1 /*
2 *******************************************************************************
3 * Copyright (C) 2007-2013, 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 U_CDECL_BEGIN
70
71 static int32_t U_CALLCONV
ures_a_codepointSort(const void * context,const void * left,const void * right)72 ures_a_codepointSort(const void *context, const void *left, const void *right) {
73 //CompareContext *cmp=(CompareContext *)context;
74 return u_strcmp(((const UResAEntry *)left)->key,
75 ((const UResAEntry *)right)->key);
76 }
77
78 U_CDECL_END
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_U, UDATPG_YEAR_FIELD, DT_SHORT, 1, 3},
148 {CAP_U, UDATPG_YEAR_FIELD, DT_LONG, 4, 0},
149 {CAP_U, UDATPG_YEAR_FIELD, DT_NARROW, 5, 0},
150 {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2},
151 {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0},
152 {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0},
153 {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
154 {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT + DT_DELTA, 3, 0},
155 {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG + DT_DELTA, 4, 0},
156 {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2},
157 {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0},
158 {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0},
159 {CAP_M, UDATPG_MONTH_FIELD, DT_NARROW, 5, 0},
160 {CAP_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
161 {CAP_L, UDATPG_MONTH_FIELD, DT_SHORT - DT_DELTA, 3, 0},
162 {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0},
163 {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0},
164 {LOW_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 1},
165 {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2},
166 {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 0},
167 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3},
168 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0},
169 {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0},
170 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2},
171 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0},
172 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
173 {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0},
174 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical
175 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0},
176 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0},
177 {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0},
178 {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2},
179 {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 3},
180 {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 0},
181 {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20}, // really internal use, so we don't care
182 {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 0},
183 {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour
184 {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2}, // 24 hour
185 {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour
186 {CAP_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // 12 hour
187 {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2},
188 {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2},
189 {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000},
190 {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 1000},
191 {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0},
192 {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
193 {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3},
194 {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0},
195 {CAP_Z, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 3},
196 {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
197 {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 5, 0},
198 {CAP_O, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0},
199 {CAP_O, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
200 {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0},
201 {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 2, 0},
202 {CAP_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0},
203 {CAP_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0},
204 {CAP_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
205 {LOW_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0},
206 {LOW_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0},
207 {LOW_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
208 {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[]
209 };
210
211 static const char* const CLDR_FIELD_APPEND[] = {
212 "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week", "Day", "*", "*", "*",
213 "Hour", "Minute", "Second", "*", "Timezone"
214 };
215
216 static const char* const CLDR_FIELD_NAME[] = {
217 "era", "year", "quarter", "month", "week", "*", "weekday", "*", "*", "day", "dayperiod",
218 "hour", "minute", "second", "*", "zone"
219 };
220
221 static const char* const Resource_Fields[] = {
222 "day", "dayperiod", "era", "hour", "minute", "month", "second", "week",
223 "weekday", "year", "zone", "quarter" };
224
225 // For appendItems
226 static const UChar UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A,
227 0x20, 0x7B, 0x31, 0x7D, 0x2524, 0}; // {0} \u251C{2}: {1}\u2524
228
229 //static const UChar repeatedPatterns[6]={CAP_G, CAP_E, LOW_Z, LOW_V, CAP_Q, 0}; // "GEzvQ"
230
231 static const char DT_DateTimePatternsTag[]="DateTimePatterns";
232 static const char DT_DateTimeCalendarTag[]="calendar";
233 static const char DT_DateTimeGregorianTag[]="gregorian";
234 static const char DT_DateTimeAppendItemsTag[]="appendItems";
235 static const char DT_DateTimeFieldsTag[]="fields";
236 static const char DT_DateTimeAvailableFormatsTag[]="availableFormats";
237 //static const UnicodeString repeatedPattern=UnicodeString(repeatedPatterns);
238
239 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateTimePatternGenerator)
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)240 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)
241 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTRedundantEnumeration)
242
243 DateTimePatternGenerator* U_EXPORT2
244 DateTimePatternGenerator::createInstance(UErrorCode& status) {
245 return createInstance(Locale::getDefault(), status);
246 }
247
248 DateTimePatternGenerator* U_EXPORT2
createInstance(const Locale & locale,UErrorCode & status)249 DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) {
250 DateTimePatternGenerator *result = new DateTimePatternGenerator(locale, 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
261 DateTimePatternGenerator* U_EXPORT2
createEmptyInstance(UErrorCode & status)262 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) {
263 DateTimePatternGenerator *result = new DateTimePatternGenerator(status);
264 if (result == NULL) {
265 status = U_MEMORY_ALLOCATION_ERROR;
266 }
267 if (U_FAILURE(status)) {
268 delete result;
269 result = NULL;
270 }
271 return result;
272 }
273
DateTimePatternGenerator(UErrorCode & status)274 DateTimePatternGenerator::DateTimePatternGenerator(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 }
286
DateTimePatternGenerator(const Locale & locale,UErrorCode & status)287 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status) :
288 skipMatcher(NULL),
289 fAvailableFormatKeyHash(NULL)
290 {
291 fp = new FormatParser();
292 dtMatcher = new DateTimeMatcher();
293 distanceInfo = new DistanceInfo();
294 patternMap = new PatternMap();
295 if (fp == NULL || dtMatcher == NULL || distanceInfo == NULL || patternMap == NULL) {
296 status = U_MEMORY_ALLOCATION_ERROR;
297 }
298 else {
299 initData(locale, status);
300 }
301 }
302
DateTimePatternGenerator(const DateTimePatternGenerator & other)303 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) :
304 UObject(),
305 skipMatcher(NULL),
306 fAvailableFormatKeyHash(NULL)
307 {
308 fp = new FormatParser();
309 dtMatcher = new DateTimeMatcher();
310 distanceInfo = new DistanceInfo();
311 patternMap = new PatternMap();
312 *this=other;
313 }
314
315 DateTimePatternGenerator&
operator =(const DateTimePatternGenerator & other)316 DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) {
317 pLocale = other.pLocale;
318 fDefaultHourFormatChar = other.fDefaultHourFormatChar;
319 *fp = *(other.fp);
320 dtMatcher->copyFrom(other.dtMatcher->skeleton);
321 *distanceInfo = *(other.distanceInfo);
322 dateTimeFormat = other.dateTimeFormat;
323 decimal = other.decimal;
324 // NUL-terminate for the C API.
325 dateTimeFormat.getTerminatedBuffer();
326 decimal.getTerminatedBuffer();
327 delete skipMatcher;
328 if ( other.skipMatcher == NULL ) {
329 skipMatcher = NULL;
330 }
331 else {
332 skipMatcher = new DateTimeMatcher(*other.skipMatcher);
333 }
334 for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) {
335 appendItemFormats[i] = other.appendItemFormats[i];
336 appendItemNames[i] = other.appendItemNames[i];
337 // NUL-terminate for the C API.
338 appendItemFormats[i].getTerminatedBuffer();
339 appendItemNames[i].getTerminatedBuffer();
340 }
341 UErrorCode status = U_ZERO_ERROR;
342 patternMap->copyFrom(*other.patternMap, status);
343 copyHashtable(other.fAvailableFormatKeyHash, status);
344 return *this;
345 }
346
347
348 UBool
operator ==(const DateTimePatternGenerator & other) const349 DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) const {
350 if (this == &other) {
351 return TRUE;
352 }
353 if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) &&
354 (dateTimeFormat==other.dateTimeFormat) && (decimal==other.decimal)) {
355 for ( int32_t i=0 ; i<UDATPG_FIELD_COUNT; ++i ) {
356 if ((appendItemFormats[i] != other.appendItemFormats[i]) ||
357 (appendItemNames[i] != other.appendItemNames[i]) ) {
358 return FALSE;
359 }
360 }
361 return TRUE;
362 }
363 else {
364 return FALSE;
365 }
366 }
367
368 UBool
operator !=(const DateTimePatternGenerator & other) const369 DateTimePatternGenerator::operator!=(const DateTimePatternGenerator& other) const {
370 return !operator==(other);
371 }
372
~DateTimePatternGenerator()373 DateTimePatternGenerator::~DateTimePatternGenerator() {
374 if (fAvailableFormatKeyHash!=NULL) {
375 delete fAvailableFormatKeyHash;
376 }
377
378 if (fp != NULL) delete fp;
379 if (dtMatcher != NULL) delete dtMatcher;
380 if (distanceInfo != NULL) delete distanceInfo;
381 if (patternMap != NULL) delete patternMap;
382 if (skipMatcher != NULL) delete skipMatcher;
383 }
384
385 void
initData(const Locale & locale,UErrorCode & status)386 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status) {
387 //const char *baseLangName = locale.getBaseName(); // unused
388
389 skipMatcher = NULL;
390 fAvailableFormatKeyHash=NULL;
391 addCanonicalItems();
392 addICUPatterns(locale, status);
393 if (U_FAILURE(status)) {
394 return;
395 }
396 addCLDRData(locale, status);
397 setDateTimeFromCalendar(locale, status);
398 setDecimalSymbols(locale, status);
399 } // DateTimePatternGenerator::initData
400
401 UnicodeString
getSkeleton(const UnicodeString & pattern,UErrorCode &)402 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode&
403 /*status*/) {
404 dtMatcher->set(pattern, fp);
405 return dtMatcher->getSkeletonPtr()->getSkeleton();
406 }
407
408 UnicodeString
getBaseSkeleton(const UnicodeString & pattern,UErrorCode &)409 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) {
410 dtMatcher->set(pattern, fp);
411 return dtMatcher->getSkeletonPtr()->getBaseSkeleton();
412 }
413
414 void
addICUPatterns(const Locale & locale,UErrorCode & status)415 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) {
416 UnicodeString dfPattern;
417 UnicodeString conflictingString;
418 DateFormat* df;
419
420 if (U_FAILURE(status)) {
421 return;
422 }
423
424 // Load with ICU patterns
425 for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) {
426 DateFormat::EStyle style = (DateFormat::EStyle)i;
427 df = DateFormat::createDateInstance(style, locale);
428 SimpleDateFormat* sdf;
429 if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
430 addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
431 }
432 // TODO Maybe we should return an error when the date format isn't simple.
433 delete df;
434 if (U_FAILURE(status)) {
435 return;
436 }
437
438 df = DateFormat::createTimeInstance(style, locale);
439 if (df != NULL && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != NULL) {
440 addPattern(sdf->toPattern(dfPattern), FALSE, conflictingString, status);
441 // HACK for hh:ss
442 if ( i==DateFormat::kMedium ) {
443 hackPattern = dfPattern;
444 }
445 }
446 // TODO Maybe we should return an error when the date format isn't simple.
447 delete df;
448 if (U_FAILURE(status)) {
449 return;
450 }
451 }
452 }
453
454 void
hackTimes(const UnicodeString & hackPattern,UErrorCode & status)455 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status) {
456 UnicodeString conflictingString;
457
458 fp->set(hackPattern);
459 UnicodeString mmss;
460 UBool gotMm=FALSE;
461 for (int32_t i=0; i<fp->itemNumber; ++i) {
462 UnicodeString field = fp->items[i];
463 if ( fp->isQuoteLiteral(field) ) {
464 if ( gotMm ) {
465 UnicodeString quoteLiteral;
466 fp->getQuoteLiteral(quoteLiteral, &i);
467 mmss += quoteLiteral;
468 }
469 }
470 else {
471 if (fp->isPatternSeparator(field) && gotMm) {
472 mmss+=field;
473 }
474 else {
475 UChar ch=field.charAt(0);
476 if (ch==LOW_M) {
477 gotMm=TRUE;
478 mmss+=field;
479 }
480 else {
481 if (ch==LOW_S) {
482 if (!gotMm) {
483 break;
484 }
485 mmss+= field;
486 addPattern(mmss, FALSE, conflictingString, status);
487 break;
488 }
489 else {
490 if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) {
491 break;
492 }
493 }
494 }
495 }
496 }
497 }
498 }
499
500 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY)
501
502 static const UChar hourFormatChars[] = { CAP_H, LOW_H, CAP_K, LOW_K, 0 }; // HhKk, the hour format characters
503
504 void
addCLDRData(const Locale & locale,UErrorCode & err)505 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& err) {
506 UResourceBundle *rb, *calTypeBundle, *calBundle;
507 UResourceBundle *patBundle, *fieldBundle, *fBundle;
508 UnicodeString rbPattern, value, field;
509 UnicodeString conflictingPattern;
510 const char *key=NULL;
511 int32_t i;
512
513 UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, LENGTHOF(UDATPG_ItemFormat)-1); // Read-only alias.
514
515 err = U_ZERO_ERROR;
516
517 fDefaultHourFormatChar = 0;
518 for (i=0; i<UDATPG_FIELD_COUNT; ++i ) {
519 appendItemNames[i]=CAP_F;
520 if (i<10) {
521 appendItemNames[i]+=(UChar)(i+0x30);
522 }
523 else {
524 appendItemNames[i]+=(UChar)0x31;
525 appendItemNames[i]+=(UChar)(i-10 + 0x30);
526 }
527 // NUL-terminate for the C API.
528 appendItemNames[i].getTerminatedBuffer();
529 }
530
531 rb = ures_open(NULL, locale.getName(), &err);
532 if (rb == NULL || U_FAILURE(err)) {
533 return;
534 }
535 const char *curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
536 const char * calendarTypeToUse = DT_DateTimeGregorianTag; // initial default
537 char calendarType[ULOC_KEYWORDS_CAPACITY]; // to be filled in with the type to use, if all goes well
538 if ( U_SUCCESS(err) ) {
539 char localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
540 // obtain a locale that always has the calendar key value that should be used
541 (void)ures_getFunctionalEquivalent(localeWithCalendarKey, ULOC_LOCALE_IDENTIFIER_CAPACITY, NULL,
542 "calendar", "calendar", locale.getName(), NULL, FALSE, &err);
543 localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination
544 // now get the calendar key value from that locale
545 int32_t calendarTypeLen = uloc_getKeywordValue(localeWithCalendarKey, "calendar", calendarType, ULOC_KEYWORDS_CAPACITY, &err);
546 if (U_SUCCESS(err) && calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
547 calendarTypeToUse = calendarType;
548 }
549 err = U_ZERO_ERROR;
550 }
551 calBundle = ures_getByKeyWithFallback(rb, DT_DateTimeCalendarTag, NULL, &err);
552 calTypeBundle = ures_getByKeyWithFallback(calBundle, calendarTypeToUse, NULL, &err);
553
554 key=NULL;
555 int32_t dtCount=0;
556 patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimePatternsTag, NULL, &err);
557 while (U_SUCCESS(err)) {
558 rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
559 dtCount++;
560 if (rbPattern.length()==0 ) {
561 break; // no more pattern
562 }
563 else {
564 if (dtCount==9) {
565 setDateTimeFormat(rbPattern);
566 } else if (dtCount==4) { // short time format
567 // set fDefaultHourFormatChar to the hour format character from this pattern
568 int32_t tfIdx, tfLen = rbPattern.length();
569 UBool ignoreChars = FALSE;
570 for (tfIdx = 0; tfIdx < tfLen; tfIdx++) {
571 UChar tfChar = rbPattern.charAt(tfIdx);
572 if ( tfChar == SINGLE_QUOTE ) {
573 ignoreChars = !ignoreChars; // toggle (handle quoted literals & '' for single quote)
574 } else if ( !ignoreChars && u_strchr(hourFormatChars, tfChar) != NULL ) {
575 fDefaultHourFormatChar = tfChar;
576 break;
577 }
578 }
579 }
580 }
581 }
582 ures_close(patBundle);
583
584 err = U_ZERO_ERROR;
585 patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAppendItemsTag, NULL, &err);
586 key=NULL;
587 UnicodeString itemKey;
588 while (U_SUCCESS(err)) {
589 rbPattern = ures_getNextUnicodeString(patBundle, &key, &err);
590 if (rbPattern.length()==0 ) {
591 break; // no more pattern
592 }
593 else {
594 setAppendItemFormat(getAppendFormatNumber(key), rbPattern);
595 }
596 }
597 ures_close(patBundle);
598
599 key=NULL;
600 err = U_ZERO_ERROR;
601 fBundle = ures_getByKeyWithFallback(rb, DT_DateTimeFieldsTag, NULL, &err);
602 for (i=0; i<MAX_RESOURCE_FIELD; ++i) {
603 err = U_ZERO_ERROR;
604 patBundle = ures_getByKeyWithFallback(fBundle, Resource_Fields[i], NULL, &err);
605 fieldBundle = ures_getByKeyWithFallback(patBundle, "dn", NULL, &err);
606 rbPattern = ures_getNextUnicodeString(fieldBundle, &key, &err);
607 ures_close(fieldBundle);
608 ures_close(patBundle);
609 if (rbPattern.length()==0 ) {
610 continue;
611 }
612 else {
613 setAppendItemName(getAppendNameNumber(Resource_Fields[i]), rbPattern);
614 }
615 }
616 ures_close(fBundle);
617
618 // add available formats
619 UBool firstTimeThrough = TRUE;
620 err = U_ZERO_ERROR;
621 initHashtable(err);
622 UBool override = TRUE;
623 while (TRUE) {
624 // At the start of the loop:
625 // - rb is the open resource bundle for the current locale being processed,
626 // whose actual name is in curLocaleName.
627 // - if U_SUCCESS(err), then calBundle and calTypeBundle are open;
628 // process contents of calTypeBundle, then close calBundle and calTypeBundle.
629 if (U_SUCCESS(err)) {
630 // process contents of calTypeBundle
631 patBundle = ures_getByKeyWithFallback(calTypeBundle, DT_DateTimeAvailableFormatsTag, NULL, &err);
632 if (U_SUCCESS(err)) {
633 int32_t numberKeys = ures_getSize(patBundle);
634 int32_t len;
635 const UChar *retPattern;
636 key=NULL;
637 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
638 UResourceBundleAIterator aiter;
639 ures_a_open(&aiter, patBundle, &err);
640 #endif
641 for(i=0; i<numberKeys; ++i) {
642 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
643 retPattern=ures_a_getNextString(&aiter, &len, &key, &err);
644 #else
645 retPattern=ures_getNextString(patBundle, &len, &key, &err);
646 #endif
647 UnicodeString format=UnicodeString(retPattern);
648 UnicodeString retKey=UnicodeString(key, -1, US_INV);
649 if ( firstTimeThrough || !isAvailableFormatSet(retKey) ) {
650 setAvailableFormat(retKey, err);
651 // Add pattern with its associated skeleton. Override any duplicate derived from std patterns,
652 // but not a previous availableFormats entry:
653 addPatternWithSkeleton(format, &retKey, override, conflictingPattern, err);
654 }
655 }
656 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
657 ures_a_close(&aiter);
658 #endif
659 ures_close(patBundle);
660 }
661 firstTimeThrough = FALSE;
662 // close calBundle and calTypeBundle
663 ures_close(calTypeBundle);
664 ures_close(calBundle);
665 }
666 if (uprv_strcmp(curLocaleName,"root")==0 || uprv_strlen(curLocaleName)==0) {
667 // we just finished handling root, nothing more to check
668 ures_close(rb);
669 break;
670 }
671 // Find the name of the appropriate parent locale (from %%Parent if present, else
672 // uloc_getParent on the actual locale name)
673 // (It would be nice to have a ures function that did this...)
674 err = U_ZERO_ERROR;
675 char parentLocale[ULOC_FULLNAME_CAPACITY];
676 int32_t locNameLen;
677 const UChar * parentUName = ures_getStringByKey(rb, "%%Parent", &locNameLen, &err);
678 if (U_SUCCESS(err) && err != U_USING_FALLBACK_WARNING && locNameLen < ULOC_FULLNAME_CAPACITY) {
679 u_UCharsToChars(parentUName, parentLocale, locNameLen + 1);
680 } else {
681 err = U_ZERO_ERROR;
682 uloc_getParent(curLocaleName, parentLocale, ULOC_FULLNAME_CAPACITY, &err);
683 if (U_FAILURE(err) || err == U_STRING_NOT_TERMINATED_WARNING) {
684 // just fallback to root, since we are not already there
685 parentLocale[0] = 0;
686 err = U_ZERO_ERROR;
687 }
688 }
689 // Close current locale bundle
690 ures_close(rb);
691 // And open its parent, which becomes the new current locale being processed
692 rb = ures_open(NULL, parentLocale, &err);
693 if ( U_FAILURE(err) ) {
694 err = U_ZERO_ERROR;
695 break;
696 }
697 // Get the name of the parent / new current locale
698 curLocaleName=ures_getLocaleByType(rb, ULOC_ACTUAL_LOCALE, &err);
699 if ( U_FAILURE(err) ) {
700 curLocaleName = parentLocale;
701 err = U_ZERO_ERROR;
702 }
703 if (uprv_strcmp(curLocaleName,"root")==0 || uprv_strlen(curLocaleName)==0) {
704 override = FALSE;
705 }
706 // Open calBundle and calTypeBundle
707 calBundle = ures_getByKeyWithFallback(rb, DT_DateTimeCalendarTag, NULL, &err);
708 if (U_SUCCESS(err)) {
709 calTypeBundle = ures_getByKeyWithFallback(calBundle, calendarTypeToUse, NULL, &err);
710 if ( U_FAILURE(err) ) {
711 ures_close(calBundle);
712 }
713 }
714 // Go to the top of the loop to process contents of calTypeBundle
715 }
716
717 if (hackPattern.length()>0) {
718 hackTimes(hackPattern, err);
719 }
720 }
721
722 void
initHashtable(UErrorCode & err)723 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
724 if (fAvailableFormatKeyHash!=NULL) {
725 return;
726 }
727 if ((fAvailableFormatKeyHash = new Hashtable(FALSE, err))==NULL) {
728 err=U_MEMORY_ALLOCATION_ERROR;
729 return;
730 }
731 }
732
733
734 void
setAppendItemFormat(UDateTimePatternField field,const UnicodeString & value)735 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
736 appendItemFormats[field] = value;
737 // NUL-terminate for the C API.
738 appendItemFormats[field].getTerminatedBuffer();
739 }
740
741 const UnicodeString&
getAppendItemFormat(UDateTimePatternField field) const742 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
743 return appendItemFormats[field];
744 }
745
746 void
setAppendItemName(UDateTimePatternField field,const UnicodeString & value)747 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
748 appendItemNames[field] = value;
749 // NUL-terminate for the C API.
750 appendItemNames[field].getTerminatedBuffer();
751 }
752
753 const UnicodeString&
getAppendItemName(UDateTimePatternField field) const754 DateTimePatternGenerator:: getAppendItemName(UDateTimePatternField field) const {
755 return appendItemNames[field];
756 }
757
758 void
getAppendName(UDateTimePatternField field,UnicodeString & value)759 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
760 value = SINGLE_QUOTE;
761 value += appendItemNames[field];
762 value += SINGLE_QUOTE;
763 }
764
765 UnicodeString
getBestPattern(const UnicodeString & patternForm,UErrorCode & status)766 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
767 return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
768 }
769
770 UnicodeString
getBestPattern(const UnicodeString & patternForm,UDateTimePatternMatchOptions options,UErrorCode & status)771 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
772 const UnicodeString *bestPattern=NULL;
773 UnicodeString dtFormat;
774 UnicodeString resultPattern;
775 int32_t flags = kDTPGNoFlags;
776
777 int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
778 int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
779
780 // Replace hour metacharacters 'j' and 'J', set flags as necessary
781 UnicodeString patternFormCopy = UnicodeString(patternForm);
782 int32_t patPos, patLen = patternFormCopy.length();
783 UBool inQuoted = FALSE;
784 for (patPos = 0; patPos < patLen; patPos++) {
785 UChar patChr = patternFormCopy.charAt(patPos);
786 if (patChr == SINGLE_QUOTE) {
787 inQuoted = !inQuoted;
788 } else if (!inQuoted) {
789 if (patChr == LOW_J) {
790 patternFormCopy.setCharAt(patPos, fDefaultHourFormatChar);
791 } else if (patChr == CAP_J) {
792 // Get pattern for skeleton with H, then replace H or k
793 // with fDefaultHourFormatChar (if different)
794 patternFormCopy.setCharAt(patPos, CAP_H);
795 flags |= kDTPGSkeletonUsesCapJ;
796 }
797 }
798 }
799
800 resultPattern.remove();
801 dtMatcher->set(patternFormCopy, fp);
802 const PtnSkeleton* specifiedSkeleton=NULL;
803 bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, &specifiedSkeleton);
804 if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
805 resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, flags, options);
806
807 return resultPattern;
808 }
809 int32_t neededFields = dtMatcher->getFieldMask();
810 UnicodeString datePattern=getBestAppending(neededFields & dateMask, flags, options);
811 UnicodeString timePattern=getBestAppending(neededFields & timeMask, flags, options);
812 if (datePattern.length()==0) {
813 if (timePattern.length()==0) {
814 resultPattern.remove();
815 }
816 else {
817 return timePattern;
818 }
819 }
820 if (timePattern.length()==0) {
821 return datePattern;
822 }
823 resultPattern.remove();
824 status = U_ZERO_ERROR;
825 dtFormat=getDateTimeFormat();
826 Formattable dateTimeObject[] = { timePattern, datePattern };
827 resultPattern = MessageFormat::format(dtFormat, dateTimeObject, 2, resultPattern, status );
828 return resultPattern;
829 }
830
831 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UErrorCode & status)832 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
833 const UnicodeString& skeleton,
834 UErrorCode& status) {
835 return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
836 }
837
838 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UDateTimePatternMatchOptions options,UErrorCode &)839 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
840 const UnicodeString& skeleton,
841 UDateTimePatternMatchOptions options,
842 UErrorCode& /*status*/) {
843 dtMatcher->set(skeleton, fp);
844 UnicodeString result = adjustFieldTypes(pattern, NULL, kDTPGNoFlags, options);
845 return result;
846 }
847
848 void
setDecimal(const UnicodeString & newDecimal)849 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
850 this->decimal = newDecimal;
851 // NUL-terminate for the C API.
852 this->decimal.getTerminatedBuffer();
853 }
854
855 const UnicodeString&
getDecimal() const856 DateTimePatternGenerator::getDecimal() const {
857 return decimal;
858 }
859
860 void
addCanonicalItems()861 DateTimePatternGenerator::addCanonicalItems() {
862 UnicodeString conflictingPattern;
863 UErrorCode status = U_ZERO_ERROR;
864
865 for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
866 addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
867 }
868 }
869
870 void
setDateTimeFormat(const UnicodeString & dtFormat)871 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
872 dateTimeFormat = dtFormat;
873 // NUL-terminate for the C API.
874 dateTimeFormat.getTerminatedBuffer();
875 }
876
877 const UnicodeString&
getDateTimeFormat() const878 DateTimePatternGenerator::getDateTimeFormat() const {
879 return dateTimeFormat;
880 }
881
882 void
setDateTimeFromCalendar(const Locale & locale,UErrorCode & status)883 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
884 const UChar *resStr;
885 int32_t resStrLen = 0;
886
887 Calendar* fCalendar = Calendar::createInstance(locale, status);
888 CalendarData calData(locale, fCalendar?fCalendar->getType():NULL, status);
889 UResourceBundle *dateTimePatterns = calData.getByKey(DT_DateTimePatternsTag, status);
890 if (U_FAILURE(status)) return;
891
892 if (ures_getSize(dateTimePatterns) <= DateFormat::kDateTime)
893 {
894 status = U_INVALID_FORMAT_ERROR;
895 return;
896 }
897 resStr = ures_getStringByIndex(dateTimePatterns, (int32_t)DateFormat::kDateTime, &resStrLen, &status);
898 setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
899
900 delete fCalendar;
901 }
902
903 void
setDecimalSymbols(const Locale & locale,UErrorCode & status)904 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
905 DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
906 if(U_SUCCESS(status)) {
907 decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
908 // NUL-terminate for the C API.
909 decimal.getTerminatedBuffer();
910 }
911 }
912
913 UDateTimePatternConflict
addPattern(const UnicodeString & pattern,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)914 DateTimePatternGenerator::addPattern(
915 const UnicodeString& pattern,
916 UBool override,
917 UnicodeString &conflictingPattern,
918 UErrorCode& status)
919 {
920 return addPatternWithSkeleton(pattern, NULL, override, conflictingPattern, status);
921 }
922
923 // For DateTimePatternGenerator::addPatternWithSkeleton -
924 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
925 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
926 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
927 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
928 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
929 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
930 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
931 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
932 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
933 UDateTimePatternConflict
addPatternWithSkeleton(const UnicodeString & pattern,const UnicodeString * skeletonToUse,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)934 DateTimePatternGenerator::addPatternWithSkeleton(
935 const UnicodeString& pattern,
936 const UnicodeString* skeletonToUse,
937 UBool override,
938 UnicodeString& conflictingPattern,
939 UErrorCode& status)
940 {
941
942 UnicodeString basePattern;
943 PtnSkeleton skeleton;
944 UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
945
946 DateTimeMatcher matcher;
947 if ( skeletonToUse == NULL ) {
948 matcher.set(pattern, fp, skeleton);
949 matcher.getBasePattern(basePattern);
950 } else {
951 matcher.set(*skeletonToUse, fp, skeleton); // no longer trims skeleton fields to max len 3, per #7930
952 matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
953 }
954 // We only care about base conflicts - and replacing the pattern associated with a base - if:
955 // 1. the conflicting previous base pattern did *not* have an explicit skeleton; in that case the previous
956 // base + pattern combination was derived from either (a) a canonical item, (b) a standard format, or
957 // (c) a pattern specified programmatically with a previous call to addPattern (which would only happen
958 // if we are getting here from a subsequent call to addPattern).
959 // 2. a skeleton is specified for the current pattern, but override=false; in that case we are checking
960 // availableFormats items from root, which should not override any previous entry with the same base.
961 UBool entryHadSpecifiedSkeleton;
962 const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
963 if (duplicatePattern != NULL && (!entryHadSpecifiedSkeleton || (skeletonToUse != NULL && !override))) {
964 conflictingStatus = UDATPG_BASE_CONFLICT;
965 conflictingPattern = *duplicatePattern;
966 if (!override) {
967 return conflictingStatus;
968 }
969 }
970 // The only time we get here with override=true and skeletonToUse!=null is when adding availableFormats
971 // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with
972 // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for
973 // the previously-specified conflicting item.
974 const PtnSkeleton* entrySpecifiedSkeleton = NULL;
975 duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
976 if (duplicatePattern != NULL ) {
977 conflictingStatus = UDATPG_CONFLICT;
978 conflictingPattern = *duplicatePattern;
979 if (!override || (skeletonToUse != NULL && entrySpecifiedSkeleton != NULL)) {
980 return conflictingStatus;
981 }
982 }
983 patternMap->add(basePattern, skeleton, pattern, skeletonToUse != NULL, status);
984 if(U_FAILURE(status)) {
985 return conflictingStatus;
986 }
987
988 return UDATPG_NO_CONFLICT;
989 }
990
991
992 UDateTimePatternField
getAppendFormatNumber(const char * field) const993 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
994 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
995 if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
996 return (UDateTimePatternField)i;
997 }
998 }
999 return UDATPG_FIELD_COUNT;
1000 }
1001
1002 UDateTimePatternField
getAppendNameNumber(const char * field) const1003 DateTimePatternGenerator::getAppendNameNumber(const char* field) const {
1004 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1005 if (uprv_strcmp(CLDR_FIELD_NAME[i],field)==0) {
1006 return (UDateTimePatternField)i;
1007 }
1008 }
1009 return UDATPG_FIELD_COUNT;
1010 }
1011
1012 const UnicodeString*
getBestRaw(DateTimeMatcher & source,int32_t includeMask,DistanceInfo * missingFields,const PtnSkeleton ** specifiedSkeletonPtr)1013 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
1014 int32_t includeMask,
1015 DistanceInfo* missingFields,
1016 const PtnSkeleton** specifiedSkeletonPtr) {
1017 int32_t bestDistance = 0x7fffffff;
1018 DistanceInfo tempInfo;
1019 const UnicodeString *bestPattern=NULL;
1020 const PtnSkeleton* specifiedSkeleton=NULL;
1021
1022 PatternMapIterator it;
1023 for (it.set(*patternMap); it.hasNext(); ) {
1024 DateTimeMatcher trial = it.next();
1025 if (trial.equals(skipMatcher)) {
1026 continue;
1027 }
1028 int32_t distance=source.getDistance(trial, includeMask, tempInfo);
1029 if (distance<bestDistance) {
1030 bestDistance=distance;
1031 bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
1032 missingFields->setTo(tempInfo);
1033 if (distance==0) {
1034 break;
1035 }
1036 }
1037 }
1038
1039 // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
1040 // then return it too. This generally happens when the caller needs to pass that skeleton
1041 // through to adjustFieldTypes so the latter can do a better job.
1042 if (bestPattern && specifiedSkeletonPtr) {
1043 *specifiedSkeletonPtr = specifiedSkeleton;
1044 }
1045 return bestPattern;
1046 }
1047
1048 UnicodeString
adjustFieldTypes(const UnicodeString & pattern,const PtnSkeleton * specifiedSkeleton,int32_t flags,UDateTimePatternMatchOptions options)1049 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
1050 const PtnSkeleton* specifiedSkeleton,
1051 int32_t flags,
1052 UDateTimePatternMatchOptions options) {
1053 UnicodeString newPattern;
1054 fp->set(pattern);
1055 for (int32_t i=0; i < fp->itemNumber; i++) {
1056 UnicodeString field = fp->items[i];
1057 if ( fp->isQuoteLiteral(field) ) {
1058
1059 UnicodeString quoteLiteral;
1060 fp->getQuoteLiteral(quoteLiteral, &i);
1061 newPattern += quoteLiteral;
1062 }
1063 else {
1064 if (fp->isPatternSeparator(field)) {
1065 newPattern+=field;
1066 continue;
1067 }
1068 int32_t canonicalIndex = fp->getCanonicalIndex(field);
1069 if (canonicalIndex < 0) {
1070 newPattern+=field;
1071 continue; // don't adjust
1072 }
1073 const dtTypeElem *row = &dtTypes[canonicalIndex];
1074 int32_t typeValue = row->field;
1075 if ((flags & kDTPGFixFractionalSeconds) != 0 && typeValue == UDATPG_SECOND_FIELD) {
1076 UnicodeString newField=dtMatcher->skeleton.original[UDATPG_FRACTIONAL_SECOND_FIELD];
1077 field = field + decimal + newField;
1078 } else if (dtMatcher->skeleton.type[typeValue]!=0) {
1079 // Here:
1080 // - "reqField" is the field from the originally requested skeleton, with length
1081 // "reqFieldLen".
1082 // - "field" is the field from the found pattern.
1083 //
1084 // The adjusted field should consist of characters from the originally requested
1085 // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1086 // UDATPG_WEEKDAY_FIELD or UDATPG_YEAR_FIELD, in which case it should consist
1087 // of characters from the found pattern.
1088 //
1089 // The length of the adjusted field (adjFieldLen) should match that in the originally
1090 // requested skeleton, except that in the following cases the length of the adjusted field
1091 // should match that in the found pattern (i.e. the length of this pattern field should
1092 // not be adjusted):
1093 // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1094 // not set (ticket #7180). Note, we may want to implement a similar change for other
1095 // numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1096 // field length, but options bits can be used to override this.
1097 // 2. There is a specified skeleton for the found pattern and one of the following is true:
1098 // a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1099 // b) The pattern field is numeric and the skeleton field is not, or vice versa.
1100
1101 UnicodeString reqField = dtMatcher->skeleton.original[typeValue];
1102 int32_t reqFieldLen = reqField.length();
1103 if (reqField.charAt(0) == CAP_E && reqFieldLen < 3)
1104 reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1105 int32_t adjFieldLen = reqFieldLen;
1106 if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1107 (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1108 (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1109 adjFieldLen = field.length();
1110 } else if (specifiedSkeleton) {
1111 UnicodeString skelField = specifiedSkeleton->original[typeValue];
1112 int32_t skelFieldLen = skelField.length();
1113 UBool patFieldIsNumeric = (row->type > 0);
1114 UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1115 if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1116 // don't adjust the field length in the found pattern
1117 adjFieldLen = field.length();
1118 }
1119 }
1120 UChar c = (typeValue!= UDATPG_HOUR_FIELD && typeValue!= UDATPG_MONTH_FIELD &&
1121 typeValue!= UDATPG_WEEKDAY_FIELD && (typeValue!= UDATPG_YEAR_FIELD || reqField.charAt(0)==CAP_Y))?
1122 reqField.charAt(0): field.charAt(0);
1123 if (typeValue == UDATPG_HOUR_FIELD && (flags & kDTPGSkeletonUsesCapJ) != 0) {
1124 c = fDefaultHourFormatChar;
1125 }
1126 field.remove();
1127 for (int32_t i=adjFieldLen; i>0; --i) {
1128 field+=c;
1129 }
1130 }
1131 newPattern+=field;
1132 }
1133 }
1134 return newPattern;
1135 }
1136
1137 UnicodeString
getBestAppending(int32_t missingFields,int32_t flags,UDateTimePatternMatchOptions options)1138 DateTimePatternGenerator::getBestAppending(int32_t missingFields, int32_t flags, UDateTimePatternMatchOptions options) {
1139 UnicodeString resultPattern, tempPattern;
1140 UErrorCode err=U_ZERO_ERROR;
1141 int32_t lastMissingFieldMask=0;
1142 if (missingFields!=0) {
1143 resultPattern=UnicodeString();
1144 const PtnSkeleton* specifiedSkeleton=NULL;
1145 tempPattern = *getBestRaw(*dtMatcher, missingFields, distanceInfo, &specifiedSkeleton);
1146 resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1147 if ( distanceInfo->missingFieldMask==0 ) {
1148 return resultPattern;
1149 }
1150 while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1151 if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1152 break; // cannot find the proper missing field
1153 }
1154 if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1155 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1156 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, flags | kDTPGFixFractionalSeconds, options);
1157 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1158 continue;
1159 }
1160 int32_t startingMask = distanceInfo->missingFieldMask;
1161 tempPattern = *getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, &specifiedSkeleton);
1162 tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1163 int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1164 int32_t topField=getTopBitNumber(foundMask);
1165 UnicodeString appendName;
1166 getAppendName((UDateTimePatternField)topField, appendName);
1167 const Formattable formatPattern[] = {
1168 resultPattern,
1169 tempPattern,
1170 appendName
1171 };
1172 UnicodeString emptyStr;
1173 resultPattern = MessageFormat::format(appendItemFormats[topField], formatPattern, 3, emptyStr, err);
1174 lastMissingFieldMask = distanceInfo->missingFieldMask;
1175 }
1176 }
1177 return resultPattern;
1178 }
1179
1180 int32_t
getTopBitNumber(int32_t foundMask)1181 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) {
1182 if ( foundMask==0 ) {
1183 return 0;
1184 }
1185 int32_t i=0;
1186 while (foundMask!=0) {
1187 foundMask >>=1;
1188 ++i;
1189 }
1190 if (i-1 >UDATPG_ZONE_FIELD) {
1191 return UDATPG_ZONE_FIELD;
1192 }
1193 else
1194 return i-1;
1195 }
1196
1197 void
setAvailableFormat(const UnicodeString & key,UErrorCode & err)1198 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1199 {
1200 fAvailableFormatKeyHash->puti(key, 1, err);
1201 }
1202
1203 UBool
isAvailableFormatSet(const UnicodeString & key) const1204 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1205 return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1206 }
1207
1208 void
copyHashtable(Hashtable * other,UErrorCode & status)1209 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1210
1211 if (other == NULL) {
1212 return;
1213 }
1214 if (fAvailableFormatKeyHash != NULL) {
1215 delete fAvailableFormatKeyHash;
1216 fAvailableFormatKeyHash = NULL;
1217 }
1218 initHashtable(status);
1219 if(U_FAILURE(status)){
1220 return;
1221 }
1222 int32_t pos = -1;
1223 const UHashElement* elem = NULL;
1224 // walk through the hash table and create a deep clone
1225 while((elem = other->nextElement(pos))!= NULL){
1226 const UHashTok otherKeyTok = elem->key;
1227 UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1228 fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1229 if(U_FAILURE(status)){
1230 return;
1231 }
1232 }
1233 }
1234
1235 StringEnumeration*
getSkeletons(UErrorCode & status) const1236 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1237 StringEnumeration* skeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status);
1238 return skeletonEnumerator;
1239 }
1240
1241 const UnicodeString&
getPatternForSkeleton(const UnicodeString & skeleton) const1242 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1243 PtnElem *curElem;
1244
1245 if (skeleton.length() ==0) {
1246 return emptyString;
1247 }
1248 curElem = patternMap->getHeader(skeleton.charAt(0));
1249 while ( curElem != NULL ) {
1250 if ( curElem->skeleton->getSkeleton()==skeleton ) {
1251 return curElem->pattern;
1252 }
1253 curElem=curElem->next;
1254 }
1255 return emptyString;
1256 }
1257
1258 StringEnumeration*
getBaseSkeletons(UErrorCode & status) const1259 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1260 StringEnumeration* baseSkeletonEnumerator = new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status);
1261 return baseSkeletonEnumerator;
1262 }
1263
1264 StringEnumeration*
getRedundants(UErrorCode & status)1265 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1266 StringEnumeration* output = new DTRedundantEnumeration();
1267 const UnicodeString *pattern;
1268 PatternMapIterator it;
1269 for (it.set(*patternMap); it.hasNext(); ) {
1270 DateTimeMatcher current = it.next();
1271 pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1272 if ( isCanonicalItem(*pattern) ) {
1273 continue;
1274 }
1275 if ( skipMatcher == NULL ) {
1276 skipMatcher = new DateTimeMatcher(current);
1277 }
1278 else {
1279 *skipMatcher = current;
1280 }
1281 UnicodeString trial = getBestPattern(current.getPattern(), status);
1282 if (trial == *pattern) {
1283 ((DTRedundantEnumeration *)output)->add(*pattern, status);
1284 }
1285 if (current.equals(skipMatcher)) {
1286 continue;
1287 }
1288 }
1289 return output;
1290 }
1291
1292 UBool
isCanonicalItem(const UnicodeString & item) const1293 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1294 if ( item.length() != 1 ) {
1295 return FALSE;
1296 }
1297 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1298 if (item.charAt(0)==Canonical_Items[i]) {
1299 return TRUE;
1300 }
1301 }
1302 return FALSE;
1303 }
1304
1305
1306 DateTimePatternGenerator*
clone() const1307 DateTimePatternGenerator::clone() const {
1308 return new DateTimePatternGenerator(*this);
1309 }
1310
PatternMap()1311 PatternMap::PatternMap() {
1312 for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1313 boot[i]=NULL;
1314 }
1315 isDupAllowed = TRUE;
1316 }
1317
1318 void
copyFrom(const PatternMap & other,UErrorCode & status)1319 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1320 this->isDupAllowed = other.isDupAllowed;
1321 for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1322 PtnElem *curElem, *otherElem, *prevElem=NULL;
1323 otherElem = other.boot[bootIndex];
1324 while (otherElem!=NULL) {
1325 if ((curElem = new PtnElem(otherElem->basePattern, otherElem->pattern))==NULL) {
1326 // out of memory
1327 status = U_MEMORY_ALLOCATION_ERROR;
1328 return;
1329 }
1330 if ( this->boot[bootIndex]== NULL ) {
1331 this->boot[bootIndex] = curElem;
1332 }
1333 if ((curElem->skeleton=new PtnSkeleton(*(otherElem->skeleton))) == NULL ) {
1334 // out of memory
1335 status = U_MEMORY_ALLOCATION_ERROR;
1336 return;
1337 }
1338
1339 if (prevElem!=NULL) {
1340 prevElem->next=curElem;
1341 }
1342 curElem->next=NULL;
1343 prevElem = curElem;
1344 otherElem = otherElem->next;
1345 }
1346
1347 }
1348 }
1349
1350 PtnElem*
getHeader(UChar baseChar)1351 PatternMap::getHeader(UChar baseChar) {
1352 PtnElem* curElem;
1353
1354 if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1355 curElem = boot[baseChar-CAP_A];
1356 }
1357 else {
1358 if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1359 curElem = boot[26+baseChar-LOW_A];
1360 }
1361 else {
1362 return NULL;
1363 }
1364 }
1365 return curElem;
1366 }
1367
~PatternMap()1368 PatternMap::~PatternMap() {
1369 for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1370 if (boot[i]!=NULL ) {
1371 delete boot[i];
1372 boot[i]=NULL;
1373 }
1374 }
1375 } // PatternMap destructor
1376
1377 void
add(const UnicodeString & basePattern,const PtnSkeleton & skeleton,const UnicodeString & value,UBool skeletonWasSpecified,UErrorCode & status)1378 PatternMap::add(const UnicodeString& basePattern,
1379 const PtnSkeleton& skeleton,
1380 const UnicodeString& value,// mapped pattern value
1381 UBool skeletonWasSpecified,
1382 UErrorCode &status) {
1383 UChar baseChar = basePattern.charAt(0);
1384 PtnElem *curElem, *baseElem;
1385 status = U_ZERO_ERROR;
1386
1387 // the baseChar must be A-Z or a-z
1388 if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1389 baseElem = boot[baseChar-CAP_A];
1390 }
1391 else {
1392 if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1393 baseElem = boot[26+baseChar-LOW_A];
1394 }
1395 else {
1396 status = U_ILLEGAL_CHARACTER;
1397 return;
1398 }
1399 }
1400
1401 if (baseElem == NULL) {
1402 if ((curElem = new PtnElem(basePattern, value)) == NULL ) {
1403 // out of memory
1404 status = U_MEMORY_ALLOCATION_ERROR;
1405 return;
1406 }
1407 if (baseChar >= LOW_A) {
1408 boot[26 + (baseChar-LOW_A)] = curElem;
1409 }
1410 else {
1411 boot[baseChar-CAP_A] = curElem;
1412 }
1413 curElem->skeleton = new PtnSkeleton(skeleton);
1414 curElem->skeletonWasSpecified = skeletonWasSpecified;
1415 }
1416 if ( baseElem != NULL ) {
1417 curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1418
1419 if (curElem == NULL) {
1420 // add new element to the list.
1421 curElem = baseElem;
1422 while( curElem -> next != NULL )
1423 {
1424 curElem = curElem->next;
1425 }
1426 if ((curElem->next = new PtnElem(basePattern, value)) == NULL ) {
1427 // out of memory
1428 status = U_MEMORY_ALLOCATION_ERROR;
1429 return;
1430 }
1431 curElem=curElem->next;
1432 curElem->skeleton = new PtnSkeleton(skeleton);
1433 curElem->skeletonWasSpecified = skeletonWasSpecified;
1434 }
1435 else {
1436 // Pattern exists in the list already.
1437 if ( !isDupAllowed ) {
1438 return;
1439 }
1440 // Overwrite the value.
1441 curElem->pattern = value;
1442 // It was a bug that we were not doing the following previously,
1443 // though that bug hid other problems by making things partly work.
1444 curElem->skeletonWasSpecified = skeletonWasSpecified;
1445 }
1446 }
1447 } // PatternMap::add
1448
1449 // Find the pattern from the given basePattern string.
1450 const UnicodeString *
getPatternFromBasePattern(UnicodeString & basePattern,UBool & skeletonWasSpecified)1451 PatternMap::getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeletonWasSpecified) { // key to search for
1452 PtnElem *curElem;
1453
1454 if ((curElem=getHeader(basePattern.charAt(0)))==NULL) {
1455 return NULL; // no match
1456 }
1457
1458 do {
1459 if ( basePattern.compare(curElem->basePattern)==0 ) {
1460 skeletonWasSpecified = curElem->skeletonWasSpecified;
1461 return &(curElem->pattern);
1462 }
1463 curElem=curElem->next;
1464 }while (curElem != NULL);
1465
1466 return NULL;
1467 } // PatternMap::getFromBasePattern
1468
1469
1470 // Find the pattern from the given skeleton.
1471 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
1472 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
1473 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
1474 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
1475 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
1476 const UnicodeString *
getPatternFromSkeleton(PtnSkeleton & skeleton,const PtnSkeleton ** specifiedSkeletonPtr)1477 PatternMap::getPatternFromSkeleton(PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) { // key to search for
1478 PtnElem *curElem;
1479
1480 if (specifiedSkeletonPtr) {
1481 *specifiedSkeletonPtr = NULL;
1482 }
1483
1484 // find boot entry
1485 UChar baseChar='\0';
1486 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1487 if (skeleton.baseOriginal[i].length() !=0 ) {
1488 baseChar = skeleton.baseOriginal[i].charAt(0);
1489 break;
1490 }
1491 }
1492
1493 if ((curElem=getHeader(baseChar))==NULL) {
1494 return NULL; // no match
1495 }
1496
1497 do {
1498 int32_t i=0;
1499 if (specifiedSkeletonPtr != NULL) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
1500 for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1501 if (curElem->skeleton->original[i].compare(skeleton.original[i]) != 0 )
1502 {
1503 break;
1504 }
1505 }
1506 } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
1507 for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1508 if (curElem->skeleton->baseOriginal[i].compare(skeleton.baseOriginal[i]) != 0 )
1509 {
1510 break;
1511 }
1512 }
1513 }
1514 if (i == UDATPG_FIELD_COUNT) {
1515 if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
1516 *specifiedSkeletonPtr = curElem->skeleton;
1517 }
1518 return &(curElem->pattern);
1519 }
1520 curElem=curElem->next;
1521 }while (curElem != NULL);
1522
1523 return NULL;
1524 }
1525
1526 UBool
equals(const PatternMap & other)1527 PatternMap::equals(const PatternMap& other) {
1528 if ( this==&other ) {
1529 return TRUE;
1530 }
1531 for (int32_t bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
1532 if ( boot[bootIndex]==other.boot[bootIndex] ) {
1533 continue;
1534 }
1535 if ( (boot[bootIndex]==NULL)||(other.boot[bootIndex]==NULL) ) {
1536 return FALSE;
1537 }
1538 PtnElem *otherElem = other.boot[bootIndex];
1539 PtnElem *myElem = boot[bootIndex];
1540 while ((otherElem!=NULL) || (myElem!=NULL)) {
1541 if ( myElem == otherElem ) {
1542 break;
1543 }
1544 if ((otherElem==NULL) || (myElem==NULL)) {
1545 return FALSE;
1546 }
1547 if ( (myElem->basePattern != otherElem->basePattern) ||
1548 (myElem->pattern != otherElem->pattern) ) {
1549 return FALSE;
1550 }
1551 if ((myElem->skeleton!=otherElem->skeleton)&&
1552 !myElem->skeleton->equals(*(otherElem->skeleton))) {
1553 return FALSE;
1554 }
1555 myElem = myElem->next;
1556 otherElem=otherElem->next;
1557 }
1558 }
1559 return TRUE;
1560 }
1561
1562 // find any key existing in the mapping table already.
1563 // return TRUE if there is an existing key, otherwise return FALSE.
1564 PtnElem*
getDuplicateElem(const UnicodeString & basePattern,const PtnSkeleton & skeleton,PtnElem * baseElem)1565 PatternMap::getDuplicateElem(
1566 const UnicodeString &basePattern,
1567 const PtnSkeleton &skeleton,
1568 PtnElem *baseElem) {
1569 PtnElem *curElem;
1570
1571 if ( baseElem == (PtnElem *)NULL ) {
1572 return (PtnElem*)NULL;
1573 }
1574 else {
1575 curElem = baseElem;
1576 }
1577 do {
1578 if ( basePattern.compare(curElem->basePattern)==0 ) {
1579 UBool isEqual=TRUE;
1580 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1581 if (curElem->skeleton->type[i] != skeleton.type[i] ) {
1582 isEqual=FALSE;
1583 break;
1584 }
1585 }
1586 if (isEqual) {
1587 return curElem;
1588 }
1589 }
1590 curElem = curElem->next;
1591 } while( curElem != (PtnElem *)NULL );
1592
1593 // end of the list
1594 return (PtnElem*)NULL;
1595
1596 } // PatternMap::getDuplicateElem
1597
DateTimeMatcher(void)1598 DateTimeMatcher::DateTimeMatcher(void) {
1599 }
1600
~DateTimeMatcher()1601 DateTimeMatcher::~DateTimeMatcher() {}
1602
DateTimeMatcher(const DateTimeMatcher & other)1603 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
1604 copyFrom(other.skeleton);
1605 }
1606
1607
1608 void
set(const UnicodeString & pattern,FormatParser * fp)1609 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
1610 PtnSkeleton localSkeleton;
1611 return set(pattern, fp, localSkeleton);
1612 }
1613
1614 void
set(const UnicodeString & pattern,FormatParser * fp,PtnSkeleton & skeletonResult)1615 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
1616 int32_t i;
1617 for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
1618 skeletonResult.type[i]=NONE;
1619 }
1620 fp->set(pattern);
1621 for (i=0; i < fp->itemNumber; i++) {
1622 UnicodeString field = fp->items[i];
1623 if ( field.charAt(0) == LOW_A ) {
1624 continue; // skip 'a'
1625 }
1626
1627 if ( fp->isQuoteLiteral(field) ) {
1628 UnicodeString quoteLiteral;
1629 fp->getQuoteLiteral(quoteLiteral, &i);
1630 continue;
1631 }
1632 int32_t canonicalIndex = fp->getCanonicalIndex(field);
1633 if (canonicalIndex < 0 ) {
1634 continue;
1635 }
1636 const dtTypeElem *row = &dtTypes[canonicalIndex];
1637 int32_t typeValue = row->field;
1638 skeletonResult.original[typeValue]=field;
1639 UChar repeatChar = row->patternChar;
1640 int32_t repeatCount = row->minLen; // #7930 removes cap at 3
1641 while (repeatCount-- > 0) {
1642 skeletonResult.baseOriginal[typeValue] += repeatChar;
1643 }
1644 int16_t subTypeValue = row->type;
1645 if ( row->type > 0) {
1646 subTypeValue += field.length();
1647 }
1648 skeletonResult.type[typeValue] = subTypeValue;
1649 }
1650 copyFrom(skeletonResult);
1651 }
1652
1653 void
getBasePattern(UnicodeString & result)1654 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
1655 result.remove(); // Reset the result first.
1656 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1657 if (skeleton.baseOriginal[i].length()!=0) {
1658 result += skeleton.baseOriginal[i];
1659 }
1660 }
1661 }
1662
1663 UnicodeString
getPattern()1664 DateTimeMatcher::getPattern() {
1665 UnicodeString result;
1666
1667 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1668 if (skeleton.original[i].length()!=0) {
1669 result += skeleton.original[i];
1670 }
1671 }
1672 return result;
1673 }
1674
1675 int32_t
getDistance(const DateTimeMatcher & other,int32_t includeMask,DistanceInfo & distanceInfo)1676 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) {
1677 int32_t result=0;
1678 distanceInfo.clear();
1679 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1680 int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
1681 int32_t otherType = other.skeleton.type[i];
1682 if (myType==otherType) {
1683 continue;
1684 }
1685 if (myType==0) {// and other is not
1686 result += EXTRA_FIELD;
1687 distanceInfo.addExtra(i);
1688 }
1689 else {
1690 if (otherType==0) {
1691 result += MISSING_FIELD;
1692 distanceInfo.addMissing(i);
1693 }
1694 else {
1695 result += abs(myType - otherType);
1696 }
1697 }
1698
1699 }
1700 return result;
1701 }
1702
1703 void
copyFrom(const PtnSkeleton & newSkeleton)1704 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
1705 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1706 this->skeleton.type[i]=newSkeleton.type[i];
1707 this->skeleton.original[i]=newSkeleton.original[i];
1708 this->skeleton.baseOriginal[i]=newSkeleton.baseOriginal[i];
1709 }
1710 }
1711
1712 void
copyFrom()1713 DateTimeMatcher::copyFrom() {
1714 // same as clear
1715 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1716 this->skeleton.type[i]=0;
1717 this->skeleton.original[i].remove();
1718 this->skeleton.baseOriginal[i].remove();
1719 }
1720 }
1721
1722 UBool
equals(const DateTimeMatcher * other) const1723 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
1724 if (other==NULL) {
1725 return FALSE;
1726 }
1727 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1728 if (this->skeleton.original[i]!=other->skeleton.original[i] ) {
1729 return FALSE;
1730 }
1731 }
1732 return TRUE;
1733 }
1734
1735 int32_t
getFieldMask()1736 DateTimeMatcher::getFieldMask() {
1737 int32_t result=0;
1738
1739 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1740 if (skeleton.type[i]!=0) {
1741 result |= (1<<i);
1742 }
1743 }
1744 return result;
1745 }
1746
1747 PtnSkeleton*
getSkeletonPtr()1748 DateTimeMatcher::getSkeletonPtr() {
1749 return &skeleton;
1750 }
1751
FormatParser()1752 FormatParser::FormatParser () {
1753 status = START;
1754 itemNumber=0;
1755 }
1756
1757
~FormatParser()1758 FormatParser::~FormatParser () {
1759 }
1760
1761
1762 // Find the next token with the starting position and length
1763 // Note: the startPos may
1764 FormatParser::TokenStatus
setTokens(const UnicodeString & pattern,int32_t startPos,int32_t * len)1765 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
1766 int32_t curLoc = startPos;
1767 if ( curLoc >= pattern.length()) {
1768 return DONE;
1769 }
1770 // check the current char is between A-Z or a-z
1771 do {
1772 UChar c=pattern.charAt(curLoc);
1773 if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
1774 curLoc++;
1775 }
1776 else {
1777 startPos = curLoc;
1778 *len=1;
1779 return ADD_TOKEN;
1780 }
1781
1782 if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
1783 break; // not the same token
1784 }
1785 } while(curLoc <= pattern.length());
1786 *len = curLoc-startPos;
1787 return ADD_TOKEN;
1788 }
1789
1790 void
set(const UnicodeString & pattern)1791 FormatParser::set(const UnicodeString& pattern) {
1792 int32_t startPos=0;
1793 TokenStatus result=START;
1794 int32_t len=0;
1795 itemNumber =0;
1796
1797 do {
1798 result = setTokens( pattern, startPos, &len );
1799 if ( result == ADD_TOKEN )
1800 {
1801 items[itemNumber++] = UnicodeString(pattern, startPos, len );
1802 startPos += len;
1803 }
1804 else {
1805 break;
1806 }
1807 } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
1808 }
1809
1810 int32_t
getCanonicalIndex(const UnicodeString & s,UBool strict)1811 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
1812 int32_t len = s.length();
1813 if (len == 0) {
1814 return -1;
1815 }
1816 UChar ch = s.charAt(0);
1817
1818 // Verify that all are the same character.
1819 for (int32_t l = 1; l < len; l++) {
1820 if (ch != s.charAt(l)) {
1821 return -1;
1822 }
1823 }
1824 int32_t i = 0;
1825 int32_t bestRow = -1;
1826 while (dtTypes[i].patternChar != '\0') {
1827 if ( dtTypes[i].patternChar != ch ) {
1828 ++i;
1829 continue;
1830 }
1831 bestRow = i;
1832 if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
1833 return i;
1834 }
1835 if (dtTypes[i+1].minLen <= len) {
1836 ++i;
1837 continue;
1838 }
1839 return i;
1840 }
1841 return strict ? -1 : bestRow;
1842 }
1843
1844 UBool
isQuoteLiteral(const UnicodeString & s) const1845 FormatParser::isQuoteLiteral(const UnicodeString& s) const {
1846 return (UBool)(s.charAt(0)==SINGLE_QUOTE);
1847 }
1848
1849 // This function aussumes the current itemIndex points to the quote literal.
1850 // Please call isQuoteLiteral prior to this function.
1851 void
getQuoteLiteral(UnicodeString & quote,int32_t * itemIndex)1852 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
1853 int32_t i=*itemIndex;
1854
1855 quote.remove();
1856 if (items[i].charAt(0)==SINGLE_QUOTE) {
1857 quote += items[i];
1858 ++i;
1859 }
1860 while ( i < itemNumber ) {
1861 if ( items[i].charAt(0)==SINGLE_QUOTE ) {
1862 if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
1863 // two single quotes e.g. 'o''clock'
1864 quote += items[i++];
1865 quote += items[i++];
1866 continue;
1867 }
1868 else {
1869 quote += items[i];
1870 break;
1871 }
1872 }
1873 else {
1874 quote += items[i];
1875 }
1876 ++i;
1877 }
1878 *itemIndex=i;
1879 }
1880
1881 UBool
isPatternSeparator(UnicodeString & field)1882 FormatParser::isPatternSeparator(UnicodeString& field) {
1883 for (int32_t i=0; i<field.length(); ++i ) {
1884 UChar c= field.charAt(i);
1885 if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
1886 (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
1887 continue;
1888 }
1889 else {
1890 return FALSE;
1891 }
1892 }
1893 return TRUE;
1894 }
1895
~DistanceInfo()1896 DistanceInfo::~DistanceInfo() {}
1897
1898 void
setTo(DistanceInfo & other)1899 DistanceInfo::setTo(DistanceInfo &other) {
1900 missingFieldMask = other.missingFieldMask;
1901 extraFieldMask= other.extraFieldMask;
1902 }
1903
PatternMapIterator()1904 PatternMapIterator::PatternMapIterator() {
1905 bootIndex = 0;
1906 nodePtr = NULL;
1907 patternMap=NULL;
1908 matcher= new DateTimeMatcher();
1909 }
1910
1911
~PatternMapIterator()1912 PatternMapIterator::~PatternMapIterator() {
1913 delete matcher;
1914 }
1915
1916 void
set(PatternMap & newPatternMap)1917 PatternMapIterator::set(PatternMap& newPatternMap) {
1918 this->patternMap=&newPatternMap;
1919 }
1920
1921 PtnSkeleton*
getSkeleton()1922 PatternMapIterator::getSkeleton() {
1923 if ( nodePtr == NULL ) {
1924 return NULL;
1925 }
1926 else {
1927 return nodePtr->skeleton;
1928 }
1929 }
1930
1931 UBool
hasNext()1932 PatternMapIterator::hasNext() {
1933 int32_t headIndex=bootIndex;
1934 PtnElem *curPtr=nodePtr;
1935
1936 if (patternMap==NULL) {
1937 return FALSE;
1938 }
1939 while ( headIndex < MAX_PATTERN_ENTRIES ) {
1940 if ( curPtr != NULL ) {
1941 if ( curPtr->next != NULL ) {
1942 return TRUE;
1943 }
1944 else {
1945 headIndex++;
1946 curPtr=NULL;
1947 continue;
1948 }
1949 }
1950 else {
1951 if ( patternMap->boot[headIndex] != NULL ) {
1952 return TRUE;
1953 }
1954 else {
1955 headIndex++;
1956 continue;
1957 }
1958 }
1959
1960 }
1961 return FALSE;
1962 }
1963
1964 DateTimeMatcher&
next()1965 PatternMapIterator::next() {
1966 while ( bootIndex < MAX_PATTERN_ENTRIES ) {
1967 if ( nodePtr != NULL ) {
1968 if ( nodePtr->next != NULL ) {
1969 nodePtr = nodePtr->next;
1970 break;
1971 }
1972 else {
1973 bootIndex++;
1974 nodePtr=NULL;
1975 continue;
1976 }
1977 }
1978 else {
1979 if ( patternMap->boot[bootIndex] != NULL ) {
1980 nodePtr = patternMap->boot[bootIndex];
1981 break;
1982 }
1983 else {
1984 bootIndex++;
1985 continue;
1986 }
1987 }
1988 }
1989 if (nodePtr!=NULL) {
1990 matcher->copyFrom(*nodePtr->skeleton);
1991 }
1992 else {
1993 matcher->copyFrom();
1994 }
1995 return *matcher;
1996 }
1997
PtnSkeleton()1998 PtnSkeleton::PtnSkeleton() {
1999 }
2000
2001
PtnSkeleton(const PtnSkeleton & other)2002 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
2003 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2004 this->type[i]=other.type[i];
2005 this->original[i]=other.original[i];
2006 this->baseOriginal[i]=other.baseOriginal[i];
2007 }
2008 }
2009
2010 UBool
equals(const PtnSkeleton & other)2011 PtnSkeleton::equals(const PtnSkeleton& other) {
2012 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2013 if ( (type[i]!= other.type[i]) ||
2014 (original[i]!=other.original[i]) ||
2015 (baseOriginal[i]!=other.baseOriginal[i]) ) {
2016 return FALSE;
2017 }
2018 }
2019 return TRUE;
2020 }
2021
2022 UnicodeString
getSkeleton()2023 PtnSkeleton::getSkeleton() {
2024 UnicodeString result;
2025
2026 for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
2027 if (original[i].length()!=0) {
2028 result += original[i];
2029 }
2030 }
2031 return result;
2032 }
2033
2034 UnicodeString
getBaseSkeleton()2035 PtnSkeleton::getBaseSkeleton() {
2036 UnicodeString result;
2037
2038 for(int32_t i=0; i< UDATPG_FIELD_COUNT; ++i) {
2039 if (baseOriginal[i].length()!=0) {
2040 result += baseOriginal[i];
2041 }
2042 }
2043 return result;
2044 }
2045
~PtnSkeleton()2046 PtnSkeleton::~PtnSkeleton() {
2047 }
2048
PtnElem(const UnicodeString & basePat,const UnicodeString & pat)2049 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
2050 basePattern(basePat),
2051 skeleton(NULL),
2052 pattern(pat),
2053 next(NULL)
2054 {
2055 }
2056
~PtnElem()2057 PtnElem::~PtnElem() {
2058
2059 if (next!=NULL) {
2060 delete next;
2061 }
2062 delete skeleton;
2063 }
2064
DTSkeletonEnumeration(PatternMap & patternMap,dtStrEnum type,UErrorCode & status)2065 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum type, UErrorCode& status) {
2066 PtnElem *curElem;
2067 PtnSkeleton *curSkeleton;
2068 UnicodeString s;
2069 int32_t bootIndex;
2070
2071 pos=0;
2072 fSkeletons = new UVector(status);
2073 if (U_FAILURE(status)) {
2074 delete fSkeletons;
2075 return;
2076 }
2077 for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2078 curElem = patternMap.boot[bootIndex];
2079 while (curElem!=NULL) {
2080 switch(type) {
2081 case DT_BASESKELETON:
2082 s=curElem->basePattern;
2083 break;
2084 case DT_PATTERN:
2085 s=curElem->pattern;
2086 break;
2087 case DT_SKELETON:
2088 curSkeleton=curElem->skeleton;
2089 s=curSkeleton->getSkeleton();
2090 break;
2091 }
2092 if ( !isCanonicalItem(s) ) {
2093 fSkeletons->addElement(new UnicodeString(s), status);
2094 if (U_FAILURE(status)) {
2095 delete fSkeletons;
2096 fSkeletons = NULL;
2097 return;
2098 }
2099 }
2100 curElem = curElem->next;
2101 }
2102 }
2103 if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=NULL) ) {
2104 status = U_BUFFER_OVERFLOW_ERROR;
2105 }
2106 }
2107
2108 const UnicodeString*
snext(UErrorCode & status)2109 DTSkeletonEnumeration::snext(UErrorCode& status) {
2110 if (U_SUCCESS(status) && pos < fSkeletons->size()) {
2111 return (const UnicodeString*)fSkeletons->elementAt(pos++);
2112 }
2113 return NULL;
2114 }
2115
2116 void
reset(UErrorCode &)2117 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2118 pos=0;
2119 }
2120
2121 int32_t
count(UErrorCode &) const2122 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2123 return (fSkeletons==NULL) ? 0 : fSkeletons->size();
2124 }
2125
2126 UBool
isCanonicalItem(const UnicodeString & item)2127 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2128 if ( item.length() != 1 ) {
2129 return FALSE;
2130 }
2131 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2132 if (item.charAt(0)==Canonical_Items[i]) {
2133 return TRUE;
2134 }
2135 }
2136 return FALSE;
2137 }
2138
~DTSkeletonEnumeration()2139 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2140 UnicodeString *s;
2141 for (int32_t i=0; i<fSkeletons->size(); ++i) {
2142 if ((s=(UnicodeString *)fSkeletons->elementAt(i))!=NULL) {
2143 delete s;
2144 }
2145 }
2146 delete fSkeletons;
2147 }
2148
DTRedundantEnumeration()2149 DTRedundantEnumeration::DTRedundantEnumeration() {
2150 pos=0;
2151 fPatterns = NULL;
2152 }
2153
2154 void
add(const UnicodeString & pattern,UErrorCode & status)2155 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2156 if (U_FAILURE(status)) return;
2157 if (fPatterns == NULL) {
2158 fPatterns = new UVector(status);
2159 if (U_FAILURE(status)) {
2160 delete fPatterns;
2161 fPatterns = NULL;
2162 return;
2163 }
2164 }
2165 fPatterns->addElement(new UnicodeString(pattern), status);
2166 if (U_FAILURE(status)) {
2167 delete fPatterns;
2168 fPatterns = NULL;
2169 return;
2170 }
2171 }
2172
2173 const UnicodeString*
snext(UErrorCode & status)2174 DTRedundantEnumeration::snext(UErrorCode& status) {
2175 if (U_SUCCESS(status) && pos < fPatterns->size()) {
2176 return (const UnicodeString*)fPatterns->elementAt(pos++);
2177 }
2178 return NULL;
2179 }
2180
2181 void
reset(UErrorCode &)2182 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2183 pos=0;
2184 }
2185
2186 int32_t
count(UErrorCode &) const2187 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2188 return (fPatterns==NULL) ? 0 : fPatterns->size();
2189 }
2190
2191 UBool
isCanonicalItem(const UnicodeString & item)2192 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) {
2193 if ( item.length() != 1 ) {
2194 return FALSE;
2195 }
2196 for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2197 if (item.charAt(0)==Canonical_Items[i]) {
2198 return TRUE;
2199 }
2200 }
2201 return FALSE;
2202 }
2203
~DTRedundantEnumeration()2204 DTRedundantEnumeration::~DTRedundantEnumeration() {
2205 UnicodeString *s;
2206 for (int32_t i=0; i<fPatterns->size(); ++i) {
2207 if ((s=(UnicodeString *)fPatterns->elementAt(i))!=NULL) {
2208 delete s;
2209 }
2210 }
2211 delete fPatterns;
2212 }
2213
2214 U_NAMESPACE_END
2215
2216
2217 #endif /* #if !UCONFIG_NO_FORMATTING */
2218
2219 //eof
2220