchLength) {
2367     UnicodeString lcase;
2368     lcase.fastCopyFrom(data).foldCase();
2369     int32_t length = lcase.length();
2370     if (length <= bestMatchLength) {
2371         // data cannot provide a better match.
2372         return FALSE;
2373     }
2374 
2375     if (lcaseText.compareBetween(0, length, lcase, 0, length) == 0) {
2376         // normal match
2377         bestMatchName = lcase;
2378         bestMatchLength = length;
2379         return TRUE;
2380     }
2381     if (lcase.charAt(--length) == 0x2e) {
2382         if (lcaseText.compareBetween(0, length, lcase, 0, length) == 0) {
2383             // The input text matches the data except for data's trailing dot.
2384             bestMatchName = lcase;
2385             bestMatchName.truncate(length);
2386             bestMatchLength = length;
2387             return TRUE;
2388         }
2389     }
2390     return FALSE;
2391 }
2392 
2393 //----------------------------------------------------------------------
2394 
2395 void
set2DigitYearStart(UDate d,UErrorCode & status)2396 SimpleDateFormat::set2DigitYearStart(UDate d, UErrorCode& status)
2397 {
2398     parseAmbiguousDatesAsAfter(d, status);
2399 }
2400 
2401 /**
2402  * Private member function that converts the parsed date strings into
2403  * timeFields. Returns -start (for ParsePosition) if failed.
2404  * @param text the time text to be parsed.
2405  * @param start where to start parsing.
2406  * @param ch the pattern character for the date field text to be parsed.
2407  * @param count the count of a pattern character.
2408  * @return the new start position if matching succeeded; a negative number
2409  * indicating matching failure, otherwise.
2410  */
subParse(const UnicodeString & text,int32_t & start,UChar ch,int32_t count,UBool obeyCount,UBool allowNegative,UBool ambiguousYear[],int32_t & saveHebrewMonth,Calendar & cal,int32_t patLoc,MessageFormat * numericLeapMonthFormatter) const2411 int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UChar ch, int32_t count,
2412                            UBool obeyCount, UBool allowNegative, UBool ambiguousYear[], int32_t& saveHebrewMonth, Calendar& cal,
2413                            int32_t patLoc, MessageFormat * numericLeapMonthFormatter) const
2414 {
2415     Formattable number;
2416     int32_t value = 0;
2417     int32_t i;
2418     int32_t ps = 0;
2419     UErrorCode status = U_ZERO_ERROR;
2420     ParsePosition pos(0);
2421     UDateFormatField patternCharIndex = DateFormatSymbols::getPatternCharIndex(ch);
2422     NumberFormat *currentNumberFormat;
2423     UnicodeString temp;
2424     UBool gotNumber = FALSE;
2425 
2426 #if defined (U_DEBUG_CAL)
2427     //fprintf(stderr, "%s:%d - [%c]  st=%d \n", __FILE__, __LINE__, (char) ch, start);
2428 #endif
2429 
2430     if (patternCharIndex == UDAT_FIELD_COUNT) {
2431         return -start;
2432     }
2433 
2434     currentNumberFormat = getNumberFormatByIndex(patternCharIndex);
2435     UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex];
2436     UnicodeString hebr("hebr", 4, US_INV);
2437 
2438     if (numericLeapMonthFormatter != NULL) {
2439         numericLeapMonthFormatter->setFormats((const Format **)&currentNumberFormat, 1);
2440     }
2441     UBool isChineseCalendar = (uprv_strcmp(cal.getType(),"chinese") == 0 || uprv_strcmp(cal.getType(),"dangi") == 0);
2442 
2443     // If there are any spaces here, skip over them.  If we hit the end
2444     // of the string, then fail.
2445     for (;;) {
2446         if (start >= text.length()) {
2447             return -start;
2448         }
2449         UChar32 c = text.char32At(start);
2450         if (!u_isUWhiteSpace(c) /*||*/ && !PatternProps::isWhiteSpace(c)) {
2451             break;
2452         }
2453         start += U16_LENGTH(c);
2454     }
2455     pos.setIndex(start);
2456 
2457     // We handle a few special cases here where we need to parse
2458     // a number value.  We handle further, more generic cases below.  We need
2459     // to handle some of them here because some fields require extra processing on
2460     // the parsed value.
2461     if (patternCharIndex == UDAT_HOUR_OF_DAY1_FIELD ||                       // k
2462         patternCharIndex == UDAT_HOUR_OF_DAY0_FIELD ||                       // H
2463         patternCharIndex == UDAT_HOUR1_FIELD ||                              // h
2464         patternCharIndex == UDAT_HOUR0_FIELD ||                              // K
2465         (patternCharIndex == UDAT_DOW_LOCAL_FIELD && count <= 2) ||          // e
2466         (patternCharIndex == UDAT_STANDALONE_DAY_FIELD && count <= 2) ||     // c
2467         (patternCharIndex == UDAT_MONTH_FIELD && count <= 2) ||              // M
2468         (patternCharIndex == UDAT_STANDALONE_MONTH_FIELD && count <= 2) ||   // L
2469         (patternCharIndex == UDAT_QUARTER_FIELD && count <= 2) ||            // Q
2470         (patternCharIndex == UDAT_STANDALONE_QUARTER_FIELD && count <= 2) || // q
2471         patternCharIndex == UDAT_YEAR_FIELD ||                               // y
2472         patternCharIndex == UDAT_YEAR_WOY_FIELD ||                           // Y
2473         patternCharIndex == UDAT_YEAR_NAME_FIELD ||                          // U (falls back to numeric)
2474         (patternCharIndex == UDAT_ERA_FIELD && isChineseCalendar) ||         // G
2475         patternCharIndex == UDAT_FRACTIONAL_SECOND_FIELD)                    // S
2476     {
2477         int32_t parseStart = pos.getIndex();
2478         // It would be good to unify this with the obeyCount logic below,
2479         // but that's going to be difficult.
2480         const UnicodeString* src;
2481 
2482         UBool parsedNumericLeapMonth = FALSE;
2483         if (numericLeapMonthFormatter != NULL && (patternCharIndex == UDAT_MONTH_FIELD || patternCharIndex == UDAT_STANDALONE_MONTH_FIELD)) {
2484             int32_t argCount;
2485             Formattable * args = numericLeapMonthFormatter->parse(text, pos, argCount);
2486             if (args != NULL && argCount == 1 && pos.getIndex() > parseStart && args[0].isNumeric()) {
2487                 parsedNumericLeapMonth = TRUE;
2488                 number.setLong(args[0].getLong());
2489                 cal.set(UCAL_IS_LEAP_MONTH, 1);
2490                 delete[] args;
2491             } else {
2492                 pos.setIndex(parseStart);
2493                 cal.set(UCAL_IS_LEAP_MONTH, 0);
2494             }
2495         }
2496 
2497         if (!parsedNumericLeapMonth) {
2498             if (obeyCount) {
2499                 if ((start+count) > text.length()) {
2500                     return -start;
2501                 }
2502 
2503                 text.extractBetween(0, start + count, temp);
2504                 src = &temp;
2505             } else {
2506                 src = &text;
2507             }
2508 
2509             parseInt(*src, number, pos, allowNegative,currentNumberFormat);
2510         }
2511 
2512         int32_t txtLoc = pos.getIndex();
2513 
2514         if (txtLoc > parseStart) {
2515             value = number.getLong();
2516             gotNumber = TRUE;
2517 
2518             // suffix processing
2519             if (value < 0 ) {
2520                 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, TRUE);
2521                 if (txtLoc != pos.getIndex()) {
2522                     value *= -1;
2523                 }
2524             }
2525             else {
2526                 txtLoc = checkIntSuffix(text, txtLoc, patLoc+1, FALSE);
2527             }
2528 
2529             if (!getBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, status)) {
2530                 // Check the range of the value
2531                 int32_t bias = gFieldRangeBias[patternCharIndex];
2532                 if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
2533                     return -start;
2534                 }
2535             }
2536 
2537             pos.setIndex(txtLoc);
2538         }
2539     }
2540 
2541     // Make sure that we got a number if
2542     // we want one, and didn't get one
2543     // if we don't want one.
2544     switch (patternCharIndex) {
2545         case UDAT_HOUR_OF_DAY1_FIELD:
2546         case UDAT_HOUR_OF_DAY0_FIELD:
2547         case UDAT_HOUR1_FIELD:
2548         case UDAT_HOUR0_FIELD:
2549             // special range check for hours:
2550             if (value < 0 || value > 24) {
2551                 return -start;
2552             }
2553 
2554             // fall through to gotNumber check
2555 
2556         case UDAT_YEAR_FIELD:
2557         case UDAT_YEAR_WOY_FIELD:
2558         case UDAT_FRACTIONAL_SECOND_FIELD:
2559             // these must be a number
2560             if (! gotNumber) {
2561                 return -start;
2562             }
2563 
2564             break;
2565 
2566         default:
2567             // we check the rest of the fields below.
2568             break;
2569     }
2570 
2571     switch (patternCharIndex) {
2572     case UDAT_ERA_FIELD:
2573         if (isChineseCalendar) {
2574             if (!gotNumber) {
2575                 return -start;
2576             }
2577             cal.set(UCAL_ERA, value);
2578             return pos.getIndex();
2579         }
2580         if (count == 5) {
2581             ps = matchString(text, start, UCAL_ERA, fSymbols->fNarrowEras, fSymbols->fNarrowErasCount, NULL, cal);
2582         } else if (count == 4) {
2583             ps = matchString(text, start, UCAL_ERA, fSymbols->fEraNames, fSymbols->fEraNamesCount, NULL, cal);
2584         } else {
2585             ps = matchString(text, start, UCAL_ERA, fSymbols->fEras, fSymbols->fErasCount, NULL, cal);
2586         }
2587 
2588         // check return position, if it equals -start, then matchString error
2589         // special case the return code so we don't necessarily fail out until we
2590         // verify no year information also
2591         if (ps == -start)
2592             ps--;
2593 
2594         return ps;
2595 
2596     case UDAT_YEAR_FIELD:
2597         // If there are 3 or more YEAR pattern characters, this indicates
2598         // that the year value is to be treated literally, without any
2599         // two-digit year adjustments (e.g., from "01" to 2001).  Otherwise
2600         // we made adjustments to place the 2-digit year in the proper
2601         // century, for parsed strings from "00" to "99".  Any other string
2602         // is treated literally:  "2250", "-1", "1", "002".
2603         if (fDateOverride.compare(hebr)==0 && value < 1000) {
2604             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2605         } else if ((pos.getIndex() - start) == 2 && !isChineseCalendar
2606             && u_isdigit(text.charAt(start))
2607             && u_isdigit(text.charAt(start+1)))
2608         {
2609         	// only adjust year for patterns less than 3.
2610         	if(count < 3) {
2611         		// Assume for example that the defaultCenturyStart is 6/18/1903.
2612         		// This means that two-digit years will be forced into the range
2613         		// 6/18/1903 to 6/17/2003.  As a result, years 00, 01, and 02
2614         		// correspond to 2000, 2001, and 2002.  Years 04, 05, etc. correspond
2615         		// to 1904, 1905, etc.  If the year is 03, then it is 2003 if the
2616         		// other fields specify a date before 6/18, or 1903 if they specify a
2617         		// date afterwards.  As a result, 03 is an ambiguous year.  All other
2618         		// two-digit years are unambiguous.
2619         		if(fHaveDefaultCentury) { // check if this formatter even has a pivot year
2620         			int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2621         			ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2622         			value += (fDefaultCenturyStartYear/100)*100 +
2623         					(value < ambiguousTwoDigitYear ? 100 : 0);
2624         		}
2625             }
2626         }
2627         cal.set(UCAL_YEAR, value);
2628 
2629         // Delayed checking for adjustment of Hebrew month numbers in non-leap years.
2630         if (saveHebrewMonth >= 0) {
2631             HebrewCalendar *hc = (HebrewCalendar*)&cal;
2632             if (!hc->isLeapYear(value) && saveHebrewMonth >= 6) {
2633                cal.set(UCAL_MONTH,saveHebrewMonth);
2634             } else {
2635                cal.set(UCAL_MONTH,saveHebrewMonth-1);
2636             }
2637             saveHebrewMonth = -1;
2638         }
2639         return pos.getIndex();
2640 
2641     case UDAT_YEAR_WOY_FIELD:
2642         // Comment is the same as for UDAT_Year_FIELDs - look above
2643         if (fDateOverride.compare(hebr)==0 && value < 1000) {
2644             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
2645         } else if ((pos.getIndex() - start) == 2
2646             && u_isdigit(text.charAt(start))
2647             && u_isdigit(text.charAt(start+1))
2648             && fHaveDefaultCentury )
2649         {
2650             int32_t ambiguousTwoDigitYear = fDefaultCenturyStartYear % 100;
2651             ambiguousYear[0] = (value == ambiguousTwoDigitYear);
2652             value += (fDefaultCenturyStartYear/100)*100 +
2653                 (value < ambiguousTwoDigitYear ? 100 : 0);
2654         }
2655         cal.set(UCAL_YEAR_WOY, value);
2656         return pos.getIndex();
2657 
2658     case UDAT_YEAR_NAME_FIELD:
2659         if (fSymbols->fShortYearNames != NULL) {
2660             int32_t newStart = matchString(text, start, UCAL_YEAR, fSymbols->fShortYearNames, fSymbols->fShortYearNamesCount, NULL, cal);
2661             if (newStart > 0) {
2662                 return newStart;
2663             }
2664         }
2665         if (gotNumber && (getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC,status) || value > fSymbols->fShortYearNamesCount)) {
2666             cal.set(UCAL_YEAR, value);
2667             return pos.getIndex();
2668         }
2669         return -start;
2670 
2671     case UDAT_MONTH_FIELD:
2672     case UDAT_STANDALONE_MONTH_FIELD:
2673         if (gotNumber) // i.e., M or MM.
2674         {
2675             // When parsing month numbers from the Hebrew Calendar, we might need to adjust the month depending on whether
2676             // 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
2677             // the year is parsed.
2678             if (!strcmp(cal.getType(),"hebrew")) {
2679                 HebrewCalendar *hc = (HebrewCalendar*)&cal;
2680                 if (cal.isSet(UCAL_YEAR)) {
2681                    UErrorCode status = U_ZERO_ERROR;
2682                    if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
2683                        cal.set(UCAL_MONTH, value);
2684                    } else {
2685                        cal.set(UCAL_MONTH, value - 1);
2686                    }
2687                 } else {
2688                     saveHebrewMonth = value;
2689                 }
2690             } else {
2691                 // Don't want to parse the month if it is a string
2692                 // while pattern uses numeric style: M/MM, L/LL
2693                 // [We computed 'value' above.]
2694                 cal.set(UCAL_MONTH, value - 1);
2695             }
2696             return pos.getIndex();
2697         } else {
2698             // count >= 3 // i.e., MMM/MMMM, LLL/LLLL
2699             // Want to be able to parse both short and long forms.
2700             // Try count == 4 first:
2701             UnicodeString * wideMonthPat = NULL;
2702             UnicodeString * shortMonthPat = NULL;
2703             if (fSymbols->fLeapMonthPatterns != NULL && fSymbols->fLeapMonthPatternsCount >= DateFormatSymbols::kMonthPatternsCount) {
2704                 if (patternCharIndex==UDAT_MONTH_FIELD) {
2705                     wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatWide];
2706                     shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternFormatAbbrev];
2707                 } else {
2708                     wideMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneWide];
2709                     shortMonthPat = &fSymbols->fLeapMonthPatterns[DateFormatSymbols::kLeapMonthPatternStandaloneAbbrev];
2710                 }
2711             }
2712             int32_t newStart = 0;
2713             if (patternCharIndex==UDAT_MONTH_FIELD) {
2714                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fMonths, fSymbols->fMonthsCount, wideMonthPat, cal); // try MMMM
2715                 if (newStart > 0) {
2716                     return newStart;
2717                 }
2718                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fShortMonths, fSymbols->fShortMonthsCount, shortMonthPat, cal); // try MMM
2719             } else {
2720                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount, wideMonthPat, cal); // try LLLL
2721                 if (newStart > 0) {
2722                     return newStart;
2723                 }
2724                 newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount, shortMonthPat, cal); // try LLL
2725             }
2726             if (newStart > 0 || !getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))  // currently we do not try to parse MMMMM/LLLLL: #8860
2727                 return newStart;
2728             // else we allowing parsing as number, below
2729         }
2730         break;
2731 
2732     case UDAT_HOUR_OF_DAY1_FIELD:
2733         // [We computed 'value' above.]
2734         if (value == cal.getMaximum(UCAL_HOUR_OF_DAY) + 1)
2735             value = 0;
2736 
2737         // fall through to set field
2738 
2739     case UDAT_HOUR_OF_DAY0_FIELD:
2740         cal.set(UCAL_HOUR_OF_DAY, value);
2741         return pos.getIndex();
2742 
2743     case UDAT_FRACTIONAL_SECOND_FIELD:
2744         // Fractional seconds left-justify
2745         i = pos.getIndex() - start;
2746         if (i < 3) {
2747             while (i < 3) {
2748                 value *= 10;
2749                 i++;
2750             }
2751         } else {
2752             int32_t a = 1;
2753             while (i > 3) {
2754                 a *= 10;
2755                 i--;
2756             }
2757             value /= a;
2758         }
2759         cal.set(UCAL_MILLISECOND, value);
2760         return pos.getIndex();
2761 
2762     case UDAT_DOW_LOCAL_FIELD:
2763         if (gotNumber) // i.e., e or ee
2764         {
2765             // [We computed 'value' above.]
2766             cal.set(UCAL_DOW_LOCAL, value);
2767             return pos.getIndex();
2768         }
2769         // else for eee-eeeee fall through to handling of EEE-EEEEE
2770         // fall through, do not break here
2771     case UDAT_DAY_OF_WEEK_FIELD:
2772         {
2773             // Want to be able to parse both short and long forms.
2774             // Try count == 4 (EEEE) wide first:
2775             int32_t newStart = 0;
2776             if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2777                                       fSymbols->fWeekdays, fSymbols->fWeekdaysCount, NULL, cal)) > 0)
2778                 return newStart;
2779             // EEEE wide failed, now try EEE abbreviated
2780             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2781                                    fSymbols->fShortWeekdays, fSymbols->fShortWeekdaysCount, NULL, cal)) > 0)
2782                 return newStart;
2783             // EEE abbreviated failed, now try EEEEEE short
2784             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2785                                    fSymbols->fShorterWeekdays, fSymbols->fShorterWeekdaysCount, NULL, cal)) > 0)
2786                 return newStart;
2787             // EEEEEE short failed, now try EEEEE narrow
2788             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2789                                    fSymbols->fNarrowWeekdays, fSymbols->fNarrowWeekdaysCount, NULL, cal)) > 0)
2790                 return newStart;
2791             else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status) || patternCharIndex == UDAT_DAY_OF_WEEK_FIELD)
2792                 return newStart;
2793             // else we allowing parsing as number, below
2794         }
2795         break;
2796 
2797     case UDAT_STANDALONE_DAY_FIELD:
2798         {
2799             if (gotNumber) // c or cc
2800             {
2801                 // [We computed 'value' above.]
2802                 cal.set(UCAL_DOW_LOCAL, value);
2803                 return pos.getIndex();
2804             }
2805             // Want to be able to parse both short and long forms.
2806             // Try count == 4 (cccc) first:
2807             int32_t newStart = 0;
2808             if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2809                                       fSymbols->fStandaloneWeekdays, fSymbols->fStandaloneWeekdaysCount, NULL, cal)) > 0)
2810                 return newStart;
2811             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2812                                           fSymbols->fStandaloneShortWeekdays, fSymbols->fStandaloneShortWeekdaysCount, NULL, cal)) > 0)
2813                 return newStart;
2814             else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
2815                                           fSymbols->fStandaloneShorterWeekdays, fSymbols->fStandaloneShorterWeekdaysCount, NULL, cal)) > 0)
2816                 return newStart;
2817             else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
2818                 return newStart;
2819             // else we allowing parsing as number, below
2820         }
2821         break;
2822 
2823     case UDAT_AM_PM_FIELD:
2824         return matchString(text, start, UCAL_AM_PM, fSymbols->fAmPms, fSymbols->fAmPmsCount, NULL, cal);
2825 
2826     case UDAT_HOUR1_FIELD:
2827         // [We computed 'value' above.]
2828         if (value == cal.getLeastMaximum(UCAL_HOUR)+1)
2829             value = 0;
2830 
2831         // fall through to set field
2832 
2833     case UDAT_HOUR0_FIELD:
2834         cal.set(UCAL_HOUR, value);
2835         return pos.getIndex();
2836 
2837     case UDAT_QUARTER_FIELD:
2838         if (gotNumber) // i.e., Q or QQ.
2839         {
2840             // Don't want to parse the month if it is a string
2841             // while pattern uses numeric style: Q or QQ.
2842             // [We computed 'value' above.]
2843             cal.set(UCAL_MONTH, (value - 1) * 3);
2844             return pos.getIndex();
2845         } else {
2846             // count >= 3 // i.e., QQQ or QQQQ
2847             // Want to be able to parse both short and long forms.
2848             // Try count == 4 first:
2849             int32_t newStart = 0;
2850 
2851             if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2852                                       fSymbols->fQuarters, fSymbols->fQuartersCount, cal)) > 0)
2853                 return newStart;
2854             else if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2855                                           fSymbols->fShortQuarters, fSymbols->fShortQuartersCount, cal)) > 0)
2856                 return newStart;
2857             else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
2858                 return newStart;
2859             // else we allowing parsing as number, below
2860         }
2861         break;
2862 
2863     case UDAT_STANDALONE_QUARTER_FIELD:
2864         if (gotNumber) // i.e., q or qq.
2865         {
2866             // Don't want to parse the month if it is a string
2867             // while pattern uses numeric style: q or q.
2868             // [We computed 'value' above.]
2869             cal.set(UCAL_MONTH, (value - 1) * 3);
2870             return pos.getIndex();
2871         } else {
2872             // count >= 3 // i.e., qqq or qqqq
2873             // Want to be able to parse both short and long forms.
2874             // Try count == 4 first:
2875             int32_t newStart = 0;
2876 
2877             if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2878                                       fSymbols->fStandaloneQuarters, fSymbols->fStandaloneQuartersCount, cal)) > 0)
2879                 return newStart;
2880             else if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
2881                                           fSymbols->fStandaloneShortQuarters, fSymbols->fStandaloneShortQuartersCount, cal)) > 0)
2882                 return newStart;
2883             else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
2884                 return newStart;
2885             // else we allowing parsing as number, below
2886         }
2887         break;
2888 
2889     case UDAT_TIMEZONE_FIELD: // 'z'
2890         {
2891             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2892             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_SPECIFIC_SHORT : UTZFMT_STYLE_SPECIFIC_LONG;
2893             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2894             if (tz != NULL) {
2895                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2896                 cal.adoptTimeZone(tz);
2897                 return pos.getIndex();
2898             }
2899         }
2900         break;
2901     case UDAT_TIMEZONE_RFC_FIELD: // 'Z'
2902         {
2903             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2904             UTimeZoneFormatStyle style = (count < 4) ?
2905                 UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL : ((count == 5) ? UTZFMT_STYLE_ISO_EXTENDED_FULL: UTZFMT_STYLE_LOCALIZED_GMT);
2906             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2907             if (tz != NULL) {
2908                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2909                 cal.adoptTimeZone(tz);
2910                 return pos.getIndex();
2911             }
2912             return -start;
2913         }
2914     case UDAT_TIMEZONE_GENERIC_FIELD: // 'v'
2915         {
2916             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2917             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_GENERIC_SHORT : UTZFMT_STYLE_GENERIC_LONG;
2918             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2919             if (tz != NULL) {
2920                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2921                 cal.adoptTimeZone(tz);
2922                 return pos.getIndex();
2923             }
2924             return -start;
2925         }
2926     case UDAT_TIMEZONE_SPECIAL_FIELD: // 'V'
2927         {
2928             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2929             UTimeZoneFormatStyle style;
2930             switch (count) {
2931             case 1:
2932                 style = UTZFMT_STYLE_ZONE_ID_SHORT;
2933                 break;
2934             case 2:
2935                 style = UTZFMT_STYLE_ZONE_ID;
2936                 break;
2937             case 3:
2938                 style = UTZFMT_STYLE_EXEMPLAR_LOCATION;
2939                 break;
2940             default:
2941                 style = UTZFMT_STYLE_GENERIC_LOCATION;
2942                 break;
2943             }
2944             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2945             if (tz != NULL) {
2946                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2947                 cal.adoptTimeZone(tz);
2948                 return pos.getIndex();
2949             }
2950             return -start;
2951         }
2952     case UDAT_TIMEZONE_LOCALIZED_GMT_OFFSET_FIELD: // 'O'
2953         {
2954             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2955             UTimeZoneFormatStyle style = (count < 4) ? UTZFMT_STYLE_LOCALIZED_GMT_SHORT : UTZFMT_STYLE_LOCALIZED_GMT;
2956             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2957             if (tz != NULL) {
2958                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2959                 cal.adoptTimeZone(tz);
2960                 return pos.getIndex();
2961             }
2962             return -start;
2963         }
2964     case UDAT_TIMEZONE_ISO_FIELD: // 'X'
2965         {
2966             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2967             UTimeZoneFormatStyle style;
2968             switch (count) {
2969             case 1:
2970                 style = UTZFMT_STYLE_ISO_BASIC_SHORT;
2971                 break;
2972             case 2:
2973                 style = UTZFMT_STYLE_ISO_BASIC_FIXED;
2974                 break;
2975             case 3:
2976                 style = UTZFMT_STYLE_ISO_EXTENDED_FIXED;
2977                 break;
2978             case 4:
2979                 style = UTZFMT_STYLE_ISO_BASIC_FULL;
2980                 break;
2981             default:
2982                 style = UTZFMT_STYLE_ISO_EXTENDED_FULL;
2983                 break;
2984             }
2985             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
2986             if (tz != NULL) {
2987                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
2988                 cal.adoptTimeZone(tz);
2989                 return pos.getIndex();
2990             }
2991             return -start;
2992         }
2993     case UDAT_TIMEZONE_ISO_LOCAL_FIELD: // 'x'
2994         {
2995             UTimeZoneFormatTimeType tzTimeType = UTZFMT_TIME_TYPE_UNKNOWN;
2996             UTimeZoneFormatStyle style;
2997             switch (count) {
2998             case 1:
2999                 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_SHORT;
3000                 break;
3001             case 2:
3002                 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FIXED;
3003                 break;
3004             case 3:
3005                 style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FIXED;
3006                 break;
3007             case 4:
3008                 style = UTZFMT_STYLE_ISO_BASIC_LOCAL_FULL;
3009                 break;
3010             default:
3011                 style = UTZFMT_STYLE_ISO_EXTENDED_LOCAL_FULL;
3012                 break;
3013             }
3014             TimeZone *tz  = tzFormat()->parse(style, text, pos, &tzTimeType);
3015             if (tz != NULL) {
3016                 ((SimpleDateFormat*)this)->tztype = tzTimeType;
3017                 cal.adoptTimeZone(tz);
3018                 return pos.getIndex();
3019             }
3020             return -start;
3021         }
3022 
3023     default:
3024         // Handle "generic" fields
3025         // this is now handled below, outside the switch block
3026         break;
3027     }
3028     // Handle "generic" fields:
3029     // switch default case now handled here (outside switch block) to allow
3030     // parsing of some string fields as digits for lenient case
3031 
3032     int32_t parseStart = pos.getIndex();
3033     const UnicodeString* src;
3034     if (obeyCount) {
3035         if ((start+count) > text.length()) {
3036             return -start;
3037         }
3038         text.extractBetween(0, start + count, temp);
3039         src = &temp;
3040     } else {
3041         src = &text;
3042     }
3043     parseInt(*src, number, pos, allowNegative,currentNumberFormat);
3044     if (pos.getIndex() != parseStart) {
3045         int32_t value = number.getLong();
3046 
3047         // Don't need suffix processing here (as in number processing at the beginning of the function);
3048         // the new fields being handled as numeric values (month, weekdays, quarters) should not have suffixes.
3049 
3050         if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status)) {
3051             // Check the range of the value
3052             int32_t bias = gFieldRangeBias[patternCharIndex];
3053             if (bias >= 0 && (value > cal.getMaximum(field) + bias || value < cal.getMinimum(field) + bias)) {
3054                 return -start;
3055             }
3056         }
3057 
3058         // For the following, need to repeat some of the "if (gotNumber)" code above:
3059         // UDAT_[STANDALONE_]MONTH_FIELD, UDAT_DOW_LOCAL_FIELD, UDAT_STANDALONE_DAY_FIELD,
3060         // UDAT_[STANDALONE_]QUARTER_FIELD
3061         switch (patternCharIndex) {
3062         case UDAT_MONTH_FIELD:
3063             // See notes under UDAT_MONTH_FIELD case above
3064             if (!strcmp(cal.getType(),"hebrew")) {
3065                 HebrewCalendar *hc = (HebrewCalendar*)&cal;
3066                 if (cal.isSet(UCAL_YEAR)) {
3067                    UErrorCode status = U_ZERO_ERROR;
3068                    if (!hc->isLeapYear(hc->get(UCAL_YEAR,status)) && value >= 6) {
3069                        cal.set(UCAL_MONTH, value);
3070                    } else {
3071                        cal.set(UCAL_MONTH, value - 1);
3072                    }
3073                 } else {
3074                     saveHebrewMonth = value;
3075                 }
3076             } else {
3077                 cal.set(UCAL_MONTH, value - 1);
3078             }
3079             break;
3080         case UDAT_STANDALONE_MONTH_FIELD:
3081             cal.set(UCAL_MONTH, value - 1);
3082             break;
3083         case UDAT_DOW_LOCAL_FIELD:
3084         case UDAT_STANDALONE_DAY_FIELD:
3085             cal.set(UCAL_DOW_LOCAL, value);
3086             break;
3087         case UDAT_QUARTER_FIELD:
3088         case UDAT_STANDALONE_QUARTER_FIELD:
3089              cal.set(UCAL_MONTH, (value - 1) * 3);
3090              break;
3091         default:
3092             cal.set(field, value);
3093             break;
3094         }
3095         return pos.getIndex();
3096     }
3097     return -start;
3098 }
3099 
3100 /**
3101  * Parse an integer using fNumberFormat.  This method is semantically
3102  * const, but actually may modify fNumberFormat.
3103  */
parseInt(const UnicodeString & text,Formattable & number,ParsePosition & pos,UBool allowNegative,NumberFormat * fmt) const3104 void SimpleDateFormat::parseInt(const UnicodeString& text,
3105                                 Formattable& number,
3106                                 ParsePosition& pos,
3107                                 UBool allowNegative,
3108                                 NumberFormat *fmt) const {
3109     parseInt(text, number, -1, pos, allowNegative,fmt);
3110 }
3111 
3112 /**
3113  * Parse an integer using fNumberFormat up to maxDigits.
3114  */
parseInt(const UnicodeString & text,Formattable & number,int32_t maxDigits,ParsePosition & pos,UBool allowNegative,NumberFormat * fmt) const3115 void SimpleDateFormat::parseInt(const UnicodeString& text,
3116                                 Formattable& number,
3117                                 int32_t maxDigits,
3118                                 ParsePosition& pos,
3119                                 UBool allowNegative,
3120                                 NumberFormat *fmt) const {
3121     UnicodeString oldPrefix;
3122     DecimalFormat* df = NULL;
3123     if (!allowNegative && (df = dynamic_cast<DecimalFormat*>(fmt)) != NULL) {
3124         df->getNegativePrefix(oldPrefix);
3125         df->setNegativePrefix(UnicodeString(TRUE, SUPPRESS_NEGATIVE_PREFIX, -1));
3126     }
3127     int32_t oldPos = pos.getIndex();
3128     fmt->parse(text, number, pos);
3129     if (df != NULL) {
3130         df->setNegativePrefix(oldPrefix);
3131     }
3132 
3133     if (maxDigits > 0) {
3134         // adjust the result to fit into
3135         // the maxDigits and move the position back
3136         int32_t nDigits = pos.getIndex() - oldPos;
3137         if (nDigits > maxDigits) {
3138             int32_t val = number.getLong();
3139             nDigits -= maxDigits;
3140             while (nDigits > 0) {
3141                 val /= 10;
3142                 nDigits--;
3143             }
3144             pos.setIndex(oldPos + maxDigits);
3145             number.setLong(val);
3146         }
3147     }
3148 }
3149 
3150 //----------------------------------------------------------------------
3151 
translatePattern(const UnicodeString & originalPattern,UnicodeString & translatedPattern,const UnicodeString & from,const UnicodeString & to,UErrorCode & status)3152 void SimpleDateFormat::translatePattern(const UnicodeString& originalPattern,
3153                                         UnicodeString& translatedPattern,
3154                                         const UnicodeString& from,
3155                                         const UnicodeString& to,
3156                                         UErrorCode& status)
3157 {
3158   // run through the pattern and convert any pattern symbols from the version
3159   // in "from" to the corresponding character ion "to".  This code takes
3160   // quoted strings into account (it doesn't try to translate them), and it signals
3161   // an error if a particular "pattern character" doesn't appear in "from".
3162   // Depending on the values of "from" and "to" this can convert from generic
3163   // to localized patterns or localized to generic.
3164   if (U_FAILURE(status))
3165     return;
3166 
3167   translatedPattern.remove();
3168   UBool inQuote = FALSE;
3169   for (int32_t i = 0; i < originalPattern.length(); ++i) {
3170     UChar c = originalPattern[i];
3171     if (inQuote) {
3172       if (c == QUOTE)
3173     inQuote = FALSE;
3174     }
3175     else {
3176       if (c == QUOTE)
3177     inQuote = TRUE;
3178       else if ((c >= 0x0061 /*'a'*/ && c <= 0x007A) /*'z'*/
3179            || (c >= 0x0041 /*'A'*/ && c <= 0x005A /*'Z'*/)) {
3180     int32_t ci = from.indexOf(c);
3181     if (ci == -1) {
3182       status = U_INVALID_FORMAT_ERROR;
3183       return;
3184     }
3185     c = to[ci];
3186       }
3187     }
3188     translatedPattern += c;
3189   }
3190   if (inQuote) {
3191     status = U_INVALID_FORMAT_ERROR;
3192     return;
3193   }
3194 }
3195 
3196 //----------------------------------------------------------------------
3197 
3198 UnicodeString&
toPattern(UnicodeString & result) const3199 SimpleDateFormat::toPattern(UnicodeString& result) const
3200 {
3201     result = fPattern;
3202     return result;
3203 }
3204 
3205 //----------------------------------------------------------------------
3206 
3207 UnicodeString&
toLocalizedPattern(UnicodeString & result,UErrorCode & status) const3208 SimpleDateFormat::toLocalizedPattern(UnicodeString& result,
3209                                      UErrorCode& status) const
3210 {
3211     translatePattern(fPattern, result,
3212                      UnicodeString(DateFormatSymbols::getPatternUChars()),
3213                      fSymbols->fLocalPatternChars, status);
3214     return result;
3215 }
3216 
3217 //----------------------------------------------------------------------
3218 
3219 void
applyPattern(const UnicodeString & pattern)3220 SimpleDateFormat::applyPattern(const UnicodeString& pattern)
3221 {
3222     fPattern = pattern;
3223 }
3224 
3225 //----------------------------------------------------------------------
3226 
3227 void
applyLocalizedPattern(const UnicodeString & pattern,UErrorCode & status)3228 SimpleDateFormat::applyLocalizedPattern(const UnicodeString& pattern,
3229                                         UErrorCode &status)
3230 {
3231     translatePattern(pattern, fPattern,
3232                      fSymbols->fLocalPatternChars,
3233                      UnicodeString(DateFormatSymbols::getPatternUChars()), status);
3234 }
3235 
3236 //----------------------------------------------------------------------
3237 
3238 const DateFormatSymbols*
getDateFormatSymbols() const3239 SimpleDateFormat::getDateFormatSymbols() const
3240 {
3241     return fSymbols;
3242 }
3243 
3244 //----------------------------------------------------------------------
3245 
3246 void
adoptDateFormatSymbols(DateFormatSymbols * newFormatSymbols)3247 SimpleDateFormat::adoptDateFormatSymbols(DateFormatSymbols* newFormatSymbols)
3248 {
3249     delete fSymbols;
3250     fSymbols = newFormatSymbols;
3251 }
3252 
3253 //----------------------------------------------------------------------
3254 void
setDateFormatSymbols(const DateFormatSymbols & newFormatSymbols)3255 SimpleDateFormat::setDateFormatSymbols(const DateFormatSymbols& newFormatSymbols)
3256 {
3257     delete fSymbols;
3258     fSymbols = new DateFormatSymbols(newFormatSymbols);
3259 }
3260 
3261 //----------------------------------------------------------------------
3262 const TimeZoneFormat*
getTimeZoneFormat(void) const3263 SimpleDateFormat::getTimeZoneFormat(void) const {
3264     return (const TimeZoneFormat*)tzFormat();
3265 }
3266 
3267 //----------------------------------------------------------------------
3268 void
adoptTimeZoneFormat(TimeZoneFormat * timeZoneFormatToAdopt)3269 SimpleDateFormat::adoptTimeZoneFormat(TimeZoneFormat* timeZoneFormatToAdopt)
3270 {
3271     delete fTimeZoneFormat;
3272     fTimeZoneFormat = timeZoneFormatToAdopt;
3273 }
3274 
3275 //----------------------------------------------------------------------
3276 void
setTimeZoneFormat(const TimeZoneFormat & newTimeZoneFormat)3277 SimpleDateFormat::setTimeZoneFormat(const TimeZoneFormat& newTimeZoneFormat)
3278 {
3279     delete fTimeZoneFormat;
3280     fTimeZoneFormat = new TimeZoneFormat(newTimeZoneFormat);
3281 }
3282 
3283 //----------------------------------------------------------------------
3284 
3285 
adoptCalendar(Calendar * calendarToAdopt)3286 void SimpleDateFormat::adoptCalendar(Calendar* calendarToAdopt)
3287 {
3288   UErrorCode status = U_ZERO_ERROR;
3289   DateFormat::adoptCalendar(calendarToAdopt);
3290   delete fSymbols;
3291   fSymbols=NULL;
3292   initializeSymbols(fLocale, fCalendar, status);  // we need new symbols
3293   initializeDefaultCentury();  // we need a new century (possibly)
3294 }
3295 
3296 
3297 //----------------------------------------------------------------------
3298 
3299 
setContext(UDisplayContext value,UErrorCode & status)3300 void SimpleDateFormat::setContext(UDisplayContext value, UErrorCode& status)
3301 {
3302     if (U_FAILURE(status))
3303         return;
3304     if ( (UDisplayContextType)((uint32_t)value >> 8) == UDISPCTX_TYPE_CAPITALIZATION ) {
3305         fCapitalizationContext = value;
3306     } else {
3307         status = U_ILLEGAL_ARGUMENT_ERROR;
3308    }
3309 }
3310 
3311 
3312 //----------------------------------------------------------------------
3313 
3314 
getContext(UDisplayContextType type,UErrorCode & status) const3315 UDisplayContext SimpleDateFormat::getContext(UDisplayContextType type, UErrorCode& status) const
3316 {
3317     if (U_FAILURE(status))
3318         return (UDisplayContext)0;
3319     if (type != UDISPCTX_TYPE_CAPITALIZATION) {
3320         status = U_ILLEGAL_ARGUMENT_ERROR;
3321         return (UDisplayContext)0;
3322     }
3323     return fCapitalizationContext;
3324 }
3325 
3326 
3327 //----------------------------------------------------------------------
3328 
3329 
3330 UBool
isFieldUnitIgnored(UCalendarDateFields field) const3331 SimpleDateFormat::isFieldUnitIgnored(UCalendarDateFields field) const {
3332     return isFieldUnitIgnored(fPattern, field);
3333 }
3334 
3335 
3336 UBool
isFieldUnitIgnored(const UnicodeString & pattern,UCalendarDateFields field)3337 SimpleDateFormat::isFieldUnitIgnored(const UnicodeString& pattern,
3338                                      UCalendarDateFields field) {
3339     int32_t fieldLevel = fgCalendarFieldToLevel[field];
3340     int32_t level;
3341     UChar ch;
3342     UBool inQuote = FALSE;
3343     UChar prevCh = 0;
3344     int32_t count = 0;
3345 
3346     for (int32_t i = 0; i < pattern.length(); ++i) {
3347         ch = pattern[i];
3348         if (ch != prevCh && count > 0) {
3349             level = fgPatternCharToLevel[prevCh - PATTERN_CHAR_BASE];
3350             // the larger the level, the smaller the field unit.
3351             if ( fieldLevel <= level ) {
3352                 return FALSE;
3353             }
3354             count = 0;
3355         }
3356         if (ch == QUOTE) {
3357             if ((i+1) < pattern.length() && pattern[i+1] == QUOTE) {
3358                 ++i;
3359             } else {
3360                 inQuote = ! inQuote;
3361             }
3362         }
3363         else if ( ! inQuote && ((ch >= 0x0061 /*'a'*/ && ch <= 0x007A /*'z'*/)
3364                     || (ch >= 0x0041 /*'A'*/ && ch <= 0x005A /*'Z'*/))) {
3365             prevCh = ch;
3366             ++count;
3367         }
3368     }
3369     if ( count > 0 ) {
3370         // last item
3371         level = fgPatternCharToLevel[prevCh - PATTERN_CHAR_BASE];
3372             if ( fieldLevel <= level ) {
3373                 return FALSE;
3374             }
3375     }
3376     return TRUE;
3377 }
3378 
3379 //----------------------------------------------------------------------
3380 
3381 const Locale&
getSmpFmtLocale(void) const3382 SimpleDateFormat::getSmpFmtLocale(void) const {
3383     return fLocale;
3384 }
3385 
3386 //----------------------------------------------------------------------
3387 
3388 int32_t
checkIntSuffix(const UnicodeString & text,int32_t start,int32_t patLoc,UBool isNegative) const3389 SimpleDateFormat::checkIntSuffix(const UnicodeString& text, int32_t start,
3390                                  int32_t patLoc, UBool isNegative) const {
3391     // local variables
3392     UnicodeString suf;
3393     int32_t patternMatch;
3394     int32_t textPreMatch;
3395     int32_t textPostMatch;
3396 
3397     // check that we are still in range
3398     if ( (start > text.length()) ||
3399          (start < 0) ||
3400          (patLoc < 0) ||
3401          (patLoc > fPattern.length())) {
3402         // out of range, don't advance location in text
3403         return start;
3404     }
3405 
3406     // get the suffix
3407     DecimalFormat* decfmt = dynamic_cast<DecimalFormat*>(fNumberFormat);
3408     if (decfmt != NULL) {
3409         if (isNegative) {
3410             suf = decfmt->getNegativeSuffix(suf);
3411         }
3412         else {
3413             suf = decfmt->getPositiveSuffix(suf);
3414         }
3415     }
3416 
3417     // check for suffix
3418     if (suf.length() <= 0) {
3419         return start;
3420     }
3421 
3422     // check suffix will be encountered in the pattern
3423     patternMatch = compareSimpleAffix(suf,fPattern,patLoc);
3424 
3425     // check if a suffix will be encountered in the text
3426     textPreMatch = compareSimpleAffix(suf,text,start);
3427 
3428     // check if a suffix was encountered in the text
3429     textPostMatch = compareSimpleAffix(suf,text,start-suf.length());
3430 
3431     // check for suffix match
3432     if ((textPreMatch >= 0) && (patternMatch >= 0) && (textPreMatch == patternMatch)) {
3433         return start;
3434     }
3435     else if ((textPostMatch >= 0) && (patternMatch >= 0) && (textPostMatch == patternMatch)) {
3436         return  start - suf.length();
3437     }
3438 
3439     // should not get here
3440     return start;
3441 }
3442 
3443 //----------------------------------------------------------------------
3444 
3445 int32_t
compareSimpleAffix(const UnicodeString & affix,const UnicodeString & input,int32_t pos) const3446 SimpleDateFormat::compareSimpleAffix(const UnicodeString& affix,
3447                    const UnicodeString& input,
3448                    int32_t pos) const {
3449     int32_t start = pos;
3450     for (int32_t i=0; i<affix.length(); ) {
3451         UChar32 c = affix.char32At(i);
3452         int32_t len = U16_LENGTH(c);
3453         if (PatternProps::isWhiteSpace(c)) {
3454             // We may have a pattern like: \u200F \u0020
3455             //        and input text like: \u200F \u0020
3456             // Note that U+200F and U+0020 are Pattern_White_Space but only
3457             // U+0020 is UWhiteSpace.  So we have to first do a direct
3458             // match of the run of Pattern_White_Space in the pattern,
3459             // then match any extra characters.
3460             UBool literalMatch = FALSE;
3461             while (pos < input.length() &&
3462                    input.char32At(pos) == c) {
3463                 literalMatch = TRUE;
3464                 i += len;
3465                 pos += len;
3466                 if (i == affix.length()) {
3467                     break;
3468                 }
3469                 c = affix.char32At(i);
3470                 len = U16_LENGTH(c);
3471                 if (!PatternProps::isWhiteSpace(c)) {
3472                     break;
3473                 }
3474             }
3475 
3476             // Advance over run in pattern
3477             i = skipPatternWhiteSpace(affix, i);
3478 
3479             // Advance over run in input text
3480             // Must see at least one white space char in input,
3481             // unless we've already matched some characters literally.
3482             int32_t s = pos;
3483             pos = skipUWhiteSpace(input, pos);
3484             if (pos == s && !literalMatch) {
3485                 return -1;
3486             }
3487 
3488             // If we skip UWhiteSpace in the input text, we need to skip it in the pattern.
3489             // Otherwise, the previous lines may have skipped over text (such as U+00A0) that
3490             // is also in the affix.
3491             i = skipUWhiteSpace(affix, i);
3492         } else {
3493             if (pos < input.length() &&
3494                 input.char32At(pos) == c) {
3495                 i += len;
3496                 pos += len;
3497             } else {
3498                 return -1;
3499             }
3500         }
3501     }
3502     return pos - start;
3503 }
3504 
3505 //----------------------------------------------------------------------
3506 
3507 int32_t
skipPatternWhiteSpace(const UnicodeString & text,int32_t pos) const3508 SimpleDateFormat::skipPatternWhiteSpace(const UnicodeString& text, int32_t pos) const {
3509     const UChar* s = text.getBuffer();
3510     return (int32_t)(PatternProps::skipWhiteSpace(s + pos, text.length() - pos) - s);
3511 }
3512 
3513 //----------------------------------------------------------------------
3514 
3515 int32_t
skipUWhiteSpace(const UnicodeString & text,int32_t pos) const3516 SimpleDateFormat::skipUWhiteSpace(const UnicodeString& text, int32_t pos) const {
3517     while (pos < text.length()) {
3518         UChar32 c = text.char32At(pos);
3519         if (!u_isUWhiteSpace(c)) {
3520             break;
3521         }
3522         pos += U16_LENGTH(c);
3523     }
3524     return pos;
3525 }
3526 
3527 //----------------------------------------------------------------------
3528 
3529 // Lazy TimeZoneFormat instantiation, semantically const.
3530 TimeZoneFormat *
tzFormat() const3531 SimpleDateFormat::tzFormat() const {
3532     if (fTimeZoneFormat == NULL) {
3533         umtx_lock(&LOCK);
3534         {
3535             if (fTimeZoneFormat == NULL) {
3536                 UErrorCode status = U_ZERO_ERROR;
3537                 TimeZoneFormat *tzfmt = TimeZoneFormat::createInstance(fLocale, status);
3538                 if (U_FAILURE(status)) {
3539                     return NULL;
3540                 }
3541 
3542                 const_cast<SimpleDateFormat *>(this)->fTimeZoneFormat = tzfmt;
3543             }
3544         }
3545         umtx_unlock(&LOCK);
3546     }
3547     return fTimeZoneFormat;
3548 }
3549 
3550 U_NAMESPACE_END
3551 
3552 #endif /* #if !UCONFIG_NO_FORMATTING */
3553 
3554 //eof
3555