1 /*
2 *******************************************************************************
3 * Copyright (C) 2007-2010, International Business Machines Corporation and
4 * others. All Rights Reserved.
5 *******************************************************************************
6 *
7 * File DTPTNGEN.CPP
8 *
9 *******************************************************************************
10 */
11
12 #include "unicode/utypes.h"
13 #if !UCONFIG_NO_FORMATTING
14
15 #include "unicode/datefmt.h"
16 #include "unicode/decimfmt.h"
17 #include "unicode/dtfmtsym.h"
18 #include "unicode/dtptngen.h"
19 #include "unicode/msgfmt.h"
20 #include "unicode/smpdtfmt.h"
21 #include "unicode/udat.h"
22 #include "unicode/udatpg.h"
23 #include "unicode/uniset.h"
24 #include "unicode/uloc.h"
25 #include "unicode/ures.h"
26 #include "unicode/ustring.h"
27 #include "unicode/rep.h"
28 #include "cpputils.h"
29 #include "ucln_in.h"
30 #include "mutex.h"
31 #include "cmemory.h"
32 #include "cstring.h"
33 #include "locbased.h"
34 #include "gregoimp.h"
35 #include "hash.h"
36 #include "uresimp.h"
37 #include "dtptngen_impl.h"
38
39 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
40
41 #if U_CHARSET_FAMILY==U_EBCDIC_FAMILY
42 /**
43 * If we are on EBCDIC, use an iterator which will
44 * traverse the bundles in ASCII order.
45 */
46 #define U_USE_ASCII_BUNDLE_ITERATOR
47 #define U_SORT_ASCII_BUNDLE_ITERATOR
48 #endif
49
50 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
51
52 #include "unicode/ustring.h"
53 #include "uarrsort.h"
54
55 struct UResAEntry {
56 UChar *key;
57 UResourceBundle *item;
58 };
59
60 struct UResourceBundleAIterator {
61 UResourceBundle *bund;
62 UResAEntry *entries;
63 int32_t num;
64 int32_t cursor;
65 };
66
67 /* Must be C linkage to pass function pointer to the sort function */
68
69 #if !defined (OS390) && !defined (OS400)
70 extern "C"
71 #endif
72 static int32_t U_CALLCONV
ures_a_codepointSort(const void * context,const void * left,const void * right)73 ures_a_codepointSort(const void *context, const void *left, const void *right) {
74 //CompareContext *cmp=(CompareContext *)context;
75 return u_strcmp(((const UResAEntry *)left)->key,
76 ((const UResAEntry *)right)->key);
77 }
78
79
ures_a_open(UResourceBundleAIterator * aiter,UResourceBundle * bund,UErrorCode * status)80 static void ures_a_open(UResourceBundleAIterator *aiter, UResourceBundle *bund, UErrorCode *status) {
81 if(U_FAILURE(*status)) {
82 return;
83 }
84 aiter->bund = bund;
85 aiter->num = ures_getSize(aiter->bund);
86 aiter->cursor = 0;
87 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
88 aiter->entries = NULL;
89 #else
90 aiter->entries = (UResAEntry*)uprv_malloc(sizeof(UResAEntry)*aiter->num);
91 for(int i=0;i<aiter->num;i++) {
92 aiter->entries[i].item = ures_getByIndex(aiter->bund, i, NULL, status);
93 const char *akey = ures_getKey(aiter->entries[i].item);
94 int32_t len = uprv_strlen(akey)+1;
95 aiter->entries[i].key = (UChar*)uprv_malloc(len*sizeof(UChar));
96 u_charsToUChars(akey, aiter->entries[i].key, len);
97 }
98 uprv_sortArray(aiter->entries, aiter->num, sizeof(UResAEntry), ures_a_codepointSort, NULL, TRUE, status);
99 #endif
100 }
101
ures_a_close(UResourceBundleAIterator * aiter)102 static void ures_a_close(UResourceBundleAIterator *aiter) {
103 #if defined(U_SORT_ASCII_BUNDLE_ITERATOR)
104 for(int i=0;i<aiter->num;i++) {
105 uprv_free(aiter->entries[i].key);
106 ures_close(aiter->entries[i].item);
107 }
108 #endif
109 }
110
ures_a_getNextString(UResourceBundleAIterator * aiter,int32_t * len,const char ** key,UErrorCode * err)111 static const UChar *ures_a_getNextString(UResourceBundleAIterator *aiter, int32_t *len, const char **key, UErrorCode *err) {
112 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
113 return ures_getNextString(aiter->bund, len, key, err);
114 #else
115 if(U_FAILURE(*err)) return NULL;
116 UResourceBundle *item = aiter->entries[aiter->cursor].item;
117 const UChar* ret = ures_getString(item, len, err);
118 *key = ures_getKey(item);
119 aiter->cursor++;
120 return ret;
121 #endif
122 }
123
124
125 #endif
126
127
128 U_NAMESPACE_BEGIN
129
130
131 // *****************************************************************************
132 // class DateTimePatternGenerator
133 // *****************************************************************************
134 static const UChar Canonical_Items[] = {
135 // GyQMwWEdDFHmsSv
136 CAP_G, LOW_Y, CAP_Q, CAP_M, LOW_W, CAP_W, CAP_E, LOW_D, CAP_D, CAP_F,
137 CAP_H, LOW_M, LOW_S, CAP_S, LOW_V, 0
138 };
139
140 static const dtTypeElem dtTypes[] = {
141 // patternChar, field, type, minLen, weight
142 {CAP_G, UDATPG_ERA_FIELD, DT_SHORT, 1, 3,},
143 {CAP_G, UDATPG_ERA_FIELD, DT_LONG, 4, 0},
144 {LOW_Y, UDATPG_YEAR_FIELD, DT_NUMERIC, 1, 20},
145 {CAP_Y, UDATPG_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 20},
146 {LOW_U, UDATPG_YEAR_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 20},
147 {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2},
148 {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0},
149 {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0},
150 {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
151 {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT + DT_DELTA, 3, 0},
152 {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG + DT_DELTA, 4, 0},
153 {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2},
154 {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0},
155 {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0},
156 {CAP_M, UDATPG_MONTH_FIELD, DT_NARROW, 5, 0},
157 {CAP_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
158 {CAP_L, UDATPG_MONTH_FIELD, DT_SHORT - DT_DELTA, 3, 0},
159 {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0},
160 {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0},
161 {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2},
162 {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 0},
163 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3},
164 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0},
165 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0},
166 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2},
167 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0},
168 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
169 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0},
170 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical
171 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0},
172 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0},
173 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0},
174 {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2},
175 {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 3},
176 {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 0},
177 {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20}, // really internal use, so we don't care
178 {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 0},
179 {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour
180 {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2},
181 {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour
182 {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
183 {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2},
184 {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2},
185 {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000},
186 {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 1000},
187 {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0},
188 {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
189 {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3},
190 {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0},
191 {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3},
192 {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
193 {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 3},
194 {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
195 {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[]
196 };
197
198 static const char* const CLDR_FIELD_APPEND[] = {
199 "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week", "Day", "*", "*", "*",
200 "Hour", "Minute", "Second", "*", "Timezone"
201 };
202
203 static const char* const CLDR_FIELD_NAME[] = {
204 "era", "year", "quarter", "month", "week", "*", "weekday", "day", "*", "*", "dayperiod",
205 "hour", "minute", "second", "*", "zone"
206 };
207
208 static const char* const Resource_Fields[] = {
209 "day", "dayperiod", "era", "hour", "minute", "month", "second", "week",
210 "weekday", "year", "zone", "quarter" };
211
212 // For appendItems
213 static const UChar UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A,
214 0x20, 0x7B, 0x31, 0x7D, 0x2524, 0}; // {0} \u251C{2}: {1}\u2524
215
216 static const UChar repeatedPatterns[6]={CAP_G, CAP_E, LOW_Z, LOW_V, CAP_Q, 0}; // "GEzvQ"
217
218 static const char DT_DateTimePatternsTag[]="DateTimePatterns";
219 static const char DT_DateTimeCalendarTag[]="calendar";
220 static const char DT_DateTimeGregorianTag[]="gregorian";
221 static const char DT_DateTimeAppendItemsTag[]="appendItems";
222 static const char DT_DateTimeFieldsTag[]="fields";
223 static const char DT_DateTimeAvailableFormatsTag[]="availableFormats";
224 //static const UnicodeString repeatedPattern=UnicodeString(repeatedPatterns);
225
226 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateTimePatternGenerator)
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)227 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)
228 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTRedundantEnumeration)
229
230 DateTimePatternGenerator* U_EXPORT2
231 DateTimePatternGenerator::createInstance(UErrorCode& status) {
232 return createInstance(Locale::getDefault(), status);
233 }
234
235 DateTimePatternGenerator* U_EXPORT2
createInstance(const Locale & locale,UErrorCode & status)236 DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) {
237 DateTimePatternGenerator *result = new DateTimePatternGenerator(locale, status);
238 if (result == NULL) {
239 status = U_MEMORY_ALLOCATION_ERROR;
240 }
241 if (U_FAILURE(status)) {
242 delete result;
243 result = NULL;
244 }
245 return result;
246 }
247
248 DateTimePatternGenerator* U_EXPORT2
createEmptyInstance(UErrorCode & status)249 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) {
250 DateTimePatternGenerator *result = new DateTimePatternGenerator(status);
251 if (result == NULL) {
252 status = U_MEMORY_ALLOCATION_ERROR;
253 }
254 if (U_FAILURE(status)) {
255 delete result;
256 result = NULL;
257 }
258 return result;
259 }
260
DateTimePatternGenerator(UErrorCode & status)261 DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) :
262 skipMatcher(NULL),
263 fAvailableFormatKeyHash(NULL)
264 {
265 fp = new FormatParser();
266 dtMatcher = new DateTimeMatcher();
267 distanceInfo = new DistanceInfo();
268 patternMap = new PatternMap();
269 if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) {
270 status = U_MEMORY_ALLOCATION_ERROR;
271 }
272 }
273
DateTimePatternGenerator(const Locale & locale,UErrorCode & status)274 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status) :
275 skipMatcher(NULL),
276 fAvailableFormatKeyHash(NULL)
277 {
278 fp = new FormatParser();
279 dtMatcher = new DateTimeMatcher();
280 distanceInfo = new DistanceInfo();
281 patternMap = new PatternMap();
282 if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) {
283 status = U_MEMORY_ALLOCATION_ERROR;
284 }
285 else {
286 initData(locale, status);
287 }
288 }
289
DateTimePatternGenerator(const DateTimePatternGenerator & other)290 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) :
291 UObject(),
292 skipMatcher(NULL),
293 fAvailableFormatKeyHash(NULL)
294 {
295 fp = new FormatParser();
296 dtMatcher = new DateTimeMatcher();
297 distanceInfo = new DistanceInfo();
298 patternMap = new PatternMap();
299 *this=other;
300 }
301
302 DateTimePatternGenerator&
operator =(const DateTimePatternGenerator & other)303 DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) {
304 pLocale = other.pLocale;
305 fDefaultHourFormatChar = other.fDefaultHourFormatChar;
306 *fp = *(other.fp);
307 dtMatcher->copyFrom(other.dtMatcher->skeleton);
308 *distanceInfo = *(other.distanceInfo);
309 dateTimeFormat = other.dateTimeFormat;
310 decimal = other.decimal;
311 // NUL-terminate for the C API.
312 dateTimeFormat.getTerminatedBuffer();
313 decimal.getTerminatedBuffer();
314 delete skipMatcher;
315 if ( other.skipMatcher == NULL ) {
316 skipMatcher = NULL;
317 }
318 else {
319 skipMatcher = new DateTimeMatcher(*other.skipMatcher);
320 }
321 for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) {
322 appendItemFormats[i] = other.appendItemFormats[i];
323 appendItemNames[i] = other.appendItemNames[i];
324 // NUL-terminate for the C API.
325 appendItemFormats[i].getTerminatedBuffer();
326 appendItemNames[i].getTerminatedBuffer();
327 }
328 UErrorCode status = U_ZERO_ERROR;
329 patternMap->copyFrom(*other.patternMap, status);
330 copyHashtable(other.fAvailableFormatKeyHash, status);
331 return *this;
332 }
333
334
335 UBool
operator ==(const DateTimePatternGenerator & other) const336 DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) const {
337 if (this == &other) {
338 return TRUE;
339 }
340 if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) &&
341 (dateTimeFormat==other.dateTimeFormat) && (decimal==other.decimal)) {
342 for ( int32_t i=0 ; i<UDATPG_FIELD_COUNT; ++i ) {
343 if ((appendItemFormats[i] != other.appendItemFormats[i]) ||
344 (appendItemNames[i] != other.appendItemNames[i]) ) {
345 return FALSE;
346 }
347 }
348 return TRUE;
349 }
350 else {
351 return FALSE;
352 }
353 }
354
355 UBool
operator !=(const DateTimePatternGenerator & other) const356 DateTimePatternGenerator::operator!=(const DateTimePatternGenerator& other) const {
357 return !operator==(other);
358 }
359
~DateTimePatternGenerator()360 DateTimePatternGenerator::~DateTimePatternGenerator() {
361 if (fAvailableFormatKeyHash!=NULL) {
362 delete fAvailableFormatKeyHash;
363 }
364
365 if (fp != NULL) delete fp;
366 if (dtMatcher != NULL) delete dtMatcher;
367 if (distanceInfo != NULL) delete distanceInfo;
368 if (patternMap != NULL) delete patternMap;
369 if (skipMatcher != NULL) delete skipMatcher;
370 }
371
372 void
initData(const Locale & locale,UErrorCode & status)373 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status) {
374 //const char *baseLangName = locale.getBaseName(); // unused
375
376 skipMatcher = NULL;
377 fAvailableFormatKeyHash=NULL;
378 addCanonicalItems();
379 addICUPatterns(locale, status);
380 if (U_FAILURE(status)) {
381 return;
382 }
383 addCLDRData(locale, status);
384 setDateTimeFromCalendar(locale, status);
385 setDecimalSymbols(locale, status);
386 } // DateTimePatternGenerator::initData
387
388 UnicodeString
getSkeleton(const UnicodeString & pattern,UErrorCode &)389 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode&
390 /*status*/) {
391 dtMatcher->set(pattern, fp);
392 return dtMatcher->getSkeletonPtr()->getSkeleton();
393 }
394
395 UnicodeString
getBaseSkeleton(const UnicodeString & pattern,UErrorCode &)396 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) {
397 dtMatcher->set(pattern, fp);
398 return dtMatcher->getSkeletonPtr()->getBaseSkeleton();
399 }
400
401 void
addICUPatterns(const Locale & locale,UErrorCode & status)402 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) {
403 UnicodeString dfPattern;
404 UnicodeString conflictingString;
405 UDateTimePatternConflict conflictingStatus;
406 DateFormat* df;
407
408 if (U_FAILURE(status)) {
409 return;
410 }
411
412 // Load with ICU patterns
413 for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) {
414 DateFormat::EStyle style = (DateFormat::EStyle)i;
415 df = DateFormat::createDateInstance(style, locale);
416 SimpleDateFormat* sdf;
417 if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
418 conflictingStatus = addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
419 }
420 // TODO Maybe we should return an error when the date format isn't simple.
421 delete df;
422 if (U_FAILURE(status)) {
423 return;
424 }
425
426 df = DateFormat::createTimeInstance(style, locale);
427 if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
428 conflictingStatus = addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
429 // HACK for hh:ss
430 if ( i==DateFormat::kMedium ) {
431 hackPattern = dfPattern;
432 }
433 }
434 // TODO Maybe we should return an error when the date format isn't simple.
435 delete df;
436 if (U_FAILURE(status)) {
437 return;
438 }
439 }
440 }
441
442 void
hackTimes(const UnicodeString & hackPattern,UErrorCode & status)443 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status) {
444 UDateTimePatternConflict conflictingStatus;
445 UnicodeString conflictingString;
446
447 fp->set(hackPattern);
448 UnicodeString mmss;
449 UBool gotMm=FALSE;
450 for (int32_t i=0; i<fp->itemNumber; ++i) {
451 UnicodeString field = fp->items[i];
452 if ( fp->isQuoteLiteral(field) ) {
453 if ( gotMm ) {
454 UnicodeString quoteLiteral;
455 fp->getQuoteLiteral(quoteLiteral, &i);
456 mmss += quoteLiteral;
457 }
458 }
459 else {
460 if (fp->isPatternSeparator(field) && gotMm) {
461 mmss+=field;
462 }
463 else {
464 UChar ch=field.charAt(0);
465 if (ch==LOW_M) {
466 gotMm=TRUE;
467 mmss+=field;
468 }
469 else {
470 if (ch==LOW_S) {
471 if (!gotMm) {
472 break;
473 }
474 mmss+= field;
475 conflictingStatus = addPattern(mmss, FALSE, conflictingString, status);
476 break;
477 }
478 else {
479 if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) {
480 break;
481 }
482 }
483 }
484 }
485 }
486 }
487 }
488
489 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY)
490
491 static const UChar hourFormatChars[] = { CAP_H, LOW_H, CAP_K, LOW_K, 0 }; // HhKk, the hour format characters
492
493 void
addCLDRData(const Locale & locale,UErrorCode & err)494 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& err) {
495 UResourceBundle *rb, *calTypeBundle, *calBundle;
496 UResourceBundle *patBundle, *fieldBundle, *fBundle;
497 UnicodeString rbPattern, value, field;
498 UnicodeString conflictingPattern;
499 UDateTimePatternConflict conflictingStatus;
500 const char *key=NULL;
501 int32_t i;
502
503 UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, LENGTHOF(UDATPG_ItemFormat)-1); // Read-only alias.
504
505 err = U_ZERO_ERROR;
506
507 fDefaultHourFormatChar = 0;
508 for (i=0; i<UDATPG_FIELD_COUNT; ++i ) {
509 appendItemNames[i]=CAP_F;
510 if (i<10) {
511 appendItemNames[i]+=(UChar)(i+0x30);
512 }
513 else {
514 appendItemNames[i]+=(UChar)0x31;
515 appendItemNames[i]+=(UChar)(i-10 + 0x30);
516 }
517 // NUL-terminate for the C API.
518 appendItemNames[i].getTerminatedBuffer();
519 }
520
521 rb = ures_open(NULL, locale.getName(), &err);
522 if (rb == NULL || U_FAILURE(err)) {
523 return;
524 }
525 const char *curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
526 const char * calendarTypeToUse = DT_DateTimeGregorianTag; // initial default
527 char calendarType[ULOC_KEYWORDS_CAPACITY]; // to be filled in with the type to use, if all goes well
528 if ( U_SUCCESS(err) ) {
529 char localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
530 // obtain a locale that always has the calendar key value that should be used
531 (void)ures_getFunctionalEquivalent(localeWithCalendarKey, ULOC_LOCALE_IDENTIFIER_CAPACITY, NULL,
532 "calendar", "calendar", locale.getName(), NULL, FALSE, &err);
533 localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination
534 // now get the calendar key value from that locale
535 int32_t calendarTypeLen = uloc_getKeywordValue(localeWithCalendarKey, "calendar", calendarType, ULOC_KEYWORDS_CAPACITY, &err);
536 if (U_SUCCESS(err) && calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
537 calendarTypeToUse = calendarType;
538 }
539 err = U_ZERO_ERROR;
540 }
541 calBundle = ures_getByKeyWithFallback(rb, DT_DateTimeCalendarTag, NULL, &err);
542 calTypeBundle = ures_getByKeyWithFallback(calBundle, calendarTypeToUse, NULL, &err);
543
544 key=NULL;
545 int32_t dtCount=0;
546 patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimePatternsTag, NULL, &err);
547 while (U_SUCCESS(err)) {
548 rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
549 dtCount++;
550 if (rbPattern.length()==0 ) {
551 break; // no more pattern
552 }
553 else {
554 if (dtCount==9) {
555 setDateTimeFormat(rbPattern);
556 } else if (dtCount==4) { // short time format
557 // set fDefaultHourFormatChar to the hour format character from this pattern
558 int32_t tfIdx, tfLen = rbPattern.length();
559 UBool ignoreChars = FALSE;
560 for (tfIdx = 0; tfIdx < tfLen; tfIdx++) {
561 UChar tfChar = rbPattern.charAt(tfIdx);
562 if ( tfChar == SINGLE_QUOTE ) {
563 ignoreChars = !ignoreChars; // toggle (handle quoted literals & '' for single quote)
564 } else if ( !ignoreChars && u_strchr(hourFormatChars, tfChar) != NULL ) {
565 fDefaultHourFormatChar = tfChar;
566 break;
567 }
568 }
569 }
570 }
571 }
572 ures_close(patBundle);
573
574 err = U_ZERO_ERROR;
575 patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAppendItemsTag, NULL, &err);
576 key=NULL;
577 UnicodeString itemKey;
578 while (U_SUCCESS(err)) {
579 rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
580 if (rbPattern.length()==0 ) {
581 break; // no more pattern
582 }
583 else {
584 setAppendItemFormat(getAppendFormatNumber(key), rbPattern);
585 }
586 }
587 ures_close(patBundle);
588
589 key=NULL;
590 err = U_ZERO_ERROR;
591 fBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeFieldsTag, NULL, &err);
592 for (i=0; i<MAX_RESOURCE_FIELD; ++i) {
593 err = U_ZERO_ERROR;
594 patBundle = ures_getByKeyWithFallback(fBundle, Resource_Fields[i], NULL, &err);
595 fieldBundle = ures_getByKeyWithFallback(patBundle, "dn", NULL, &err);
596 rbPattern = ures_getNextUnicodeString(fieldBundle, &key, &err);
597 ures_close(fieldBundle);
598 ures_close(patBundle);
599 if (rbPattern.length()==0 ) {
600 continue;
601 }
602 else {
603 setAppendItemName(getAppendNameNumber(Resource_Fields[i]), rbPattern);
604 }
605 }
606 ures_close(fBundle);
607
608 // add available formats
609 err = U_ZERO_ERROR;
610 initHashtable(err);
611 patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAvailableFormatsTag, NULL, &err);
612 if (U_SUCCESS(err)) {
613 int32_t numberKeys = ures_getSize(patBundle);
614 int32_t len;
615 const UChar *retPattern;
616 key=NULL;
617 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
618 UResourceBundleAIterator aiter;
619 ures_a_open(&aiter, patBundle, &err);
620 #endif
621 for(i=0; i<numberKeys; ++i) {
622 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
623 retPattern=ures_a_getNextString(&aiter, &len, &key, &err);
624 #else
625 retPattern=ures_getNextString(patBundle, &len, &key, &err);
626 #endif
627 UnicodeString format=UnicodeString(retPattern);
628 UnicodeString retKey=UnicodeString(key, -1, US_INV);
629 setAvailableFormat(retKey, err);
630 // Add pattern with its associated skeleton. Override any duplicate derived from std patterns,
631 // but not a previous availableFormats entry:
632 conflictingStatus = addPatternWithSkeleton(format, &retKey, TRUE, conflictingPattern, err);
633 }
634 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
635 ures_a_close(&aiter);
636 #endif
637 }
638 ures_close(patBundle);
639 ures_close(calTypeBundle);
640 ures_close(calBundle);
641 ures_close(rb);
642
643 err = U_ZERO_ERROR;
644 char parentLocale[50];
645 int32_t localeNameLen=0;
646 uprv_strcpy(parentLocale, curLocaleName);
647 while((localeNameLen=uloc_getParent(parentLocale, parentLocale, 50, &err))>=0 ) {
648 rb = ures_open(NULL, parentLocale, &err);
649 curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
650 uprv_strcpy(parentLocale, curLocaleName);
651 calBundle = ures_getByKey(rb, DT_DateTimeCalendarTag, NULL, &err);
652 calTypeBundle = ures_getByKey(calBundle, calendarTypeToUse, NULL, &err);
653 patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAvailableFormatsTag, NULL, &err);
654 if (U_SUCCESS(err)) {
655 int32_t numberKeys = ures_getSize(patBundle);
656 int32_t len;
657 const UChar *retPattern;
658 key=NULL;
659 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
660 UResourceBundleAIterator aiter;
661 ures_a_open(&aiter, patBundle, &err);
662 #endif
663 for(i=0; i<numberKeys; ++i) {
664 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
665 retPattern=ures_a_getNextString(&aiter, &len, &key, &err);
666 #else
667 retPattern=ures_getNextString(patBundle, &len, &key, &err);
668 #endif
669 UnicodeString format=UnicodeString(retPattern);
670 UnicodeString retKey=UnicodeString(key, -1, US_INV);
671 if ( !isAvailableFormatSet(retKey) ) {
672 setAvailableFormat(retKey, err);
673 // Add pattern with its associated skeleton. Override any duplicate derived from std patterns,
674 // but not a previous availableFormats entry:
675 conflictingStatus = addPatternWithSkeleton(format, &retKey, TRUE, conflictingPattern, err);
676 }
677 }
678 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
679 ures_a_close(&aiter);
680 #endif
681 }
682 err = U_ZERO_ERROR; // reset; if this locale lacks the necessary data, need to keep checking up to root.
683 ures_close(patBundle);
684 ures_close(calTypeBundle);
685 ures_close(calBundle);
686 ures_close(rb);
687 if (localeNameLen==0) {
688 break;
689 }
690 }
691
692 if (hackPattern.length()>0) {
693 hackTimes(hackPattern, err);
694 }
695 }
696
697 void
initHashtable(UErrorCode & err)698 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
699 if (fAvailableFormatKeyHash!=NULL) {
700 return;
701 }
702 if ((fAvailableFormatKeyHash = new Hashtable(FALSE, err))==NULL) {
703 err=U_MEMORY_ALLOCATION_ERROR;
704 return;
705 }
706 }
707
708
709 void
setAppendItemFormat(UDateTimePatternField field,const UnicodeString & value)710 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
711 appendItemFormats[field] = value;
712 // NUL-terminate for the C API.
713 appendItemFormats[field].getTerminatedBuffer();
714 }
715
716 const UnicodeString&
getAppendItemFormat(UDateTimePatternField field) const717 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
718 return appendItemFormats[field];
719 }
720
721 void
setAppendItemName(UDateTimePatternField field,const UnicodeString & value)722 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
723 appendItemNames[field] = value;
724 // NUL-terminate for the C API.
725 appendItemNames[field].getTerminatedBuffer();
726 }
727
728 const UnicodeString&
getAppendItemName(UDateTimePatternField field) const729 DateTimePatternGenerator:: getAppendItemName(UDateTimePatternField field) const {
730 return appendItemNames[field];
731 }
732
733 void
getAppendName(UDateTimePatternField field,UnicodeString & value)734 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
735 value = SINGLE_QUOTE;
736 value += appendItemNames[field];
737 value += SINGLE_QUOTE;
738 }
739
740 UnicodeString
getBestPattern(const UnicodeString & patternForm,UErrorCode & status)741 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
742 return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
743 }
744
745 UnicodeString
getBestPattern(const UnicodeString & patternForm,UDateTimePatternMatchOptions options,UErrorCode & status)746 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
747 const UnicodeString *bestPattern=NULL;
748 UnicodeString dtFormat;
749 UnicodeString resultPattern;
750
751 int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
752 int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
753
754 UnicodeString patternFormCopy = UnicodeString(patternForm);
755 patternFormCopy.findAndReplace(UnicodeString(LOW_J), UnicodeString(fDefaultHourFormatChar));
756
757 resultPattern.remove();
758 dtMatcher->set(patternFormCopy, fp);
759 const PtnSkeleton* specifiedSkeleton=NULL;
760 bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, &specifiedSkeleton);
761 if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
762 resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, FALSE, options);
763
764 return resultPattern;
765 }
766 int32_t neededFields = dtMatcher->getFieldMask();
767 UnicodeString datePattern=getBestAppending(neededFields & dateMask, options);
768 UnicodeString timePattern=getBestAppending(neededFields & timeMask, options);
769 if (datePattern.length()==0) {
770 if (timePattern.length()==0) {
771 resultPattern.remove();
772 }
773 else {
774 return timePattern;
775 }
776 }
777 if (timePattern.length()==0) {
778 return datePattern;
779 }
780 resultPattern.remove();
781 status = U_ZERO_ERROR;
782 dtFormat=getDateTimeFormat();
783 Formattable dateTimeObject[] = { timePattern, datePattern };
784 resultPattern = MessageFormat::format(dtFormat, dateTimeObject, 2, resultPattern, status );
785 return resultPattern;
786 }
787
788 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UErrorCode & status)789 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
790 const UnicodeString& skeleton,
791 UErrorCode& status) {
792 return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
793 }
794
795 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UDateTimePatternMatchOptions options,UErrorCode &)796 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
797 const UnicodeString& skeleton,
798 UDateTimePatternMatchOptions options,
799 UErrorCode& /*status*/) {
800 dtMatcher->set(skeleton, fp);
801 UnicodeString result = adjustFieldTypes(pattern, NULL, FALSE, options);
802 return result;
803 }
804
805 void
setDecimal(const UnicodeString & newDecimal)806 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
807 this->decimal = newDecimal;
808 // NUL-terminate for the C API.
809 this->decimal.getTerminatedBuffer();
810 }
811
812 const UnicodeString&
getDecimal() const813 DateTimePatternGenerator::getDecimal() const {
814 return decimal;
815 }
816
817 void
addCanonicalItems()818 DateTimePatternGenerator::addCanonicalItems() {
819 UnicodeString conflictingPattern;
820 UDateTimePatternConflict conflictingStatus;
821 UErrorCode status = U_ZERO_ERROR;
822
823 for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
824 conflictingStatus = addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
825 }
826 }
827
828 void
setDateTimeFormat(const UnicodeString & dtFormat)829 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
830 dateTimeFormat = dtFormat;
831 // NUL-terminate for the C API.
832 dateTimeFormat.getTerminatedBuffer();
833 }
834
835 const UnicodeString&
getDateTimeFormat() const836 DateTimePatternGenerator::getDateTimeFormat() const {
837 return dateTimeFormat;
838 }
839
840 void
setDateTimeFromCalendar(const Locale & locale,UErrorCode & status)841 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
842 const UChar *resStr;
843 int32_t resStrLen = 0;
844
845 Calendar* fCalendar = Calendar::createInstance(locale, status);
846 CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
847 UResourceBundle *dateTimePatterns = calData.getByKey(DT_DateTimePatternsTag, status);
848 if (U_FAILURE(status)) return;
849
850 if (ures_getSize(dateTimePatterns) <= DateFormat::kDateTime)
851 {
852 status = U_INVALID_FORMAT_ERROR;
853 return;
854 }
855 resStr = ures_getStringByIndex(dateTimePatterns, (int32_t)DateFormat::kDateTime, &resStrLen, &status);
856 setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
857
858 delete fCalendar;
859 }
860
861 void
setDecimalSymbols(const Locale & locale,UErrorCode & status)862 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
863 DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
864 if(U_SUCCESS(status)) {
865 decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
866 // NUL-terminate for the C API.
867 decimal.getTerminatedBuffer();
868 }
869 }
870
871 UDateTimePatternConflict
addPattern(const UnicodeString & pattern,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)872 DateTimePatternGenerator::addPattern(
873 const UnicodeString& pattern,
874 UBool override,
875 UnicodeString &conflictingPattern,
876 UErrorCode& status)
877 {
878 return addPatternWithSkeleton(pattern, NULL, override, conflictingPattern, status);
879 }
880
881 // For DateTimePatternGenerator::addPatternWithSkeleton -
882 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
883 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
884 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
885 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
886 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
887 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
888 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
889 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
890 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
891 UDateTimePatternConflict
addPatternWithSkeleton(const UnicodeString & pattern,const UnicodeString * skeletonToUse,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)892 DateTimePatternGenerator::addPatternWithSkeleton(
893 const UnicodeString& pattern,
894 const UnicodeString* skeletonToUse,
895 UBool override,
896 UnicodeString& conflictingPattern,
897 UErrorCode& status)
898 {
899
900 UnicodeString basePattern;
901 PtnSkeleton skeleton;
902 UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
903
904 DateTimeMatcher matcher;
905 if ( skeletonToUse == NULL ) {
906 matcher.set(pattern, fp, skeleton);
907 matcher.getBasePattern(basePattern);
908 } else {
909 matcher.set(*skeletonToUse, fp, skeleton); // this still trims skeleton fields to max len 3, may need to change it.
910 matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
911 }
912 UBool entryHadSpecifiedSkeleton;
913 const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
914 if (duplicatePattern != NULL ) {
915 conflictingStatus = UDATPG_BASE_CONFLICT;
916 conflictingPattern = *duplicatePattern;
917 if (!override || (skeletonToUse != NULL && entryHadSpecifiedSkeleton)) {
918 return conflictingStatus;
919 }
920 }
921 const PtnSkeleton* entrySpecifiedSkeleton = NULL;
922 duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
923 if (duplicatePattern != NULL ) {
924 conflictingStatus = UDATPG_CONFLICT;
925 conflictingPattern = *duplicatePattern;
926 if (!override || (skeletonToUse != NULL && entrySpecifiedSkeleton != NULL)) {
927 return conflictingStatus;
928 }
929 }
930 patternMap->add(basePattern, skeleton, pattern, skeletonToUse != NULL, status);
931 if(U_FAILURE(status)) {
932 return conflictingStatus;
933 }
934
935 return UDATPG_NO_CONFLICT;
936 }
937
938
939 UDateTimePatternField
getAppendFormatNumber(const char * field) const940 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
941 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
942 if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
943 return (UDateTimePatternField)i;
944 }
945 }
946 return UDATPG_FIELD_COUNT;
947 }
948
949 UDateTimePatternField
getAppendNameNumber(const char * field) const950 DateTimePatternGenerator::getAppendNameNumber(const char* field) const {
951 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
952 if (uprv_strcmp(CLDR_FIELD_NAME[i],field)==0) {
953 return (UDateTimePatternField)i;
954 }
955 }
956 return UDATPG_FIELD_COUNT;
957 }
958
959 const UnicodeString*
getBestRaw(DateTimeMatcher & source,int32_t includeMask,DistanceInfo * missingFields,const PtnSkeleton ** specifiedSkeletonPtr)960 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
961 int32_t includeMask,
962 DistanceInfo* missingFields,
963 const PtnSkeleton** specifiedSkeletonPtr) {
964 int32_t bestDistance = 0x7fffffff;
965 DistanceInfo tempInfo;
966 const UnicodeString *bestPattern=NULL;
967 const PtnSkeleton* specifiedSkeleton=NULL;
968
969 PatternMapIterator it;
970 for (it.set(*patternMap); it.hasNext(); ) {
971 DateTimeMatcher trial = it.next();
972 if (trial.equals(skipMatcher)) {
973 continue;
974 }
975 int32_t distance=source.getDistance(trial, includeMask, tempInfo);
976 if (distance<bestDistance) {
977 bestDistance=distance;
978 bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
979 missingFields->setTo(tempInfo);
980 if (distance==0) {
981 break;
982 }
983 }
984 }
985
986 // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
987 // then return it too. This generally happens when the caller needs to pass that skeleton
988 // through to adjustFieldTypes so the latter can do a better job.
989 if (bestPattern && specifiedSkeletonPtr) {
990 *specifiedSkeletonPtr = specifiedSkeleton;
991 }
992 return bestPattern;
993 }
994
995 UnicodeString
adjustFieldTypes(const UnicodeString & pattern,const PtnSkeleton * specifiedSkeleton,UBool fixFractionalSeconds,UDateTimePatternMatchOptions options)996 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
997 const PtnSkeleton* specifiedSkeleton,
998 UBool fixFractionalSeconds,
999 UDateTimePatternMatchOptions options) {
1000 UnicodeString newPattern;
1001 fp->set(pattern);
1002 for (int32_t i=0; i < fp->itemNumber; i++) {
1003 UnicodeString field = fp->items[i];
1004 if ( fp->isQuoteLiteral(field) ) {
1005
1006 UnicodeString quoteLiteral;
1007 fp->getQuoteLiteral(quoteLiteral, &i);
1008 newPattern += quoteLiteral;
1009 }
1010 else {
1011 if (fp->isPatternSeparator(field)) {
1012 newPattern+=field;
1013 continue;
1014 }
1015 int32_t canonicalIndex = fp->getCanonicalIndex(field);
1016 if (canonicalIndex < 0) {
1017 newPattern+=field;
1018 continue; // don't adjust
1019 }
1020 const dtTypeElem *row = &dtTypes[canonicalIndex];
1021 int32_t typeValue = row->field;
1022 if (fixFractionalSeconds && typeValue == UDATPG_SECOND_FIELD) {
1023 UnicodeString newField=dtMatcher->skeleton.original[UDATPG_FRACTIONAL_SECOND_FIELD];
1024 field = field + decimal + newField;
1025 }
1026 else {
1027 if (dtMatcher->skeleton.type[typeValue]!=0) {
1028 // Here:
1029 // - "reqField" is the field from the originally requested skeleton, with length
1030 // "reqFieldLen".
1031 // - "field" is the field from the found pattern.
1032 //
1033 // The adjusted field should consist of characters from the originally requested
1034 // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1035 // UDATPG_WEEKDAY_FIELD, in which case it should consist of characters from the
1036 // found pattern.
1037 //
1038 // The length of the adjusted field (adjFieldLen) should match that in the originally
1039 // requested skeleton, except that in the following cases the length of the adjusted field
1040 // should match that in the found pattern (i.e. the length of this pattern field should
1041 // not be adjusted):
1042 // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1043 // not set (ticket #7180). Note, we may want to implement a similar change for other
1044 // numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1045 // field length, but options bits can be used to override this.
1046 // 2. There is a specified skeleton for the found pattern and one of the following is true:
1047 // a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1048 // b) The pattern field is numeric and the skeleton field is not, or vice versa.
1049
1050 UnicodeString reqField = dtMatcher->skeleton.original[typeValue];
1051 int32_t reqFieldLen = reqField.length();
1052 if (reqField.charAt(0) == CAP_E && reqFieldLen < 3)
1053 reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1054 int32_t adjFieldLen = reqFieldLen;
1055 if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1056 (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1057 (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1058 adjFieldLen = field.length();
1059 } else if (specifiedSkeleton) {
1060 UnicodeString skelField = specifiedSkeleton->original[typeValue];
1061 int32_t skelFieldLen = skelField.length();
1062 UBool patFieldIsNumeric = (row->type > 0);
1063 UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1064 if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1065 // don't adjust the field length in the found pattern
1066 adjFieldLen = field.length();
1067 }
1068 }
1069 UChar c = (typeValue!= UDATPG_HOUR_FIELD && typeValue!= UDATPG_MONTH_FIELD && typeValue!= UDATPG_WEEKDAY_FIELD)?
1070 reqField.charAt(0): field.charAt(0);
1071 field.remove();
1072 for (int32_t i=adjFieldLen; i>0; --i) {
1073 field+=c;
1074 }
1075 }
1076 newPattern+=field;
1077 }
1078 }
1079 }
1080 return newPattern;
1081 }
1082
1083 UnicodeString
getBestAppending(int32_t missingFields,UDateTimePatternMatchOptions options)1084 DateTimePatternGenerator::getBestAppending(int32_t missingFields, UDateTimePatternMatchOptions options) {
1085 UnicodeString resultPattern, tempPattern, formattedPattern;
1086 UErrorCode err=U_ZERO_ERROR;
1087 int32_t lastMissingFieldMask=0;
1088 if (missingFields!=0) {
1089 resultPattern=UnicodeString();
1090 const PtnSkeleton* specifiedSkeleton=NULL;
1091 tempPattern = *getBestRaw(*dtMatcher, missingFields, distanceInfo, &specifiedSkeleton);
1092 resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options);
1093 if ( distanceInfo->missingFieldMask==0 ) {
1094 return resultPattern;
1095 }
1096 while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1097 if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1098 break; // cannot find the proper missing field
1099 }
1100 if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1101 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1102 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, FALSE, options);
1103 //resultPattern = tempPattern;
1104 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1105 continue;
1106 }
1107 int32_t startingMask = distanceInfo->missingFieldMask;
1108 tempPattern = *getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, &specifiedSkeleton);
1109 tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, FALSE, options);
1110 int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1111 int32_t topField=getTopBitNumber(foundMask);
1112 UnicodeString appendName;
1113 getAppendName((UDateTimePatternField)topField, appendName);
1114 const Formattable formatPattern[] = {
1115 resultPattern,
1116 tempPattern,
1117 appendName
1118 };
1119 UnicodeString emptyStr;
1120 formattedPattern = MessageFormat::format(appendItemFormats[topField], formatPattern, 3, emptyStr, err);
1121 lastMissingFieldMask = distanceInfo->missingFieldMask;
1122 }
1123 }
1124 return formattedPattern;
1125 }
1126
1127 int32_t
getTopBitNumber(int32_t foundMask)1128 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) {
1129 if ( foundMask==0 ) {
1130 return 0;
1131 }
1132 int32_t i=0;
1133 while (foundMask!=0) {
1134 foundMask >>=1;
1135 ++i;
1136 }
1137 if (i-1 >UDATPG_ZONE_FIELD) {
1138 return UDATPG_ZONE_FIELD;
1139 }
1140 else
1141 return i-1;
1142 }
1143
1144 void
setAvailableFormat(const UnicodeString & key,UErrorCode & err)1145 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1146 {
1147 fAvailableFormatKeyHash->puti(key, 1, err);
1148 }
1149
1150 UBool
isAvailableFormatSet(const UnicodeString & key) const1151 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1152 return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1153 }
1154
1155 void
copyHashtable(Hashtable * other,UErrorCode & status)1156 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1157
1158 if (other == NULL) {
1159 return;
1160 }
1161 if (fAvailableFormatKeyHash != NULL) {
1162 delete fAvailableFormatKeyHash;
1163 fAvailableFormatKeyHash = NULL;
1164 }
1165 initHashtable(status);
1166 if(U_FAILURE(status)){
1167 return;
1168 }
1169 int32_t pos = -1;
1170 const UHashElement* elem = NULL;
1171 // walk through the hash table and create a deep clone
1172 while((elem = other->nextElement(pos))!= NULL){
1173 const UHashTok otherKeyTok = elem->key;
1174 UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1175 fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1176 if(U_FAILURE(status)){
1177 return;
1178 }
1179 }
1180 }
1181
1182 StringEnumeration*
getSkeletons(UErrorCode & status) const1183 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1184 StringEnumeration* skeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status);
1185 return skeletonEnumerator;
1186 }
1187
1188 const UnicodeString&
getPatternForSkeleton(const UnicodeString & skeleton) const1189 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1190 PtnElem *curElem;
1191
1192 if (skeleton.length() ==0) {
1193 return emptyString;
1194 }
1195 curElem = patternMap->getHeader(skeleton.charAt(0));
1196 while ( curElem != NULL ) {
1197 if ( curElem->skeleton->getSkeleton()==skeleton ) {
1198 return curElem->pattern;
1199 }
1200 curElem=curElem->next;
1201 }
1202 return emptyString;
1203 }
1204
1205 StringEnumeration*
getBaseSkeletons(UErrorCode & status) const1206 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1207 StringEnumeration* baseSkeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status);
1208 return baseSkeletonEnumerator;
1209 }
1210
1211 StringEnumeration*
getRedundants(UErrorCode & status)1212 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1213 StringEnumeration* output = new DTRedundantEnumeration();
1214 const UnicodeString *pattern;
1215 PatternMapIterator it;
1216 for (it.set(*patternMap); it.hasNext(); ) {
1217 DateTimeMatcher current = it.next();
1218 pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1219 if ( isCanonicalItem(*pattern) ) {
1220 continue;
1221 }
1222 if ( skipMatcher == NULL ) {
1223 skipMatcher = new DateTimeMatcher(current);
1224 }
1225 else {
1226 *skipMatcher = current;
1227 }
1228 UnicodeString trial = getBestPattern(current.getPattern(), status);
1229 if (trial == *pattern) {
1230 ((DTRedundantEnumeration *)output)->add(*pattern, status);
1231 }
1232 if (current.equals(skipMatcher)) {
1233 continue;
1234 }
1235 }
1236 return output;
1237 }
1238
1239 UBool
isCanonicalItem(const UnicodeString & item) const1240 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1241 if ( item.length() != 1 ) {
1242 return FALSE;
1243 }
1244 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1245 if (item.charAt(0)==Canonical_Items[i]) {
1246 return TRUE;
1247 }
1248 }
1249 return FALSE;
1250 }
1251
1252
1253 DateTimePatternGenerator*
clone() const1254 DateTimePatternGenerator::clone() const {
1255 return new DateTimePatternGenerator(*this);
1256 }
1257
PatternMap()1258 PatternMap::PatternMap() {
1259 for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1260 boot[i]=NULL;
1261 }
1262 isDupAllowed = TRUE;
1263 }
1264
1265 void
copyFrom(const PatternMap & other,UErrorCode & status)1266 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1267 this->isDupAllowed = other.isDupAllowed;
1268 for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1269 PtnElem *curElem, *otherElem, *prevElem=NULL;
1270 otherElem = other.boot[bootIndex];
1271 while (otherElem!=NULL) {
1272 if ((curElem = new PtnElem(otherElem->basePattern, otherElem->pattern))==NULL) {
1273 // out of memory
1274 status = U_MEMORY_ALLOCATION_ERROR;
1275 return;
1276 }
1277 if ( this->boot[bootIndex]== NULL ) {
1278 this->boot[bootIndex] = curElem;
1279 }
1280 if ((curElem->skeleton=new PtnSkeleton(*(otherElem->skeleton))) == NULL ) {
1281 // out of memory
1282 status = U_MEMORY_ALLOCATION_ERROR;
1283 return;
1284 }
1285
1286 if (prevElem!=NULL) {
1287 prevElem->next=curElem;
1288 }
1289 curElem->next=NULL;
1290 prevElem = curElem;
1291 otherElem = otherElem->next;
1292 }
1293
1294 }
1295 }
1296
1297 PtnElem*
getHeader(UChar baseChar)1298 PatternMap::getHeader(UChar baseChar) {
1299 PtnElem* curElem;
1300
1301 if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1302 curElem = boot[baseChar-CAP_A];
1303 }
1304 else {
1305 if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1306 curElem = boot[26+baseChar-LOW_A];
1307 }
1308 else {
1309 return NULL;
1310 }
1311 }
1312 return curElem;
1313 }
1314
~PatternMap()1315 PatternMap::~PatternMap() {
1316 for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1317 if (boot[i]!=NULL ) {
1318 delete boot[i];
1319 boot[i]=NULL;
1320 }
1321 }
1322 } // PatternMap destructor
1323
1324 void
add(const UnicodeString & basePattern,const PtnSkeleton & skeleton,const UnicodeString & value,UBool skeletonWasSpecified,UErrorCode & status)1325 PatternMap::add(const UnicodeString& basePattern,
1326 const PtnSkeleton& skeleton,
1327 const UnicodeString& value,// mapped pattern value
1328 UBool skeletonWasSpecified,
1329 UErrorCode &status) {
1330 UChar baseChar = basePattern.charAt(0);
1331 PtnElem *curElem, *baseElem;
1332 status = U_ZERO_ERROR;
1333
1334 // the baseChar must be A-Z or a-z
1335 if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1336 baseElem = boot[baseChar-CAP_A];
1337 }
1338 else {
1339 if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1340 baseElem = boot[26+baseChar-LOW_A];
1341 }
1342 else {
1343 status = U_ILLEGAL_CHARACTER;
1344 return;
1345 }
1346 }
1347
1348 if (baseElem == NULL) {
1349 if ((curElem = new PtnElem(basePattern, value)) == NULL ) {
1350 // out of memory
1351 status = U_MEMORY_ALLOCATION_ERROR;
1352 return;
1353 }
1354 if (baseChar >= LOW_A) {
1355 boot[26 + (baseChar-LOW_A)] = curElem;
1356 }
1357 else {
1358 boot[baseChar-CAP_A] = curElem;
1359 }
1360 curElem->skeleton = new PtnSkeleton(skeleton);
1361 curElem->skeletonWasSpecified = skeletonWasSpecified;
1362 }
1363 if ( baseElem != NULL ) {
1364 curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1365
1366 if (curElem == NULL) {
1367 // add new element to the list.
1368 curElem = baseElem;
1369 while( curElem -> next != NULL )
1370 {
1371 curElem = curElem->next;
1372 }
1373 if ((curElem->next = new PtnElem(basePattern, value)) == NULL ) {
1374 // out of memory
1375 status = U_MEMORY_ALLOCATION_ERROR;
1376 return;
1377 }
1378 curElem=curElem->next;
1379 curElem->skeleton = new PtnSkeleton(skeleton);
1380 curElem->skeletonWasSpecified = skeletonWasSpecified;
1381 }
1382 else {
1383 // Pattern exists in the list already.
1384 if ( !isDupAllowed ) {
1385 return;
1386 }
1387 // Overwrite the value.
1388 curElem->pattern = value;
1389 }
1390 }
1391 } // PatternMap::add
1392
1393 // Find the pattern from the given basePattern string.
1394 const UnicodeString *
getPatternFromBasePattern(UnicodeString & basePattern,UBool & skeletonWasSpecified)1395 PatternMap::getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeletonWasSpecified) { // key to search for
1396 PtnElem *curElem;
1397
1398 if ((curElem=getHeader(basePattern.charAt(0)))==NULL) {
1399 return NULL; // no match
1400 }
1401
1402 do {
1403 if ( basePattern.compare(curElem->basePattern)==0 ) {
1404 skeletonWasSpecified = curElem->skeletonWasSpecified;
1405 return &(curElem->pattern);
1406 }
1407 curElem=curElem->next;
1408 }while (curElem != NULL);
1409
1410 return NULL;
1411 } // PatternMap::getFromBasePattern
1412
1413
1414 // Find the pattern from the given skeleton.
1415 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
1416 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
1417 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
1418 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
1419 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
1420 const UnicodeString *
getPatternFromSkeleton(PtnSkeleton & skeleton,const PtnSkeleton ** specifiedSkeletonPtr)1421 PatternMap::getPatternFromSkeleton(PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) { // key to search for
1422 PtnElem *curElem;
1423
1424 if (specifiedSkeletonPtr) {
1425 *specifiedSkeletonPtr = NULL;
1426 }
1427
1428 // find boot entry
1429 UChar baseChar='\0';
1430 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1431 if (skeleton.baseOriginal[i].length() !=0 ) {
1432 baseChar = skeleton.baseOriginal[i].charAt(0);
1433 break;
1434 }
1435 }
1436
1437 if ((curElem=getHeader(baseChar))==NULL) {
1438 return NULL; // no match
1439 }
1440
1441 do {
1442 int32_t i=0;
1443 if (specifiedSkeletonPtr != NULL) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
1444 for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1445 if (curElem->skeleton->original[i].compare(skeleton.original[i]) != 0 )
1446 {
1447 break;
1448 }
1449 }
1450 } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
1451 for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1452 if (curElem->skeleton->baseOriginal[i].compare(skeleton.baseOriginal[i]) != 0 )
1453 {
1454 break;
1455 }
1456 }
1457 }
1458 if (i == UDATPG_FIELD_COUNT) {
1459 if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
1460 *specifiedSkeletonPtr = curElem->skeleton;
1461 }
1462 return &(curElem->pattern);
1463 }
1464 curElem=curElem->next;
1465 }while (curElem != NULL);
1466
1467 return NULL;
1468 }
1469
1470 UBool
equals(const PatternMap & other)1471 PatternMap::equals(const PatternMap& other) {
1472 if ( this==&other ) {
1473 return TRUE;
1474 }
1475 for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1476 if ( boot[bootIndex]==other.boot[bootIndex] ) {
1477 continue;
1478 }
1479 if ( (boot[bootIndex]==NULL)||(other.boot[bootIndex]==NULL) ) {
1480 return FALSE;
1481 }
1482 PtnElem *otherElem = other.boot[bootIndex];
1483 PtnElem *myElem = boot[bootIndex];
1484 while ((otherElem!=NULL) || (myElem!=NULL)) {
1485 if ( myElem == otherElem ) {
1486 break;
1487 }
1488 if ((otherElem==NULL) || (myElem==NULL)) {
1489 return FALSE;
1490 }
1491 if ( (myElem->basePattern != otherElem->basePattern) ||
1492 (myElem->pattern != otherElem->pattern) ) {
1493 return FALSE;
1494 }
1495 if ((myElem->skeleton!=otherElem->skeleton)&&
1496 !myElem->skeleton->equals(*(otherElem->skeleton))) {
1497 return FALSE;
1498 }
1499 myElem = myElem->next;
1500 otherElem=otherElem->next;
1501 }
1502 }
1503 return TRUE;
1504 }
1505
1506 // find any key existing in the mapping table already.
1507 // return TRUE if there is an existing key, otherwise return FALSE.
1508 PtnElem*
getDuplicateElem(const UnicodeString & basePattern,const PtnSkeleton & skeleton,PtnElem * baseElem)1509 PatternMap::getDuplicateElem(
1510 const UnicodeString &basePattern,
1511 const PtnSkeleton &skeleton,
1512 PtnElem *baseElem) {
1513 PtnElem *curElem;
1514
1515 if ( baseElem == (PtnElem *)NULL ) {
1516 return (PtnElem*)NULL;
1517 }
1518 else {
1519 curElem = baseElem;
1520 }
1521 do {
1522 if ( basePattern.compare(curElem->basePattern)==0 ) {
1523 UBool isEqual=TRUE;
1524 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1525 if (curElem->skeleton->type[i] != skeleton.type[i] ) {
1526 isEqual=FALSE;
1527 break;
1528 }
1529 }
1530 if (isEqual) {
1531 return curElem;
1532 }
1533 }
1534 curElem = curElem->next;
1535 } while( curElem != (PtnElem *)NULL );
1536
1537 // end of the list
1538 return (PtnElem*)NULL;
1539
1540 } // PatternMap::getDuplicateElem
1541
DateTimeMatcher(void)1542 DateTimeMatcher::DateTimeMatcher(void) {
1543 }
1544
DateTimeMatcher(const DateTimeMatcher & other)1545 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
1546 copyFrom(other.skeleton);
1547 }
1548
1549
1550 void
set(const UnicodeString & pattern,FormatParser * fp)1551 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
1552 PtnSkeleton localSkeleton;
1553 return set(pattern, fp, localSkeleton);
1554 }
1555
1556 void
set(const UnicodeString & pattern,FormatParser * fp,PtnSkeleton & skeletonResult)1557 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
1558 int32_t i;
1559 for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1560 skeletonResult.type[i]=NONE;
1561 }
1562 fp->set(pattern);
1563 for (i=0; i < fp->itemNumber; i++) {
1564 UnicodeString field = fp->items[i];
1565 if ( field.charAt(0) == LOW_A ) {
1566 continue; // skip 'a'
1567 }
1568
1569 if ( fp->isQuoteLiteral(field) ) {
1570 UnicodeString quoteLiteral;
1571 fp->getQuoteLiteral(quoteLiteral, &i);
1572 continue;
1573 }
1574 int32_t canonicalIndex = fp->getCanonicalIndex(field);
1575 if (canonicalIndex < 0 ) {
1576 continue;
1577 }
1578 const dtTypeElem *row = &dtTypes[canonicalIndex];
1579 int32_t typeValue = row->field;
1580 skeletonResult.original[typeValue]=field;
1581 UChar repeatChar = row->patternChar;
1582 int32_t repeatCount = row->minLen > 3 ? 3: row->minLen;
1583 while (repeatCount-- > 0) {
1584 skeletonResult.baseOriginal[typeValue] += repeatChar;
1585 }
1586 int16_t subTypeValue = row->type;
1587 if ( row->type > 0) {
1588 subTypeValue += field.length();
1589 }
1590 skeletonResult.type[typeValue] = subTypeValue;
1591 }
1592 copyFrom(skeletonResult);
1593 }
1594
1595 void
getBasePattern(UnicodeString & result)1596 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
1597 result.remove(); // Reset the result first.
1598 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1599 if (skeleton.baseOriginal[i].length()!=0) {
1600 result += skeleton.baseOriginal[i];
1601 }
1602 }
1603 }
1604
1605 UnicodeString
getPattern()1606 DateTimeMatcher::getPattern() {
1607 UnicodeString result;
1608
1609 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1610 if (skeleton.original[i].length()!=0) {
1611 result += skeleton.original[i];
1612 }
1613 }
1614 return result;
1615 }
1616
1617 int32_t
getDistance(const DateTimeMatcher & other,int32_t includeMask,DistanceInfo & distanceInfo)1618 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) {
1619 int32_t result=0;
1620 distanceInfo.clear();
1621 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1622 int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
1623 int32_t otherType = other.skeleton.type[i];
1624 if (myType==otherType) {
1625 continue;
1626 }
1627 if (myType==0) {// and other is not
1628 result += EXTRA_FIELD;
1629 distanceInfo.addExtra(i);
1630 }
1631 else {
1632 if (otherType==0) {
1633 result += MISSING_FIELD;
1634 distanceInfo.addMissing(i);
1635 }
1636 else {
1637 result += abs(myType - otherType);
1638 }
1639 }
1640
1641 }
1642 return result;
1643 }
1644
1645 void
copyFrom(const PtnSkeleton & newSkeleton)1646 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
1647 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1648 this->skeleton.type[i]=newSkeleton.type[i];
1649 this->skeleton.original[i]=newSkeleton.original[i];
1650 this->skeleton.baseOriginal[i]=newSkeleton.baseOriginal[i];
1651 }
1652 }
1653
1654 void
copyFrom()1655 DateTimeMatcher::copyFrom() {
1656 // same as clear
1657 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1658 this->skeleton.type[i]=0;
1659 this->skeleton.original[i].remove();
1660 this->skeleton.baseOriginal[i].remove();
1661 }
1662 }
1663
1664 UBool
equals(const DateTimeMatcher * other) const1665 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
1666 if (other==NULL) {
1667 return FALSE;
1668 }
1669 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1670 if (this->skeleton.original[i]!=other->skeleton.original[i] ) {
1671 return FALSE;
1672 }
1673 }
1674 return TRUE;
1675 }
1676
1677 int32_t
getFieldMask()1678 DateTimeMatcher::getFieldMask() {
1679 int32_t result=0;
1680
1681 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1682 if (skeleton.type[i]!=0) {
1683 result |= (1<<i);
1684 }
1685 }
1686 return result;
1687 }
1688
1689 PtnSkeleton*
getSkeletonPtr()1690 DateTimeMatcher::getSkeletonPtr() {
1691 return &skeleton;
1692 }
1693
FormatParser()1694 FormatParser::FormatParser () {
1695 status = START;
1696 itemNumber=0;
1697 }
1698
1699
~FormatParser()1700 FormatParser::~FormatParser () {
1701 }
1702
1703
1704 // Find the next token with the starting position and length
1705 // Note: the startPos may
1706 FormatParser::TokenStatus
setTokens(const UnicodeString & pattern,int32_t startPos,int32_t * len)1707 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
1708 int32_t curLoc = startPos;
1709 if ( curLoc >= pattern.length()) {
1710 return DONE;
1711 }
1712 // check the current char is between A-Z or a-z
1713 do {
1714 UChar c=pattern.charAt(curLoc);
1715 if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
1716 curLoc++;
1717 }
1718 else {
1719 startPos = curLoc;
1720 *len=1;
1721 return ADD_TOKEN;
1722 }
1723
1724 if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
1725 break; // not the same token
1726 }
1727 } while(curLoc <= pattern.length());
1728 *len = curLoc-startPos;
1729 return ADD_TOKEN;
1730 }
1731
1732 void
set(const UnicodeString & pattern)1733 FormatParser::set(const UnicodeString& pattern) {
1734 int32_t startPos=0;
1735 TokenStatus result=START;
1736 int32_t len=0;
1737 itemNumber =0;
1738
1739 do {
1740 result = setTokens( pattern, startPos, &len );
1741 if ( result == ADD_TOKEN )
1742 {
1743 items[itemNumber++] = UnicodeString(pattern, startPos, len );
1744 startPos += len;
1745 }
1746 else {
1747 break;
1748 }
1749 } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
1750 }
1751
1752 int32_t
getCanonicalIndex(const UnicodeString & s,UBool strict)1753 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
1754 int32_t len = s.length();
1755 if (len == 0) {
1756 return -1;
1757 }
1758 UChar ch = s.charAt(0);
1759
1760 // Verify that all are the same character.
1761 for (int32_t l = 1; l < len; l++) {
1762 if (ch != s.charAt(l)) {
1763 return -1;
1764 }
1765 }
1766 int32_t i = 0;
1767 int32_t bestRow = -1;
1768 while (dtTypes[i].patternChar != '\0') {
1769 if ( dtTypes[i].patternChar != ch ) {
1770 ++i;
1771 continue;
1772 }
1773 bestRow = i;
1774 if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
1775 return i;
1776 }
1777 if (dtTypes[i+1].minLen <= len) {
1778 ++i;
1779 continue;
1780 }
1781 return i;
1782 }
1783 return strict ? -1 : bestRow;
1784 }
1785
1786 UBool
isQuoteLiteral(const UnicodeString & s) const1787 FormatParser::isQuoteLiteral(const UnicodeString& s) const {
1788 return (UBool)(s.charAt(0)==SINGLE_QUOTE);
1789 }
1790
1791 // This function aussumes the current itemIndex points to the quote literal.
1792 // Please call isQuoteLiteral prior to this function.
1793 void
getQuoteLiteral(UnicodeString & quote,int32_t * itemIndex)1794 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
1795 int32_t i=*itemIndex;
1796
1797 quote.remove();
1798 if (items[i].charAt(0)==SINGLE_QUOTE) {
1799 quote += items[i];
1800 ++i;
1801 }
1802 while ( i < itemNumber ) {
1803 if ( items[i].charAt(0)==SINGLE_QUOTE ) {
1804 if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
1805 // two single quotes e.g. 'o''clock'
1806 quote += items[i++];
1807 quote += items[i++];
1808 continue;
1809 }
1810 else {
1811 quote += items[i];
1812 break;
1813 }
1814 }
1815 else {
1816 quote += items[i];
1817 }
1818 ++i;
1819 }
1820 *itemIndex=i;
1821 }
1822
1823 UBool
isPatternSeparator(UnicodeString & field)1824 FormatParser::isPatternSeparator(UnicodeString& field) {
1825 for (int32_t i=0; i<field.length(); ++i ) {
1826 UChar c= field.charAt(i);
1827 if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
1828 (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
1829 continue;
1830 }
1831 else {
1832 return FALSE;
1833 }
1834 }
1835 return TRUE;
1836 }
1837
1838 void
setTo(DistanceInfo & other)1839 DistanceInfo::setTo(DistanceInfo &other) {
1840 missingFieldMask = other.missingFieldMask;
1841 extraFieldMask= other.extraFieldMask;
1842 }
1843
PatternMapIterator()1844 PatternMapIterator::PatternMapIterator() {
1845 bootIndex = 0;
1846 nodePtr = NULL;
1847 patternMap=NULL;
1848 matcher= new DateTimeMatcher();
1849 }
1850
1851
~PatternMapIterator()1852 PatternMapIterator::~PatternMapIterator() {
1853 delete matcher;
1854 }
1855
1856 void
set(PatternMap & newPatternMap)1857 PatternMapIterator::set(PatternMap& newPatternMap) {
1858 this->patternMap=&newPatternMap;
1859 }
1860
1861 PtnSkeleton*
getSkeleton()1862 PatternMapIterator::getSkeleton() {
1863 if ( nodePtr == NULL ) {
1864 return NULL;
1865 }
1866 else {
1867 return nodePtr->skeleton;
1868 }
1869 }
1870
1871 UBool
hasNext()1872 PatternMapIterator::hasNext() {
1873 int32_t headIndex=bootIndex;
1874 PtnElem *curPtr=nodePtr;
1875
1876 if (patternMap==NULL) {
1877 return FALSE;
1878 }
1879 while ( headIndex < MAX_PATTERN_ENTRIES ) {
1880 if ( curPtr != NULL ) {
1881 if ( curPtr->next != NULL ) {
1882 return TRUE;
1883 }
1884 else {
1885 headIndex++;
1886 curPtr=NULL;
1887 continue;
1888 }
1889 }
1890 else {
1891 if ( patternMap->boot[headIndex] != NULL ) {
1892 return TRUE;
1893 }
1894 else {
1895 headIndex++;
1896 continue;
1897 }
1898 }
1899
1900 }
1901 return FALSE;
1902 }
1903
1904 DateTimeMatcher&
next()1905 PatternMapIterator::next() {
1906 while ( bootIndex < MAX_PATTERN_ENTRIES ) {
1907 if ( nodePtr != NULL ) {
1908 if ( nodePtr->next != NULL ) {
1909 nodePtr = nodePtr->next;
1910 break;
1911 }
1912 else {
1913 bootIndex++;
1914 nodePtr=NULL;
1915 continue;
1916 }
1917 }
1918 else {
1919 if ( patternMap->boot[bootIndex] != NULL ) {
1920 nodePtr = patternMap->boot[bootIndex];
1921 break;
1922 }
1923 else {
1924 bootIndex++;
1925 continue;
1926 }
1927 }
1928 }
1929 if (nodePtr!=NULL) {
1930 matcher->copyFrom(*nodePtr->skeleton);
1931 }
1932 else {
1933 matcher->copyFrom();
1934 }
1935 return *matcher;
1936 }
1937
PtnSkeleton()1938 PtnSkeleton::PtnSkeleton() {
1939 }
1940
1941
PtnSkeleton(const PtnSkeleton & other)1942 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
1943 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1944 this->type[i]=other.type[i];
1945 this->original[i]=other.original[i];
1946 this->baseOriginal[i]=other.baseOriginal[i];
1947 }
1948 }
1949
1950 UBool
equals(const PtnSkeleton & other)1951 PtnSkeleton::equals(const PtnSkeleton& other) {
1952 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1953 if ( (type[i]!= other.type[i]) ||
1954 (original[i]!=other.original[i]) ||
1955 (baseOriginal[i]!=other.baseOriginal[i]) ) {
1956 return FALSE;
1957 }
1958 }
1959 return TRUE;
1960 }
1961
1962 UnicodeString
getSkeleton()1963 PtnSkeleton::getSkeleton() {
1964 UnicodeString result;
1965
1966 for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
1967 if (original[i].length()!=0) {
1968 result += original[i];
1969 }
1970 }
1971 return result;
1972 }
1973
1974 UnicodeString
getBaseSkeleton()1975 PtnSkeleton::getBaseSkeleton() {
1976 UnicodeString result;
1977
1978 for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
1979 if (baseOriginal[i].length()!=0) {
1980 result += baseOriginal[i];
1981 }
1982 }
1983 return result;
1984 }
1985
~PtnSkeleton()1986 PtnSkeleton::~PtnSkeleton() {
1987 }
1988
PtnElem(const UnicodeString & basePat,const UnicodeString & pat)1989 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
1990 basePattern(basePat),
1991 skeleton(NULL),
1992 pattern(pat),
1993 next(NULL)
1994 {
1995 }
1996
~PtnElem()1997 PtnElem::~PtnElem() {
1998
1999 if (next!=NULL) {
2000 delete next;
2001 }
2002 delete skeleton;
2003 }
2004
DTSkeletonEnumeration(PatternMap & patternMap,dtStrEnum type,UErrorCode & status)2005 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum type, UErrorCode& status) {
2006 PtnElem *curElem;
2007 PtnSkeleton *curSkeleton;
2008 UnicodeString s;
2009 int32_t bootIndex;
2010
2011 pos=0;
2012 fSkeletons = new UVector(status);
2013 if (U_FAILURE(status)) {
2014 delete fSkeletons;
2015 return;
2016 }
2017 for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2018 curElem = patternMap.boot[bootIndex];
2019 while (curElem!=NULL) {
2020 switch(type) {
2021 case DT_BASESKELETON:
2022 s=curElem->basePattern;
2023 break;
2024 case DT_PATTERN:
2025 s=curElem->pattern;
2026 break;
2027 case DT_SKELETON:
2028 curSkeleton=curElem->skeleton;
2029 s=curSkeleton->getSkeleton();
2030 break;
2031 }
2032 if ( !isCanonicalItem(s) ) {
2033 fSkeletons->addElement(new UnicodeString(s), status);
2034 if (U_FAILURE(status)) {
2035 delete fSkeletons;
2036 fSkeletons = NULL;
2037 return;
2038 }
2039 }
2040 curElem = curElem->next;
2041 }
2042 }
2043 if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=NULL) ) {
2044 status = U_BUFFER_OVERFLOW_ERROR;
2045 }
2046 }
2047
2048 const UnicodeString*
snext(UErrorCode & status)2049 DTSkeletonEnumeration::snext(UErrorCode& status) {
2050 if (U_SUCCESS(status) && pos < fSkeletons->size()) {
2051 return (const UnicodeString*)fSkeletons->elementAt(pos++);
2052 }
2053 return NULL;
2054 }
2055
2056 void
reset(UErrorCode &)2057 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2058 pos=0;
2059 }
2060
2061 int32_t
count(UErrorCode &) const2062 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2063 return (fSkeletons==NULL) ? 0 : fSkeletons->size();
2064 }
2065
2066 UBool
isCanonicalItem(const UnicodeString & item)2067 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2068 if ( item.length() != 1 ) {
2069 return FALSE;
2070 }
2071 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2072 if (item.charAt(0)==Canonical_Items[i]) {
2073 return TRUE;
2074 }
2075 }
2076 return FALSE;
2077 }
2078
~DTSkeletonEnumeration()2079 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2080 UnicodeString *s;
2081 for (int32_t i=0; i<fSkeletons->size(); ++i) {
2082 if ((s=(UnicodeString *)fSkeletons->elementAt(i))!=NULL) {
2083 delete s;
2084 }
2085 }
2086 delete fSkeletons;
2087 }
2088
DTRedundantEnumeration()2089 DTRedundantEnumeration::DTRedundantEnumeration() {
2090 pos=0;
2091 fPatterns = NULL;
2092 }
2093
2094 void
add(const UnicodeString & pattern,UErrorCode & status)2095 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2096 if (U_FAILURE(status)) return;
2097 if (fPatterns == NULL) {
2098 fPatterns = new UVector(status);
2099 if (U_FAILURE(status)) {
2100 delete fPatterns;
2101 fPatterns = NULL;
2102 return;
2103 }
2104 }
2105 fPatterns->addElement(new UnicodeString(pattern), status);
2106 if (U_FAILURE(status)) {
2107 delete fPatterns;
2108 fPatterns = NULL;
2109 return;
2110 }
2111 }
2112
2113 const UnicodeString*
snext(UErrorCode & status)2114 DTRedundantEnumeration::snext(UErrorCode& status) {
2115 if (U_SUCCESS(status) && pos < fPatterns->size()) {
2116 return (const UnicodeString*)fPatterns->elementAt(pos++);
2117 }
2118 return NULL;
2119 }
2120
2121 void
reset(UErrorCode &)2122 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2123 pos=0;
2124 }
2125
2126 int32_t
count(UErrorCode &) const2127 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2128 return (fPatterns==NULL) ? 0 : fPatterns->size();
2129 }
2130
2131 UBool
isCanonicalItem(const UnicodeString & item)2132 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) {
2133 if ( item.length() != 1 ) {
2134 return FALSE;
2135 }
2136 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2137 if (item.charAt(0)==Canonical_Items[i]) {
2138 return TRUE;
2139 }
2140 }
2141 return FALSE;
2142 }
2143
~DTRedundantEnumeration()2144 DTRedundantEnumeration::~DTRedundantEnumeration() {
2145 UnicodeString *s;
2146 for (int32_t i=0; i<fPatterns->size(); ++i) {
2147 if ((s=(UnicodeString *)fPatterns->elementAt(i))!=NULL) {
2148 delete s;
2149 }
2150 }
2151 delete fPatterns;
2152 }
2153
2154 U_NAMESPACE_END
2155
2156
2157 #endif /* #if !UCONFIG_NO_FORMATTING */
2158
2159 //eof
2160