• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 * Copyright (C) 2007-2016, International Business Machines Corporation and
6 * others. All Rights Reserved.
7 *******************************************************************************
8 *
9 * File DTPTNGEN.CPP
10 *
11 *******************************************************************************
12 */
13 
14 #include "unicode/utypes.h"
15 #if !UCONFIG_NO_FORMATTING
16 
17 #include "unicode/datefmt.h"
18 #include "unicode/decimfmt.h"
19 #include "unicode/dtfmtsym.h"
20 #include "unicode/dtptngen.h"
21 #include "unicode/localpointer.h"
22 #include "unicode/simpleformatter.h"
23 #include "unicode/smpdtfmt.h"
24 #include "unicode/udat.h"
25 #include "unicode/udatpg.h"
26 #include "unicode/uniset.h"
27 #include "unicode/uloc.h"
28 #include "unicode/ures.h"
29 #include "unicode/ustring.h"
30 #include "unicode/rep.h"
31 #include "cpputils.h"
32 #include "mutex.h"
33 #include "umutex.h"
34 #include "cmemory.h"
35 #include "cstring.h"
36 #include "locbased.h"
37 #include "hash.h"
38 #include "uhash.h"
39 #include "uresimp.h"
40 #include "dtptngen_impl.h"
41 #include "ucln_in.h"
42 #include "charstr.h"
43 #include "uassert.h"
44 
45 #if U_CHARSET_FAMILY==U_EBCDIC_FAMILY
46 /**
47  * If we are on EBCDIC, use an iterator which will
48  * traverse the bundles in ASCII order.
49  */
50 #define U_USE_ASCII_BUNDLE_ITERATOR
51 #define U_SORT_ASCII_BUNDLE_ITERATOR
52 #endif
53 
54 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
55 
56 #include "unicode/ustring.h"
57 #include "uarrsort.h"
58 
59 struct UResAEntry {
60     UChar *key;
61     UResourceBundle *item;
62 };
63 
64 struct UResourceBundleAIterator {
65     UResourceBundle  *bund;
66     UResAEntry *entries;
67     int32_t num;
68     int32_t cursor;
69 };
70 
71 /* Must be C linkage to pass function pointer to the sort function */
72 
73 U_CDECL_BEGIN
74 
75 static int32_t U_CALLCONV
ures_a_codepointSort(const void * context,const void * left,const void * right)76 ures_a_codepointSort(const void *context, const void *left, const void *right) {
77     //CompareContext *cmp=(CompareContext *)context;
78     return u_strcmp(((const UResAEntry *)left)->key,
79                     ((const UResAEntry *)right)->key);
80 }
81 
82 U_CDECL_END
83 
ures_a_open(UResourceBundleAIterator * aiter,UResourceBundle * bund,UErrorCode * status)84 static void ures_a_open(UResourceBundleAIterator *aiter, UResourceBundle *bund, UErrorCode *status) {
85     if(U_FAILURE(*status)) {
86         return;
87     }
88     aiter->bund = bund;
89     aiter->num = ures_getSize(aiter->bund);
90     aiter->cursor = 0;
91 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
92     aiter->entries = nullptr;
93 #else
94     aiter->entries = (UResAEntry*)uprv_malloc(sizeof(UResAEntry)*aiter->num);
95     for(int i=0;i<aiter->num;i++) {
96         aiter->entries[i].item = ures_getByIndex(aiter->bund, i, nullptr, status);
97         const char *akey = ures_getKey(aiter->entries[i].item);
98         int32_t len = uprv_strlen(akey)+1;
99         aiter->entries[i].key = (UChar*)uprv_malloc(len*sizeof(UChar));
100         u_charsToUChars(akey, aiter->entries[i].key, len);
101     }
102     uprv_sortArray(aiter->entries, aiter->num, sizeof(UResAEntry), ures_a_codepointSort, nullptr, TRUE, status);
103 #endif
104 }
105 
ures_a_close(UResourceBundleAIterator * aiter)106 static void ures_a_close(UResourceBundleAIterator *aiter) {
107 #if defined(U_SORT_ASCII_BUNDLE_ITERATOR)
108     for(int i=0;i<aiter->num;i++) {
109         uprv_free(aiter->entries[i].key);
110         ures_close(aiter->entries[i].item);
111     }
112 #endif
113 }
114 
ures_a_getNextString(UResourceBundleAIterator * aiter,int32_t * len,const char ** key,UErrorCode * err)115 static const UChar *ures_a_getNextString(UResourceBundleAIterator *aiter, int32_t *len, const char **key, UErrorCode *err) {
116 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
117     return ures_getNextString(aiter->bund, len, key, err);
118 #else
119     if(U_FAILURE(*err)) return nullptr;
120     UResourceBundle *item = aiter->entries[aiter->cursor].item;
121     const UChar* ret = ures_getString(item, len, err);
122     *key = ures_getKey(item);
123     aiter->cursor++;
124     return ret;
125 #endif
126 }
127 
128 
129 #endif
130 
131 
132 U_NAMESPACE_BEGIN
133 
134 // *****************************************************************************
135 // class DateTimePatternGenerator
136 // *****************************************************************************
137 static const UChar Canonical_Items[] = {
138     // GyQMwWEDFdaHmsSv
139     CAP_G, LOW_Y, CAP_Q, CAP_M, LOW_W, CAP_W, CAP_E,
140     CAP_D, CAP_F, LOW_D, LOW_A, // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J
141     CAP_H, LOW_M, LOW_S, CAP_S, LOW_V, 0
142 };
143 
144 static const dtTypeElem dtTypes[] = {
145     // patternChar, field, type, minLen, weight
146     {CAP_G, UDATPG_ERA_FIELD, DT_SHORT, 1, 3,},
147     {CAP_G, UDATPG_ERA_FIELD, DT_LONG,  4, 0},
148     {CAP_G, UDATPG_ERA_FIELD, DT_NARROW, 5, 0},
149 
150     {LOW_Y, UDATPG_YEAR_FIELD, DT_NUMERIC, 1, 20},
151     {CAP_Y, UDATPG_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 20},
152     {LOW_U, UDATPG_YEAR_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 20},
153     {LOW_R, UDATPG_YEAR_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20},
154     {CAP_U, UDATPG_YEAR_FIELD, DT_SHORT, 1, 3},
155     {CAP_U, UDATPG_YEAR_FIELD, DT_LONG, 4, 0},
156     {CAP_U, UDATPG_YEAR_FIELD, DT_NARROW, 5, 0},
157 
158     {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2},
159     {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0},
160     {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0},
161     {CAP_Q, UDATPG_QUARTER_FIELD, DT_NARROW, 5, 0},
162     {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
163     {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT - DT_DELTA, 3, 0},
164     {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG - DT_DELTA, 4, 0},
165     {LOW_Q, UDATPG_QUARTER_FIELD, DT_NARROW - DT_DELTA, 5, 0},
166 
167     {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2},
168     {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0},
169     {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0},
170     {CAP_M, UDATPG_MONTH_FIELD, DT_NARROW, 5, 0},
171     {CAP_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
172     {CAP_L, UDATPG_MONTH_FIELD, DT_SHORT - DT_DELTA, 3, 0},
173     {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0},
174     {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0},
175     {LOW_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 1},
176 
177     {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2},
178 
179     {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC, 1, 0},
180 
181     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3},
182     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0},
183     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0},
184     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER, 6, 0},
185     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2},
186     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0},
187     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
188     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0},
189     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORTER - 2*DT_DELTA, 6, 0},
190     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical
191     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0},
192     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0},
193     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0},
194     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER - DT_DELTA, 6, 0},
195 
196     {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2},
197     {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 20}, // really internal use, so we don't care
198 
199     {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC, 1, 3},
200 
201     {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC, 1, 0},
202 
203     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 3},
204     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_LONG, 4, 0},
205     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_NARROW, 5, 0},
206     {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - DT_DELTA, 1, 3},
207     {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - DT_DELTA, 4, 0},
208     {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - DT_DELTA, 5, 0},
209     // b needs to be closer to a than to B, so we make this 3*DT_DELTA
210     {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - 3*DT_DELTA, 1, 3},
211     {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - 3*DT_DELTA, 4, 0},
212     {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - 3*DT_DELTA, 5, 0},
213 
214     {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour
215     {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2}, // 24 hour
216     {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour
217     {CAP_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // 12 hour
218     // The C code has had versions of the following 3, keep & update. Should not need these, but...
219     // Without these, certain tests using e.g. staticGetSkeleton fail because j/J in patterns
220     // get skipped instead of mapped to the right hour chars, for example in
221     //   DateFormatTest::TestPatternFromSkeleton
222     //   IntlTestDateTimePatternGeneratorAPI:: testStaticGetSkeleton
223     //   DateIntervalFormatTest::testTicket11985
224     // Need to investigate better handling of jJC replacement e.g. in staticGetSkeleton.
225     {CAP_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 5*DT_DELTA, 1, 2}, // 12/24 hour no AM/PM
226     {LOW_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 6*DT_DELTA, 1, 6}, // 12/24 hour
227     {CAP_C, UDATPG_HOUR_FIELD, DT_NUMERIC + 7*DT_DELTA, 1, 6}, // 12/24 hour with preferred dayPeriods for 12
228 
229     {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2},
230 
231     {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2},
232     {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000},
233 
234     {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC, 1, 1000},
235 
236     {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0},
237     {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
238     {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3},
239     {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0},
240     {CAP_Z, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 3},
241     {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
242     {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 5, 0},
243     {CAP_O, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0},
244     {CAP_O, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
245     {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0},
246     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 2, 0},
247     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-1 - DT_DELTA, 3, 0},
248     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-2 - DT_DELTA, 4, 0},
249     {CAP_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0},
250     {CAP_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0},
251     {CAP_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
252     {LOW_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0},
253     {LOW_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0},
254     {LOW_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
255 
256     {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[]
257  };
258 
259 static const char* const CLDR_FIELD_APPEND[] = {
260     "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week",
261     "*", "*", "Day", "*", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J
262     "Hour", "Minute", "Second", "*", "Timezone"
263 };
264 
265 static const char* const CLDR_FIELD_NAME[UDATPG_FIELD_COUNT] = {
266     "era", "year", "quarter", "month", "week", "weekOfMonth", "weekday",
267     "dayOfYear", "weekdayOfMonth", "day", "dayperiod", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J
268     "hour", "minute", "second", "*", "zone"
269 };
270 
271 static const char* const CLDR_FIELD_WIDTH[] = { // [UDATPG_WIDTH_COUNT]
272     "", "-short", "-narrow"
273 };
274 
275 // TODO(ticket:13619): remove when definition uncommented in dtptngen.h.
276 static const int32_t UDATPG_WIDTH_COUNT = UDATPG_NARROW + 1;
277 static constexpr UDateTimePGDisplayWidth UDATPG_WIDTH_APPENDITEM = UDATPG_WIDE;
278 static constexpr int32_t UDATPG_FIELD_KEY_MAX = 24; // max length of CLDR field tag (type + width)
279 
280 // For appendItems
281 static const UChar UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A,
282     0x20, 0x7B, 0x31, 0x7D, 0x2524, 0};  // {0} \u251C{2}: {1}\u2524
283 
284 //static const UChar repeatedPatterns[6]={CAP_G, CAP_E, LOW_Z, LOW_V, CAP_Q, 0}; // "GEzvQ"
285 
286 static const char DT_DateTimePatternsTag[]="DateTimePatterns";
287 static const char DT_DateTimeCalendarTag[]="calendar";
288 static const char DT_DateTimeGregorianTag[]="gregorian";
289 static const char DT_DateTimeAppendItemsTag[]="appendItems";
290 static const char DT_DateTimeFieldsTag[]="fields";
291 static const char DT_DateTimeAvailableFormatsTag[]="availableFormats";
292 //static const UnicodeString repeatedPattern=UnicodeString(repeatedPatterns);
293 
294 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateTimePatternGenerator)
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)295 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)
296 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTRedundantEnumeration)
297 
298 DateTimePatternGenerator*  U_EXPORT2
299 DateTimePatternGenerator::createInstance(UErrorCode& status) {
300     return createInstance(Locale::getDefault(), status);
301 }
302 
303 DateTimePatternGenerator* U_EXPORT2
createInstance(const Locale & locale,UErrorCode & status)304 DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) {
305     if (U_FAILURE(status)) {
306         return nullptr;
307     }
308     LocalPointer<DateTimePatternGenerator> result(
309             new DateTimePatternGenerator(locale, status), status);
310     return U_SUCCESS(status) ? result.orphan() : nullptr;
311 }
312 
313 DateTimePatternGenerator*  U_EXPORT2
createEmptyInstance(UErrorCode & status)314 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) {
315     if (U_FAILURE(status)) {
316         return nullptr;
317     }
318     LocalPointer<DateTimePatternGenerator> result(
319             new DateTimePatternGenerator(status), status);
320     return U_SUCCESS(status) ? result.orphan() : nullptr;
321 }
322 
DateTimePatternGenerator(UErrorCode & status)323 DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) :
324     skipMatcher(nullptr),
325     fAvailableFormatKeyHash(nullptr),
326     internalErrorCode(U_ZERO_ERROR)
327 {
328     fp = new FormatParser();
329     dtMatcher = new DateTimeMatcher();
330     distanceInfo = new DistanceInfo();
331     patternMap = new PatternMap();
332     if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) {
333         internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR;
334     }
335 }
336 
DateTimePatternGenerator(const Locale & locale,UErrorCode & status)337 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status) :
338     skipMatcher(nullptr),
339     fAvailableFormatKeyHash(nullptr),
340     internalErrorCode(U_ZERO_ERROR)
341 {
342     fp = new FormatParser();
343     dtMatcher = new DateTimeMatcher();
344     distanceInfo = new DistanceInfo();
345     patternMap = new PatternMap();
346     if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) {
347         internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR;
348     }
349     else {
350         initData(locale, status);
351     }
352 }
353 
DateTimePatternGenerator(const DateTimePatternGenerator & other)354 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) :
355     UObject(),
356     skipMatcher(nullptr),
357     fAvailableFormatKeyHash(nullptr),
358     internalErrorCode(U_ZERO_ERROR)
359 {
360     fp = new FormatParser();
361     dtMatcher = new DateTimeMatcher();
362     distanceInfo = new DistanceInfo();
363     patternMap = new PatternMap();
364     if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) {
365         internalErrorCode = U_MEMORY_ALLOCATION_ERROR;
366     }
367     *this=other;
368 }
369 
370 DateTimePatternGenerator&
operator =(const DateTimePatternGenerator & other)371 DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) {
372     // reflexive case
373     if (&other == this) {
374         return *this;
375     }
376     internalErrorCode = other.internalErrorCode;
377     pLocale = other.pLocale;
378     fDefaultHourFormatChar = other.fDefaultHourFormatChar;
379     *fp = *(other.fp);
380     dtMatcher->copyFrom(other.dtMatcher->skeleton);
381     *distanceInfo = *(other.distanceInfo);
382     dateTimeFormat = other.dateTimeFormat;
383     decimal = other.decimal;
384     // NUL-terminate for the C API.
385     dateTimeFormat.getTerminatedBuffer();
386     decimal.getTerminatedBuffer();
387     delete skipMatcher;
388     if ( other.skipMatcher == nullptr ) {
389         skipMatcher = nullptr;
390     }
391     else {
392         skipMatcher = new DateTimeMatcher(*other.skipMatcher);
393         if (skipMatcher == nullptr)
394         {
395             internalErrorCode = U_MEMORY_ALLOCATION_ERROR;
396             return *this;
397         }
398     }
399     for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) {
400         appendItemFormats[i] = other.appendItemFormats[i];
401         appendItemFormats[i].getTerminatedBuffer(); // NUL-terminate for the C API.
402         for (int32_t j=0; j< UDATPG_WIDTH_COUNT; ++j ) {
403             fieldDisplayNames[i][j] = other.fieldDisplayNames[i][j];
404             fieldDisplayNames[i][j].getTerminatedBuffer(); // NUL-terminate for the C API.
405         }
406     }
407     patternMap->copyFrom(*other.patternMap, internalErrorCode);
408     copyHashtable(other.fAvailableFormatKeyHash, internalErrorCode);
409     return *this;
410 }
411 
412 
413 UBool
operator ==(const DateTimePatternGenerator & other) const414 DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) const {
415     if (this == &other) {
416         return TRUE;
417     }
418     if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) &&
419         (dateTimeFormat==other.dateTimeFormat) && (decimal==other.decimal)) {
420         for ( int32_t i=0 ; i<UDATPG_FIELD_COUNT; ++i ) {
421             if (appendItemFormats[i] != other.appendItemFormats[i]) {
422                 return FALSE;
423             }
424             for (int32_t j=0; j< UDATPG_WIDTH_COUNT; ++j ) {
425                 if (fieldDisplayNames[i][j] != other.fieldDisplayNames[i][j]) {
426                     return FALSE;
427                 }
428             }
429         }
430         return TRUE;
431     }
432     else {
433         return FALSE;
434     }
435 }
436 
437 UBool
operator !=(const DateTimePatternGenerator & other) const438 DateTimePatternGenerator::operator!=(const DateTimePatternGenerator& other) const {
439     return  !operator==(other);
440 }
441 
~DateTimePatternGenerator()442 DateTimePatternGenerator::~DateTimePatternGenerator() {
443     if (fAvailableFormatKeyHash!=nullptr) {
444         delete fAvailableFormatKeyHash;
445     }
446 
447     if (fp != nullptr) delete fp;
448     if (dtMatcher != nullptr) delete dtMatcher;
449     if (distanceInfo != nullptr) delete distanceInfo;
450     if (patternMap != nullptr) delete patternMap;
451     if (skipMatcher != nullptr) delete skipMatcher;
452 }
453 
454 namespace {
455 
456 UInitOnce initOnce = U_INITONCE_INITIALIZER;
457 UHashtable *localeToAllowedHourFormatsMap = nullptr;
458 
459 // Value deleter for hashmap.
deleteAllowedHourFormats(void * ptr)460 U_CFUNC void U_CALLCONV deleteAllowedHourFormats(void *ptr) {
461     uprv_free(ptr);
462 }
463 
464 // Close hashmap at cleanup.
allowedHourFormatsCleanup()465 U_CFUNC UBool U_CALLCONV allowedHourFormatsCleanup() {
466     uhash_close(localeToAllowedHourFormatsMap);
467     return TRUE;
468 }
469 
470 enum AllowedHourFormat{
471     ALLOWED_HOUR_FORMAT_UNKNOWN = -1,
472     ALLOWED_HOUR_FORMAT_h,
473     ALLOWED_HOUR_FORMAT_H,
474     ALLOWED_HOUR_FORMAT_hb,
475     ALLOWED_HOUR_FORMAT_Hb,
476     ALLOWED_HOUR_FORMAT_hB,
477     ALLOWED_HOUR_FORMAT_HB
478 };
479 
480 }  // namespace
481 
482 void
initData(const Locale & locale,UErrorCode & status)483 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status) {
484     //const char *baseLangName = locale.getBaseName(); // unused
485 
486     skipMatcher = nullptr;
487     fAvailableFormatKeyHash=nullptr;
488     addCanonicalItems(status);
489     addICUPatterns(locale, status);
490     addCLDRData(locale, status);
491     setDateTimeFromCalendar(locale, status);
492     setDecimalSymbols(locale, status);
493     umtx_initOnce(initOnce, loadAllowedHourFormatsData, status);
494     getAllowedHourFormats(locale, status);
495     // If any of the above methods failed then the object is in an invalid state.
496     internalErrorCode = status;
497 } // DateTimePatternGenerator::initData
498 
499 namespace {
500 
501 struct AllowedHourFormatsSink : public ResourceSink {
502     // Initialize sub-sinks.
AllowedHourFormatsSink__anonbd2c24a10211::AllowedHourFormatsSink503     AllowedHourFormatsSink() {}
504     virtual ~AllowedHourFormatsSink();
505 
put__anonbd2c24a10211::AllowedHourFormatsSink506     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
507                      UErrorCode &errorCode) {
508         ResourceTable timeData = value.getTable(errorCode);
509         if (U_FAILURE(errorCode)) { return; }
510         for (int32_t i = 0; timeData.getKeyAndValue(i, key, value); ++i) {
511             const char *regionOrLocale = key;
512             ResourceTable formatList = value.getTable(errorCode);
513             if (U_FAILURE(errorCode)) { return; }
514             for (int32_t j = 0; formatList.getKeyAndValue(j, key, value); ++j) {
515                 if (uprv_strcmp(key, "allowed") == 0) {  // Ignore "preferred" list.
516                     LocalMemory<int32_t> list;
517                     int32_t length;
518                     if (value.getType() == URES_STRING) {
519                         if (list.allocateInsteadAndReset(2) == nullptr) {
520                             errorCode = U_MEMORY_ALLOCATION_ERROR;
521                             return;
522                         }
523                         list[0] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
524                         length = 1;
525                     }
526                     else {
527                         ResourceArray allowedFormats = value.getArray(errorCode);
528                         length = allowedFormats.getSize();
529                         if (list.allocateInsteadAndReset(length + 1) == nullptr) {
530                             errorCode = U_MEMORY_ALLOCATION_ERROR;
531                             return;
532                         }
533                         for (int32_t k = 0; k < length; ++k) {
534                             allowedFormats.getValue(k, value);
535                             list[k] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
536                         }
537                     }
538                     list[length] = ALLOWED_HOUR_FORMAT_UNKNOWN;
539                     uhash_put(localeToAllowedHourFormatsMap,
540                               const_cast<char *>(regionOrLocale), list.orphan(), &errorCode);
541                     if (U_FAILURE(errorCode)) { return; }
542                 }
543             }
544         }
545     }
546 
getHourFormatFromUnicodeString__anonbd2c24a10211::AllowedHourFormatsSink547     AllowedHourFormat getHourFormatFromUnicodeString(const UnicodeString &s) {
548         if (s.length() == 1) {
549             if (s[0] == LOW_H) { return ALLOWED_HOUR_FORMAT_h; }
550             if (s[0] == CAP_H) { return ALLOWED_HOUR_FORMAT_H; }
551         } else if (s.length() == 2) {
552             if (s[0] == LOW_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_hb; }
553             if (s[0] == CAP_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Hb; }
554             if (s[0] == LOW_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_hB; }
555             if (s[0] == CAP_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_HB; }
556         }
557 
558         return ALLOWED_HOUR_FORMAT_UNKNOWN;
559     }
560 };
561 
562 }  // namespace
563 
~AllowedHourFormatsSink()564 AllowedHourFormatsSink::~AllowedHourFormatsSink() {}
565 
loadAllowedHourFormatsData(UErrorCode & status)566 U_CFUNC void U_CALLCONV DateTimePatternGenerator::loadAllowedHourFormatsData(UErrorCode &status) {
567     if (U_FAILURE(status)) { return; }
568     localeToAllowedHourFormatsMap = uhash_open(
569         uhash_hashChars, uhash_compareChars, nullptr, &status);
570     if (U_FAILURE(status)) { return; }
571 
572     uhash_setValueDeleter(localeToAllowedHourFormatsMap, deleteAllowedHourFormats);
573     ucln_i18n_registerCleanup(UCLN_I18N_ALLOWED_HOUR_FORMATS, allowedHourFormatsCleanup);
574 
575     LocalUResourceBundlePointer rb(ures_openDirect(nullptr, "supplementalData", &status));
576     if (U_FAILURE(status)) { return; }
577 
578     AllowedHourFormatsSink sink;
579     // TODO: Currently in the enumeration each table allocates a new array.
580     // Try to reduce the number of memory allocations. Consider storing a
581     // UVector32 with the concatenation of all of the sub-arrays, put the start index
582     // into the hashmap, store 6 single-value sub-arrays right at the beginning of the
583     // vector (at index enum*2) for easy data sharing, copy sub-arrays into runtime
584     // object. Remember to clean up the vector, too.
585     ures_getAllItemsWithFallback(rb.getAlias(), "timeData", sink, status);
586 }
587 
getAllowedHourFormats(const Locale & locale,UErrorCode & status)588 void DateTimePatternGenerator::getAllowedHourFormats(const Locale &locale, UErrorCode &status) {
589     if (U_FAILURE(status)) { return; }
590     const char *localeID = locale.getName();
591     char maxLocaleID[ULOC_FULLNAME_CAPACITY];
592     int32_t length = uloc_addLikelySubtags(localeID, maxLocaleID, ULOC_FULLNAME_CAPACITY, &status);
593     if (U_FAILURE(status)) {
594         return;
595     } else if (length == ULOC_FULLNAME_CAPACITY) {  // no room for NUL
596         status = U_BUFFER_OVERFLOW_ERROR;
597         return;
598     }
599     Locale maxLocale = Locale(maxLocaleID);
600 
601     const char *country = maxLocale.getCountry();
602     if (*country == '\0') { country = "001"; }
603     const char *language = maxLocale.getLanguage();
604 
605     CharString langCountry;
606     langCountry.append(language, static_cast<int32_t>(uprv_strlen(language)), status);
607     langCountry.append('_', status);
608     langCountry.append(country, static_cast<int32_t>(uprv_strlen(country)), status);
609 
610     int32_t *allowedFormats;
611     allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, langCountry.data());
612     if (allowedFormats == nullptr) {
613         allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, const_cast<char *>(country));
614     }
615 
616     if (allowedFormats != nullptr) {  // Lookup is successful
617         for (int32_t i = 0; i < UPRV_LENGTHOF(fAllowedHourFormats); ++i) {
618             fAllowedHourFormats[i] = allowedFormats[i];
619             if (allowedFormats[i] == ALLOWED_HOUR_FORMAT_UNKNOWN) {
620                 break;
621             }
622         }
623     } else {  // Lookup failed, twice
624         fAllowedHourFormats[0] = ALLOWED_HOUR_FORMAT_H;
625         fAllowedHourFormats[1] = ALLOWED_HOUR_FORMAT_UNKNOWN;
626     }
627 }
628 
629 UnicodeString
getSkeleton(const UnicodeString & pattern,UErrorCode &)630 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode&
631 /*status*/) {
632     FormatParser fp2;
633     DateTimeMatcher matcher;
634     PtnSkeleton localSkeleton;
635     matcher.set(pattern, &fp2, localSkeleton);
636     return localSkeleton.getSkeleton();
637 }
638 
639 UnicodeString
staticGetSkeleton(const UnicodeString & pattern,UErrorCode &)640 DateTimePatternGenerator::staticGetSkeleton(
641         const UnicodeString& pattern, UErrorCode& /*status*/) {
642     FormatParser fp;
643     DateTimeMatcher matcher;
644     PtnSkeleton localSkeleton;
645     matcher.set(pattern, &fp, localSkeleton);
646     return localSkeleton.getSkeleton();
647 }
648 
649 UnicodeString
getBaseSkeleton(const UnicodeString & pattern,UErrorCode &)650 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) {
651     FormatParser fp2;
652     DateTimeMatcher matcher;
653     PtnSkeleton localSkeleton;
654     matcher.set(pattern, &fp2, localSkeleton);
655     return localSkeleton.getBaseSkeleton();
656 }
657 
658 UnicodeString
staticGetBaseSkeleton(const UnicodeString & pattern,UErrorCode &)659 DateTimePatternGenerator::staticGetBaseSkeleton(
660         const UnicodeString& pattern, UErrorCode& /*status*/) {
661     FormatParser fp;
662     DateTimeMatcher matcher;
663     PtnSkeleton localSkeleton;
664     matcher.set(pattern, &fp, localSkeleton);
665     return localSkeleton.getBaseSkeleton();
666 }
667 
668 void
addICUPatterns(const Locale & locale,UErrorCode & status)669 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) {
670     if (U_FAILURE(status)) { return; }
671     UnicodeString dfPattern;
672     UnicodeString conflictingString;
673     DateFormat* df;
674 
675     // Load with ICU patterns
676     for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) {
677         DateFormat::EStyle style = (DateFormat::EStyle)i;
678         df = DateFormat::createDateInstance(style, locale);
679         SimpleDateFormat* sdf;
680         if (df != nullptr && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != nullptr) {
681             sdf->toPattern(dfPattern);
682             addPattern(dfPattern, FALSE, conflictingString, status);
683         }
684         // TODO Maybe we should return an error when the date format isn't simple.
685         delete df;
686         if (U_FAILURE(status)) { return; }
687 
688         df = DateFormat::createTimeInstance(style, locale);
689         if (df != nullptr && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != nullptr) {
690             sdf->toPattern(dfPattern);
691             addPattern(dfPattern, FALSE, conflictingString, status);
692 
693             // TODO: C++ and Java are inconsistent (see #12568).
694             // C++ uses MEDIUM, but Java uses SHORT.
695             if ( i==DateFormat::kShort && !dfPattern.isEmpty() ) {
696                 consumeShortTimePattern(dfPattern, status);
697             }
698         }
699         // TODO Maybe we should return an error when the date format isn't simple.
700         delete df;
701         if (U_FAILURE(status)) { return; }
702     }
703 }
704 
705 void
hackTimes(const UnicodeString & hackPattern,UErrorCode & status)706 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status)  {
707     UnicodeString conflictingString;
708 
709     fp->set(hackPattern);
710     UnicodeString mmss;
711     UBool gotMm=FALSE;
712     for (int32_t i=0; i<fp->itemNumber; ++i) {
713         UnicodeString field = fp->items[i];
714         if ( fp->isQuoteLiteral(field) ) {
715             if ( gotMm ) {
716                UnicodeString quoteLiteral;
717                fp->getQuoteLiteral(quoteLiteral, &i);
718                mmss += quoteLiteral;
719             }
720         }
721         else {
722             if (fp->isPatternSeparator(field) && gotMm) {
723                 mmss+=field;
724             }
725             else {
726                 UChar ch=field.charAt(0);
727                 if (ch==LOW_M) {
728                     gotMm=TRUE;
729                     mmss+=field;
730                 }
731                 else {
732                     if (ch==LOW_S) {
733                         if (!gotMm) {
734                             break;
735                         }
736                         mmss+= field;
737                         addPattern(mmss, FALSE, conflictingString, status);
738                         break;
739                     }
740                     else {
741                         if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) {
742                             break;
743                         }
744                     }
745                 }
746             }
747         }
748     }
749 }
750 
751 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY)
752 
753 static const UChar hourFormatChars[] = { CAP_H, LOW_H, CAP_K, LOW_K, 0 }; // HhKk, the hour format characters
754 
755 void
getCalendarTypeToUse(const Locale & locale,CharString & destination,UErrorCode & err)756 DateTimePatternGenerator::getCalendarTypeToUse(const Locale& locale, CharString& destination, UErrorCode& err) {
757     destination.clear().append(DT_DateTimeGregorianTag, -1, err); // initial default
758     if ( U_SUCCESS(err) ) {
759         char localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
760         // obtain a locale that always has the calendar key value that should be used
761         ures_getFunctionalEquivalent(
762             localeWithCalendarKey,
763             ULOC_LOCALE_IDENTIFIER_CAPACITY,
764             nullptr,
765             "calendar",
766             "calendar",
767             locale.getName(),
768             nullptr,
769             FALSE,
770             &err);
771         if (U_FAILURE(err)) { return; }
772         localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination
773         // now get the calendar key value from that locale
774         char calendarType[ULOC_KEYWORDS_CAPACITY];
775         int32_t calendarTypeLen = uloc_getKeywordValue(
776             localeWithCalendarKey,
777             "calendar",
778             calendarType,
779             ULOC_KEYWORDS_CAPACITY,
780             &err);
781         if (U_FAILURE(err)) { return; }
782         if (calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
783             destination.clear().append(calendarType, -1, err);
784             if (U_FAILURE(err)) { return; }
785         }
786         err = U_ZERO_ERROR;
787     }
788 }
789 
790 void
consumeShortTimePattern(const UnicodeString & shortTimePattern,UErrorCode & status)791 DateTimePatternGenerator::consumeShortTimePattern(const UnicodeString& shortTimePattern,
792         UErrorCode& status) {
793     if (U_FAILURE(status)) { return; }
794     // set fDefaultHourFormatChar to the hour format character from this pattern
795     int32_t tfIdx, tfLen = shortTimePattern.length();
796     UBool ignoreChars = FALSE;
797     for (tfIdx = 0; tfIdx < tfLen; tfIdx++) {
798         UChar tfChar = shortTimePattern.charAt(tfIdx);
799         if ( tfChar == SINGLE_QUOTE ) {
800             ignoreChars = !ignoreChars; // toggle (handle quoted literals & '' for single quote)
801         } else if ( !ignoreChars && u_strchr(hourFormatChars, tfChar) != nullptr ) {
802             fDefaultHourFormatChar = tfChar;
803             break;
804         }
805     }
806 
807     // HACK for hh:ss
808     hackTimes(shortTimePattern, status);
809 }
810 
811 struct DateTimePatternGenerator::AppendItemFormatsSink : public ResourceSink {
812 
813     // Destination for data, modified via setters.
814     DateTimePatternGenerator& dtpg;
815 
AppendItemFormatsSinkDateTimePatternGenerator::AppendItemFormatsSink816     AppendItemFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
817     virtual ~AppendItemFormatsSink();
818 
putDateTimePatternGenerator::AppendItemFormatsSink819     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
820             UErrorCode &errorCode) {
821         ResourceTable itemsTable = value.getTable(errorCode);
822         if (U_FAILURE(errorCode)) { return; }
823         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
824             UDateTimePatternField field = dtpg.getAppendFormatNumber(key);
825             if (field == UDATPG_FIELD_COUNT) { continue; }
826             const UnicodeString& valueStr = value.getUnicodeString(errorCode);
827             if (dtpg.getAppendItemFormat(field).isEmpty() && !valueStr.isEmpty()) {
828                 dtpg.setAppendItemFormat(field, valueStr);
829             }
830         }
831     }
832 
fillInMissingDateTimePatternGenerator::AppendItemFormatsSink833     void fillInMissing() {
834         UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, UPRV_LENGTHOF(UDATPG_ItemFormat)-1);  // Read-only alias.
835         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
836             UDateTimePatternField field = (UDateTimePatternField)i;
837             if (dtpg.getAppendItemFormat(field).isEmpty()) {
838                 dtpg.setAppendItemFormat(field, defaultItemFormat);
839             }
840         }
841     }
842 };
843 
844 struct DateTimePatternGenerator::AppendItemNamesSink : public ResourceSink {
845 
846     // Destination for data, modified via setters.
847     DateTimePatternGenerator& dtpg;
848 
AppendItemNamesSinkDateTimePatternGenerator::AppendItemNamesSink849     AppendItemNamesSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
850     virtual ~AppendItemNamesSink();
851 
putDateTimePatternGenerator::AppendItemNamesSink852     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
853             UErrorCode &errorCode) {
854         ResourceTable itemsTable = value.getTable(errorCode);
855         if (U_FAILURE(errorCode)) { return; }
856         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
857             UDateTimePGDisplayWidth width;
858             UDateTimePatternField field = dtpg.getFieldAndWidthIndices(key, &width);
859             if (field == UDATPG_FIELD_COUNT) { continue; }
860             ResourceTable detailsTable = value.getTable(errorCode);
861             if (U_FAILURE(errorCode)) { return; }
862             for (int32_t j = 0; detailsTable.getKeyAndValue(j, key, value); ++j) {
863                 if (uprv_strcmp(key, "dn") != 0) { continue; }
864                 const UnicodeString& valueStr = value.getUnicodeString(errorCode);
865                 if (dtpg.getFieldDisplayName(field,width).isEmpty() && !valueStr.isEmpty()) {
866                     dtpg.setFieldDisplayName(field,width,valueStr);
867                 }
868                 break;
869             }
870         }
871     }
872 
fillInMissingDateTimePatternGenerator::AppendItemNamesSink873     void fillInMissing() {
874         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
875             UnicodeString& valueStr = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, UDATPG_WIDE);
876             if (valueStr.isEmpty()) {
877                 valueStr = CAP_F;
878                 U_ASSERT(i < 20);
879                 if (i < 10) {
880                     // F0, F1, ..., F9
881                     valueStr += (UChar)(i+0x30);
882                 } else {
883                     // F10, F11, ...
884                     valueStr += (UChar)0x31;
885                     valueStr += (UChar)(i-10 + 0x30);
886                 }
887                 // NUL-terminate for the C API.
888                 valueStr.getTerminatedBuffer();
889             }
890             for (int32_t j = 1; j < UDATPG_WIDTH_COUNT; j++) {
891                 UnicodeString& valueStr2 = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)j);
892                 if (valueStr2.isEmpty()) {
893                     valueStr2 = dtpg.getFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)(j-1));
894                 }
895             }
896         }
897     }
898 };
899 
900 struct DateTimePatternGenerator::AvailableFormatsSink : public ResourceSink {
901 
902     // Destination for data, modified via setters.
903     DateTimePatternGenerator& dtpg;
904 
905     // Temporary variable, required for calling addPatternWithSkeleton.
906     UnicodeString conflictingPattern;
907 
AvailableFormatsSinkDateTimePatternGenerator::AvailableFormatsSink908     AvailableFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
909     virtual ~AvailableFormatsSink();
910 
putDateTimePatternGenerator::AvailableFormatsSink911     virtual void put(const char *key, ResourceValue &value, UBool isRoot,
912             UErrorCode &errorCode) {
913         ResourceTable itemsTable = value.getTable(errorCode);
914         if (U_FAILURE(errorCode)) { return; }
915         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
916             const UnicodeString formatKey(key, -1, US_INV);
917             if (!dtpg.isAvailableFormatSet(formatKey) ) {
918                 dtpg.setAvailableFormat(formatKey, errorCode);
919                 // Add pattern with its associated skeleton. Override any duplicate
920                 // derived from std patterns, but not a previous availableFormats entry:
921                 const UnicodeString& formatValue = value.getUnicodeString(errorCode);
922                 conflictingPattern.remove();
923                 dtpg.addPatternWithSkeleton(formatValue, &formatKey, !isRoot, conflictingPattern, errorCode);
924             }
925         }
926     }
927 };
928 
929 // Virtual destructors must be defined out of line.
~AppendItemFormatsSink()930 DateTimePatternGenerator::AppendItemFormatsSink::~AppendItemFormatsSink() {}
~AppendItemNamesSink()931 DateTimePatternGenerator::AppendItemNamesSink::~AppendItemNamesSink() {}
~AvailableFormatsSink()932 DateTimePatternGenerator::AvailableFormatsSink::~AvailableFormatsSink() {}
933 
934 void
addCLDRData(const Locale & locale,UErrorCode & errorCode)935 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& errorCode) {
936     if (U_FAILURE(errorCode)) { return; }
937     UnicodeString rbPattern, value, field;
938     CharString path;
939 
940     LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &errorCode));
941     if (U_FAILURE(errorCode)) { return; }
942 
943     CharString calendarTypeToUse; // to be filled in with the type to use, if all goes well
944     getCalendarTypeToUse(locale, calendarTypeToUse, errorCode);
945     if (U_FAILURE(errorCode)) { return; }
946 
947     // Local err to ignore resource not found exceptions
948     UErrorCode err = U_ZERO_ERROR;
949 
950     // Load append item formats.
951     AppendItemFormatsSink appendItemFormatsSink(*this);
952     path.clear()
953         .append(DT_DateTimeCalendarTag, errorCode)
954         .append('/', errorCode)
955         .append(calendarTypeToUse, errorCode)
956         .append('/', errorCode)
957         .append(DT_DateTimeAppendItemsTag, errorCode); // i.e., calendar/xxx/appendItems
958     if (U_FAILURE(errorCode)) { return; }
959     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), appendItemFormatsSink, err);
960     appendItemFormatsSink.fillInMissing();
961 
962     // Load CLDR item names.
963     err = U_ZERO_ERROR;
964     AppendItemNamesSink appendItemNamesSink(*this);
965     ures_getAllItemsWithFallback(rb.getAlias(), DT_DateTimeFieldsTag, appendItemNamesSink, err);
966     appendItemNamesSink.fillInMissing();
967 
968     // Load the available formats from CLDR.
969     err = U_ZERO_ERROR;
970     initHashtable(errorCode);
971     if (U_FAILURE(errorCode)) { return; }
972     AvailableFormatsSink availableFormatsSink(*this);
973     path.clear()
974         .append(DT_DateTimeCalendarTag, errorCode)
975         .append('/', errorCode)
976         .append(calendarTypeToUse, errorCode)
977         .append('/', errorCode)
978         .append(DT_DateTimeAvailableFormatsTag, errorCode); // i.e., calendar/xxx/availableFormats
979     if (U_FAILURE(errorCode)) { return; }
980     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), availableFormatsSink, err);
981 }
982 
983 void
initHashtable(UErrorCode & err)984 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
985     if (U_FAILURE(err)) { return; }
986     if (fAvailableFormatKeyHash!=nullptr) {
987         return;
988     }
989     LocalPointer<Hashtable> hash(new Hashtable(FALSE, err), err);
990     if (U_SUCCESS(err)) {
991         fAvailableFormatKeyHash = hash.orphan();
992     }
993 }
994 
995 void
setAppendItemFormat(UDateTimePatternField field,const UnicodeString & value)996 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
997     appendItemFormats[field] = value;
998     // NUL-terminate for the C API.
999     appendItemFormats[field].getTerminatedBuffer();
1000 }
1001 
1002 const UnicodeString&
getAppendItemFormat(UDateTimePatternField field) const1003 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
1004     return appendItemFormats[field];
1005 }
1006 
1007 void
setAppendItemName(UDateTimePatternField field,const UnicodeString & value)1008 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
1009     setFieldDisplayName(field, UDATPG_WIDTH_APPENDITEM, value);
1010 }
1011 
1012 const UnicodeString&
getAppendItemName(UDateTimePatternField field) const1013 DateTimePatternGenerator::getAppendItemName(UDateTimePatternField field) const {
1014     return fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1015 }
1016 
1017 void
setFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width,const UnicodeString & value)1018 DateTimePatternGenerator::setFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width, const UnicodeString& value) {
1019     fieldDisplayNames[field][width] = value;
1020     // NUL-terminate for the C API.
1021     fieldDisplayNames[field][width].getTerminatedBuffer();
1022 }
1023 
1024 UnicodeString
getFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width) const1025 DateTimePatternGenerator::getFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) const {
1026     return fieldDisplayNames[field][width];
1027 }
1028 
1029 UnicodeString&
getMutableFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width)1030 DateTimePatternGenerator::getMutableFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) {
1031     return fieldDisplayNames[field][width];
1032 }
1033 
1034 void
getAppendName(UDateTimePatternField field,UnicodeString & value)1035 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
1036     value = SINGLE_QUOTE;
1037     value += fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1038     value += SINGLE_QUOTE;
1039 }
1040 
1041 UnicodeString
getBestPattern(const UnicodeString & patternForm,UErrorCode & status)1042 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
1043     return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
1044 }
1045 
1046 UnicodeString
getBestPattern(const UnicodeString & patternForm,UDateTimePatternMatchOptions options,UErrorCode & status)1047 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
1048     if (U_FAILURE(status)) {
1049         return UnicodeString();
1050     }
1051     if (U_FAILURE(internalErrorCode)) {
1052         status = internalErrorCode;
1053         return UnicodeString();
1054     }
1055     const UnicodeString *bestPattern = nullptr;
1056     UnicodeString dtFormat;
1057     UnicodeString resultPattern;
1058     int32_t flags = kDTPGNoFlags;
1059 
1060     int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
1061     int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
1062 
1063     // Replace hour metacharacters 'j', 'C' and 'J', set flags as necessary
1064     UnicodeString patternFormMapped = mapSkeletonMetacharacters(patternForm, &flags, status);
1065     if (U_FAILURE(status)) {
1066         return UnicodeString();
1067     }
1068 
1069     resultPattern.remove();
1070     dtMatcher->set(patternFormMapped, fp);
1071     const PtnSkeleton* specifiedSkeleton = nullptr;
1072     bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, status, &specifiedSkeleton);
1073     if (U_FAILURE(status)) {
1074         return UnicodeString();
1075     }
1076 
1077     if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
1078         resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, flags, options);
1079 
1080         return resultPattern;
1081     }
1082     int32_t neededFields = dtMatcher->getFieldMask();
1083     UnicodeString datePattern=getBestAppending(neededFields & dateMask, flags, status, options);
1084     UnicodeString timePattern=getBestAppending(neededFields & timeMask, flags, status, options);
1085     if (U_FAILURE(status)) {
1086         return UnicodeString();
1087     }
1088     if (datePattern.length()==0) {
1089         if (timePattern.length()==0) {
1090             resultPattern.remove();
1091         }
1092         else {
1093             return timePattern;
1094         }
1095     }
1096     if (timePattern.length()==0) {
1097         return datePattern;
1098     }
1099     resultPattern.remove();
1100     status = U_ZERO_ERROR;
1101     dtFormat=getDateTimeFormat();
1102     SimpleFormatter(dtFormat, 2, 2, status).format(timePattern, datePattern, resultPattern, status);
1103     return resultPattern;
1104 }
1105 
1106 /*
1107  * Map a skeleton that may have metacharacters jJC to one without, by replacing
1108  * the metacharacters with locale-appropriate fields of h/H/k/K and of a/b/B
1109  * (depends on fDefaultHourFormatChar and fAllowedHourFormats being set, which in
1110  * turn depends on initData having been run). This method also updates the flags
1111  * as necessary. Returns the updated skeleton.
1112  */
1113 UnicodeString
mapSkeletonMetacharacters(const UnicodeString & patternForm,int32_t * flags,UErrorCode & status)1114 DateTimePatternGenerator::mapSkeletonMetacharacters(const UnicodeString& patternForm, int32_t* flags, UErrorCode& status) {
1115     UnicodeString patternFormMapped;
1116     patternFormMapped.remove();
1117     UBool inQuoted = FALSE;
1118     int32_t patPos, patLen = patternForm.length();
1119     for (patPos = 0; patPos < patLen; patPos++) {
1120         UChar patChr = patternForm.charAt(patPos);
1121         if (patChr == SINGLE_QUOTE) {
1122             inQuoted = !inQuoted;
1123         } else if (!inQuoted) {
1124             // Handle special mappings for 'j' and 'C' in which fields lengths
1125             // 1,3,5 => hour field length 1
1126             // 2,4,6 => hour field length 2
1127             // 1,2 => abbreviated dayPeriod (field length 1..3)
1128             // 3,4 => long dayPeriod (field length 4)
1129             // 5,6 => narrow dayPeriod (field length 5)
1130             if (patChr == LOW_J || patChr == CAP_C) {
1131                 int32_t extraLen = 0; // 1 less than total field length
1132                 while (patPos+1 < patLen && patternForm.charAt(patPos+1)==patChr) {
1133                     extraLen++;
1134                     patPos++;
1135                 }
1136                 int32_t hourLen = 1 + (extraLen & 1);
1137                 int32_t dayPeriodLen = (extraLen < 2)? 1: 3 + (extraLen >> 1);
1138                 UChar hourChar = LOW_H;
1139                 UChar dayPeriodChar = LOW_A;
1140                 if (patChr == LOW_J) {
1141                     hourChar = fDefaultHourFormatChar;
1142                 } else {
1143                     AllowedHourFormat preferred;
1144                     if (fAllowedHourFormats[0] != ALLOWED_HOUR_FORMAT_UNKNOWN) {
1145                         preferred = (AllowedHourFormat)fAllowedHourFormats[0];
1146                     } else {
1147                         status = U_INVALID_FORMAT_ERROR;
1148                         return UnicodeString();
1149                     }
1150                     if (preferred == ALLOWED_HOUR_FORMAT_H || preferred == ALLOWED_HOUR_FORMAT_HB || preferred == ALLOWED_HOUR_FORMAT_Hb) {
1151                         hourChar = CAP_H;
1152                     }
1153                     // in #13183 just add b/B to skeleton, no longer need to set special flags
1154                     if (preferred == ALLOWED_HOUR_FORMAT_HB || preferred == ALLOWED_HOUR_FORMAT_hB) {
1155                         dayPeriodChar = CAP_B;
1156                     } else if (preferred == ALLOWED_HOUR_FORMAT_Hb || preferred == ALLOWED_HOUR_FORMAT_hb) {
1157                         dayPeriodChar = LOW_B;
1158                     }
1159                 }
1160                 if (hourChar==CAP_H || hourChar==LOW_K) {
1161                     dayPeriodLen = 0;
1162                 }
1163                 while (dayPeriodLen-- > 0) {
1164                     patternFormMapped.append(dayPeriodChar);
1165                 }
1166                 while (hourLen-- > 0) {
1167                     patternFormMapped.append(hourChar);
1168                 }
1169             } else if (patChr == CAP_J) {
1170                 // Get pattern for skeleton with H, then replace H or k
1171                 // with fDefaultHourFormatChar (if different)
1172                 patternFormMapped.append(CAP_H);
1173                 *flags |= kDTPGSkeletonUsesCapJ;
1174             } else {
1175                 patternFormMapped.append(patChr);
1176             }
1177         }
1178     }
1179     return patternFormMapped;
1180 }
1181 
1182 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UErrorCode & status)1183 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1184                                             const UnicodeString& skeleton,
1185                                             UErrorCode& status) {
1186     return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
1187 }
1188 
1189 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UDateTimePatternMatchOptions options,UErrorCode & status)1190 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1191                                             const UnicodeString& skeleton,
1192                                             UDateTimePatternMatchOptions options,
1193                                             UErrorCode& status) {
1194     if (U_FAILURE(status)) {
1195         return UnicodeString();
1196     }
1197     if (U_FAILURE(internalErrorCode)) {
1198         status = internalErrorCode;
1199         return UnicodeString();
1200     }
1201     dtMatcher->set(skeleton, fp);
1202     UnicodeString result = adjustFieldTypes(pattern, nullptr, kDTPGNoFlags, options);
1203     return result;
1204 }
1205 
1206 void
setDecimal(const UnicodeString & newDecimal)1207 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
1208     this->decimal = newDecimal;
1209     // NUL-terminate for the C API.
1210     this->decimal.getTerminatedBuffer();
1211 }
1212 
1213 const UnicodeString&
getDecimal() const1214 DateTimePatternGenerator::getDecimal() const {
1215     return decimal;
1216 }
1217 
1218 void
addCanonicalItems(UErrorCode & status)1219 DateTimePatternGenerator::addCanonicalItems(UErrorCode& status) {
1220     if (U_FAILURE(status)) { return; }
1221     UnicodeString  conflictingPattern;
1222 
1223     for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
1224         if (Canonical_Items[i] > 0) {
1225             addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
1226         }
1227         if (U_FAILURE(status)) { return; }
1228     }
1229 }
1230 
1231 void
setDateTimeFormat(const UnicodeString & dtFormat)1232 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
1233     dateTimeFormat = dtFormat;
1234     // NUL-terminate for the C API.
1235     dateTimeFormat.getTerminatedBuffer();
1236 }
1237 
1238 const UnicodeString&
getDateTimeFormat() const1239 DateTimePatternGenerator::getDateTimeFormat() const {
1240     return dateTimeFormat;
1241 }
1242 
1243 void
setDateTimeFromCalendar(const Locale & locale,UErrorCode & status)1244 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
1245     if (U_FAILURE(status)) { return; }
1246 
1247     const UChar *resStr;
1248     int32_t resStrLen = 0;
1249 
1250     LocalPointer<Calendar> fCalendar(Calendar::createInstance(locale, status), status);
1251     if (U_FAILURE(status)) { return; }
1252 
1253     LocalUResourceBundlePointer calData(ures_open(nullptr, locale.getBaseName(), &status));
1254     if (U_FAILURE(status)) { return; }
1255     ures_getByKey(calData.getAlias(), DT_DateTimeCalendarTag, calData.getAlias(), &status);
1256     if (U_FAILURE(status)) { return; }
1257 
1258     LocalUResourceBundlePointer dateTimePatterns;
1259     if (fCalendar->getType() != nullptr && *fCalendar->getType() != '\0'
1260             && uprv_strcmp(fCalendar->getType(), DT_DateTimeGregorianTag) != 0) {
1261         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), fCalendar->getType(),
1262                                                                 nullptr, &status));
1263         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1264                                   dateTimePatterns.getAlias(), &status);
1265     }
1266 
1267     if (dateTimePatterns.isNull() || status == U_MISSING_RESOURCE_ERROR) {
1268         status = U_ZERO_ERROR;
1269         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), DT_DateTimeGregorianTag,
1270                                                                 dateTimePatterns.orphan(), &status));
1271         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1272                                   dateTimePatterns.getAlias(), &status);
1273     }
1274     if (U_FAILURE(status)) { return; }
1275 
1276     if (ures_getSize(dateTimePatterns.getAlias()) <= DateFormat::kDateTime)
1277     {
1278         status = U_INVALID_FORMAT_ERROR;
1279         return;
1280     }
1281     resStr = ures_getStringByIndex(dateTimePatterns.getAlias(), (int32_t)DateFormat::kDateTime, &resStrLen, &status);
1282     setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
1283 }
1284 
1285 void
setDecimalSymbols(const Locale & locale,UErrorCode & status)1286 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
1287     DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
1288     if(U_SUCCESS(status)) {
1289         decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
1290         // NUL-terminate for the C API.
1291         decimal.getTerminatedBuffer();
1292     }
1293 }
1294 
1295 UDateTimePatternConflict
addPattern(const UnicodeString & pattern,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1296 DateTimePatternGenerator::addPattern(
1297     const UnicodeString& pattern,
1298     UBool override,
1299     UnicodeString &conflictingPattern,
1300     UErrorCode& status)
1301 {
1302     if (U_FAILURE(internalErrorCode)) {
1303         status = internalErrorCode;
1304         return UDATPG_NO_CONFLICT;
1305     }
1306 
1307     return addPatternWithSkeleton(pattern, nullptr, override, conflictingPattern, status);
1308 }
1309 
1310 // For DateTimePatternGenerator::addPatternWithSkeleton -
1311 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
1312 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
1313 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
1314 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
1315 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
1316 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
1317 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
1318 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
1319 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
1320 UDateTimePatternConflict
addPatternWithSkeleton(const UnicodeString & pattern,const UnicodeString * skeletonToUse,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1321 DateTimePatternGenerator::addPatternWithSkeleton(
1322     const UnicodeString& pattern,
1323     const UnicodeString* skeletonToUse,
1324     UBool override,
1325     UnicodeString& conflictingPattern,
1326     UErrorCode& status)
1327 {
1328     if (U_FAILURE(internalErrorCode)) {
1329         status = internalErrorCode;
1330         return UDATPG_NO_CONFLICT;
1331     }
1332 
1333     UnicodeString basePattern;
1334     PtnSkeleton   skeleton;
1335     UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
1336 
1337     DateTimeMatcher matcher;
1338     if ( skeletonToUse == nullptr ) {
1339         matcher.set(pattern, fp, skeleton);
1340         matcher.getBasePattern(basePattern);
1341     } else {
1342         matcher.set(*skeletonToUse, fp, skeleton); // no longer trims skeleton fields to max len 3, per #7930
1343         matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
1344     }
1345     // We only care about base conflicts - and replacing the pattern associated with a base - if:
1346     // 1. the conflicting previous base pattern did *not* have an explicit skeleton; in that case the previous
1347     // base + pattern combination was derived from either (a) a canonical item, (b) a standard format, or
1348     // (c) a pattern specified programmatically with a previous call to addPattern (which would only happen
1349     // if we are getting here from a subsequent call to addPattern).
1350     // 2. a skeleton is specified for the current pattern, but override=false; in that case we are checking
1351     // availableFormats items from root, which should not override any previous entry with the same base.
1352     UBool entryHadSpecifiedSkeleton;
1353     const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
1354     if (duplicatePattern != nullptr && (!entryHadSpecifiedSkeleton || (skeletonToUse != nullptr && !override))) {
1355         conflictingStatus = UDATPG_BASE_CONFLICT;
1356         conflictingPattern = *duplicatePattern;
1357         if (!override) {
1358             return conflictingStatus;
1359         }
1360     }
1361     // The only time we get here with override=true and skeletonToUse!=null is when adding availableFormats
1362     // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with
1363     // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for
1364     // the previously-specified conflicting item.
1365     const PtnSkeleton* entrySpecifiedSkeleton = nullptr;
1366     duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
1367     if (duplicatePattern != nullptr ) {
1368         conflictingStatus = UDATPG_CONFLICT;
1369         conflictingPattern = *duplicatePattern;
1370         if (!override || (skeletonToUse != nullptr && entrySpecifiedSkeleton != nullptr)) {
1371             return conflictingStatus;
1372         }
1373     }
1374     patternMap->add(basePattern, skeleton, pattern, skeletonToUse != nullptr, status);
1375     if(U_FAILURE(status)) {
1376         return conflictingStatus;
1377     }
1378 
1379     return UDATPG_NO_CONFLICT;
1380 }
1381 
1382 
1383 UDateTimePatternField
getAppendFormatNumber(const char * field) const1384 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
1385     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1386         if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
1387             return (UDateTimePatternField)i;
1388         }
1389     }
1390     return UDATPG_FIELD_COUNT;
1391 }
1392 
1393 UDateTimePatternField
getFieldAndWidthIndices(const char * key,UDateTimePGDisplayWidth * widthP) const1394 DateTimePatternGenerator::getFieldAndWidthIndices(const char* key, UDateTimePGDisplayWidth* widthP) const {
1395     char cldrFieldKey[UDATPG_FIELD_KEY_MAX + 1];
1396     uprv_strncpy(cldrFieldKey, key, UDATPG_FIELD_KEY_MAX);
1397     cldrFieldKey[UDATPG_FIELD_KEY_MAX]=0; // ensure termination
1398     *widthP = UDATPG_WIDE;
1399     char* hyphenPtr = uprv_strchr(cldrFieldKey, '-');
1400     if (hyphenPtr) {
1401         for (int32_t i=UDATPG_WIDTH_COUNT-1; i>0; --i) {
1402             if (uprv_strcmp(CLDR_FIELD_WIDTH[i], hyphenPtr)==0) {
1403                 *widthP=(UDateTimePGDisplayWidth)i;
1404                 break;
1405             }
1406         }
1407         *hyphenPtr = 0; // now delete width portion of key
1408     }
1409     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1410         if (uprv_strcmp(CLDR_FIELD_NAME[i],cldrFieldKey)==0) {
1411             return (UDateTimePatternField)i;
1412         }
1413     }
1414     return UDATPG_FIELD_COUNT;
1415 }
1416 
1417 const UnicodeString*
getBestRaw(DateTimeMatcher & source,int32_t includeMask,DistanceInfo * missingFields,UErrorCode & status,const PtnSkeleton ** specifiedSkeletonPtr)1418 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
1419                                      int32_t includeMask,
1420                                      DistanceInfo* missingFields,
1421                                      UErrorCode &status,
1422                                      const PtnSkeleton** specifiedSkeletonPtr) {
1423     int32_t bestDistance = 0x7fffffff;
1424     DistanceInfo tempInfo;
1425     const UnicodeString *bestPattern=nullptr;
1426     const PtnSkeleton* specifiedSkeleton=nullptr;
1427 
1428     PatternMapIterator it(status);
1429     if (U_FAILURE(status)) { return nullptr; }
1430 
1431     for (it.set(*patternMap); it.hasNext(); ) {
1432         DateTimeMatcher trial = it.next();
1433         if (trial.equals(skipMatcher)) {
1434             continue;
1435         }
1436         int32_t distance=source.getDistance(trial, includeMask, tempInfo);
1437         if (distance<bestDistance) {
1438             bestDistance=distance;
1439             bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
1440             missingFields->setTo(tempInfo);
1441             if (distance==0) {
1442                 break;
1443             }
1444         }
1445     }
1446 
1447     // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
1448     // then return it too. This generally happens when the caller needs to pass that skeleton
1449     // through to adjustFieldTypes so the latter can do a better job.
1450     if (bestPattern && specifiedSkeletonPtr) {
1451         *specifiedSkeletonPtr = specifiedSkeleton;
1452     }
1453     return bestPattern;
1454 }
1455 
1456 UnicodeString
adjustFieldTypes(const UnicodeString & pattern,const PtnSkeleton * specifiedSkeleton,int32_t flags,UDateTimePatternMatchOptions options)1457 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
1458                                            const PtnSkeleton* specifiedSkeleton,
1459                                            int32_t flags,
1460                                            UDateTimePatternMatchOptions options) {
1461     UnicodeString newPattern;
1462     fp->set(pattern);
1463     for (int32_t i=0; i < fp->itemNumber; i++) {
1464         UnicodeString field = fp->items[i];
1465         if ( fp->isQuoteLiteral(field) ) {
1466 
1467             UnicodeString quoteLiteral;
1468             fp->getQuoteLiteral(quoteLiteral, &i);
1469             newPattern += quoteLiteral;
1470         }
1471         else {
1472             if (fp->isPatternSeparator(field)) {
1473                 newPattern+=field;
1474                 continue;
1475             }
1476             int32_t canonicalIndex = fp->getCanonicalIndex(field);
1477             if (canonicalIndex < 0) {
1478                 newPattern+=field;
1479                 continue;  // don't adjust
1480             }
1481             const dtTypeElem *row = &dtTypes[canonicalIndex];
1482             int32_t typeValue = row->field;
1483 
1484             // handle day periods - with #13183, no longer need special handling here, integrated with normal types
1485 
1486             if ((flags & kDTPGFixFractionalSeconds) != 0 && typeValue == UDATPG_SECOND_FIELD) {
1487                 field += decimal;
1488                 dtMatcher->skeleton.original.appendFieldTo(UDATPG_FRACTIONAL_SECOND_FIELD, field);
1489             } else if (dtMatcher->skeleton.type[typeValue]!=0) {
1490                     // Here:
1491                     // - "reqField" is the field from the originally requested skeleton, with length
1492                     // "reqFieldLen".
1493                     // - "field" is the field from the found pattern.
1494                     //
1495                     // The adjusted field should consist of characters from the originally requested
1496                     // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1497                     // UDATPG_WEEKDAY_FIELD or UDATPG_YEAR_FIELD, in which case it should consist
1498                     // of characters from the  found pattern.
1499                     //
1500                     // The length of the adjusted field (adjFieldLen) should match that in the originally
1501                     // requested skeleton, except that in the following cases the length of the adjusted field
1502                     // should match that in the found pattern (i.e. the length of this pattern field should
1503                     // not be adjusted):
1504                     // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1505                     //    not set (ticket #7180). Note, we may want to implement a similar change for other
1506                     //    numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1507                     //    field length, but options bits can be used to override this.
1508                     // 2. There is a specified skeleton for the found pattern and one of the following is true:
1509                     //    a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1510                     //    b) The pattern field is numeric and the skeleton field is not, or vice versa.
1511 
1512                     UChar reqFieldChar = dtMatcher->skeleton.original.getFieldChar(typeValue);
1513                     int32_t reqFieldLen = dtMatcher->skeleton.original.getFieldLength(typeValue);
1514                     if (reqFieldChar == CAP_E && reqFieldLen < 3)
1515                         reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1516                     int32_t adjFieldLen = reqFieldLen;
1517                     if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1518                          (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1519                          (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1520                          adjFieldLen = field.length();
1521                     } else if (specifiedSkeleton) {
1522                         int32_t skelFieldLen = specifiedSkeleton->original.getFieldLength(typeValue);
1523                         UBool patFieldIsNumeric = (row->type > 0);
1524                         UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1525                         if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1526                             // don't adjust the field length in the found pattern
1527                             adjFieldLen = field.length();
1528                         }
1529                     }
1530                     UChar c = (typeValue!= UDATPG_HOUR_FIELD
1531                             && typeValue!= UDATPG_MONTH_FIELD
1532                             && typeValue!= UDATPG_WEEKDAY_FIELD
1533                             && (typeValue!= UDATPG_YEAR_FIELD || reqFieldChar==CAP_Y))
1534                             ? reqFieldChar
1535                             : field.charAt(0);
1536                     if (typeValue == UDATPG_HOUR_FIELD && (flags & kDTPGSkeletonUsesCapJ) != 0) {
1537                         c = fDefaultHourFormatChar;
1538                     }
1539                     field.remove();
1540                     for (int32_t j=adjFieldLen; j>0; --j) {
1541                         field += c;
1542                     }
1543             }
1544             newPattern+=field;
1545         }
1546     }
1547     return newPattern;
1548 }
1549 
1550 UnicodeString
getBestAppending(int32_t missingFields,int32_t flags,UErrorCode & status,UDateTimePatternMatchOptions options)1551 DateTimePatternGenerator::getBestAppending(int32_t missingFields, int32_t flags, UErrorCode &status, UDateTimePatternMatchOptions options) {
1552     if (U_FAILURE(status)) {
1553         return UnicodeString();
1554     }
1555     UnicodeString  resultPattern, tempPattern;
1556     const UnicodeString* tempPatternPtr;
1557     int32_t lastMissingFieldMask=0;
1558     if (missingFields!=0) {
1559         resultPattern=UnicodeString();
1560         const PtnSkeleton* specifiedSkeleton=nullptr;
1561         tempPatternPtr = getBestRaw(*dtMatcher, missingFields, distanceInfo, status, &specifiedSkeleton);
1562         if (U_FAILURE(status)) {
1563             return UnicodeString();
1564         }
1565         tempPattern = *tempPatternPtr;
1566         resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1567         if ( distanceInfo->missingFieldMask==0 ) {
1568             return resultPattern;
1569         }
1570         while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1571             if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1572                 break;  // cannot find the proper missing field
1573             }
1574             if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1575                 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1576                 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, flags | kDTPGFixFractionalSeconds, options);
1577                 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1578                 continue;
1579             }
1580             int32_t startingMask = distanceInfo->missingFieldMask;
1581             tempPatternPtr = getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, status, &specifiedSkeleton);
1582             if (U_FAILURE(status)) {
1583                 return UnicodeString();
1584             }
1585             tempPattern = *tempPatternPtr;
1586             tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1587             int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1588             int32_t topField=getTopBitNumber(foundMask);
1589 
1590             if (appendItemFormats[topField].length() != 0) {
1591                 UnicodeString appendName;
1592                 getAppendName((UDateTimePatternField)topField, appendName);
1593                 const UnicodeString *values[3] = {
1594                     &resultPattern,
1595                     &tempPattern,
1596                     &appendName
1597                 };
1598                 SimpleFormatter(appendItemFormats[topField], 2, 3, status).
1599                     formatAndReplace(values, 3, resultPattern, nullptr, 0, status);
1600             }
1601             lastMissingFieldMask = distanceInfo->missingFieldMask;
1602         }
1603     }
1604     return resultPattern;
1605 }
1606 
1607 int32_t
getTopBitNumber(int32_t foundMask) const1608 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) const {
1609     if ( foundMask==0 ) {
1610         return 0;
1611     }
1612     int32_t i=0;
1613     while (foundMask!=0) {
1614         foundMask >>=1;
1615         ++i;
1616     }
1617     if (i-1 >UDATPG_ZONE_FIELD) {
1618         return UDATPG_ZONE_FIELD;
1619     }
1620     else
1621         return i-1;
1622 }
1623 
1624 void
setAvailableFormat(const UnicodeString & key,UErrorCode & err)1625 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1626 {
1627     fAvailableFormatKeyHash->puti(key, 1, err);
1628 }
1629 
1630 UBool
isAvailableFormatSet(const UnicodeString & key) const1631 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1632     return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1633 }
1634 
1635 void
copyHashtable(Hashtable * other,UErrorCode & status)1636 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1637     if (other == nullptr || U_FAILURE(status)) {
1638         return;
1639     }
1640     if (fAvailableFormatKeyHash != nullptr) {
1641         delete fAvailableFormatKeyHash;
1642         fAvailableFormatKeyHash = nullptr;
1643     }
1644     initHashtable(status);
1645     if(U_FAILURE(status)){
1646         return;
1647     }
1648     int32_t pos = UHASH_FIRST;
1649     const UHashElement* elem = nullptr;
1650     // walk through the hash table and create a deep clone
1651     while((elem = other->nextElement(pos))!= nullptr){
1652         const UHashTok otherKeyTok = elem->key;
1653         UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1654         fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1655         if(U_FAILURE(status)){
1656             return;
1657         }
1658     }
1659 }
1660 
1661 StringEnumeration*
getSkeletons(UErrorCode & status) const1662 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1663     if (U_FAILURE(status)) {
1664         return nullptr;
1665     }
1666     if (U_FAILURE(internalErrorCode)) {
1667         status = internalErrorCode;
1668         return nullptr;
1669     }
1670     LocalPointer<StringEnumeration> skeletonEnumerator(
1671         new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status), status);
1672 
1673     return U_SUCCESS(status) ? skeletonEnumerator.orphan() : nullptr;
1674 }
1675 
1676 const UnicodeString&
getPatternForSkeleton(const UnicodeString & skeleton) const1677 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1678     PtnElem *curElem;
1679 
1680     if (skeleton.length() ==0) {
1681         return emptyString;
1682     }
1683     curElem = patternMap->getHeader(skeleton.charAt(0));
1684     while ( curElem != nullptr ) {
1685         if ( curElem->skeleton->getSkeleton()==skeleton ) {
1686             return curElem->pattern;
1687         }
1688         curElem = curElem->next.getAlias();
1689     }
1690     return emptyString;
1691 }
1692 
1693 StringEnumeration*
getBaseSkeletons(UErrorCode & status) const1694 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1695     if (U_FAILURE(status)) {
1696         return nullptr;
1697     }
1698     if (U_FAILURE(internalErrorCode)) {
1699         status = internalErrorCode;
1700         return nullptr;
1701     }
1702     LocalPointer<StringEnumeration> baseSkeletonEnumerator(
1703         new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status), status);
1704 
1705     return U_SUCCESS(status) ? baseSkeletonEnumerator.orphan() : nullptr;
1706 }
1707 
1708 StringEnumeration*
getRedundants(UErrorCode & status)1709 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1710     if (U_FAILURE(status)) { return nullptr; }
1711     if (U_FAILURE(internalErrorCode)) {
1712         status = internalErrorCode;
1713         return nullptr;
1714     }
1715     LocalPointer<StringEnumeration> output(new DTRedundantEnumeration(), status);
1716     if (U_FAILURE(status)) { return nullptr; }
1717     const UnicodeString *pattern;
1718     PatternMapIterator it(status);
1719     if (U_FAILURE(status)) { return nullptr; }
1720 
1721     for (it.set(*patternMap); it.hasNext(); ) {
1722         DateTimeMatcher current = it.next();
1723         pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1724         if ( isCanonicalItem(*pattern) ) {
1725             continue;
1726         }
1727         if ( skipMatcher == nullptr ) {
1728             skipMatcher = new DateTimeMatcher(current);
1729             if (skipMatcher == nullptr) {
1730                 status = U_MEMORY_ALLOCATION_ERROR;
1731                 return nullptr;
1732             }
1733         }
1734         else {
1735             *skipMatcher = current;
1736         }
1737         UnicodeString trial = getBestPattern(current.getPattern(), status);
1738         if (U_FAILURE(status)) { return nullptr; }
1739         if (trial == *pattern) {
1740             ((DTRedundantEnumeration *)output.getAlias())->add(*pattern, status);
1741             if (U_FAILURE(status)) { return nullptr; }
1742         }
1743         if (current.equals(skipMatcher)) {
1744             continue;
1745         }
1746     }
1747     return output.orphan();
1748 }
1749 
1750 UBool
isCanonicalItem(const UnicodeString & item) const1751 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1752     if ( item.length() != 1 ) {
1753         return FALSE;
1754     }
1755     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1756         if (item.charAt(0)==Canonical_Items[i]) {
1757             return TRUE;
1758         }
1759     }
1760     return FALSE;
1761 }
1762 
1763 
1764 DateTimePatternGenerator*
clone() const1765 DateTimePatternGenerator::clone() const {
1766     return new DateTimePatternGenerator(*this);
1767 }
1768 
PatternMap()1769 PatternMap::PatternMap() {
1770    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1771        boot[i] = nullptr;
1772    }
1773    isDupAllowed = TRUE;
1774 }
1775 
1776 void
copyFrom(const PatternMap & other,UErrorCode & status)1777 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1778     if (U_FAILURE(status)) {
1779         return;
1780     }
1781     this->isDupAllowed = other.isDupAllowed;
1782     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
1783         PtnElem *curElem, *otherElem, *prevElem=nullptr;
1784         otherElem = other.boot[bootIndex];
1785         while (otherElem != nullptr) {
1786             LocalPointer<PtnElem> newElem(new PtnElem(otherElem->basePattern, otherElem->pattern), status);
1787             if (U_FAILURE(status)) {
1788                 return; // out of memory
1789             }
1790             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(*(otherElem->skeleton)), status);
1791             if (U_FAILURE(status)) {
1792                 return; // out of memory
1793             }
1794             newElem->skeletonWasSpecified = otherElem->skeletonWasSpecified;
1795 
1796             // Release ownership from the LocalPointer of the PtnElem object.
1797             // The PtnElem will now be owned by either the boot (for the first entry in the linked-list)
1798             // or owned by the previous PtnElem object in the linked-list.
1799             curElem = newElem.orphan();
1800 
1801             if (this->boot[bootIndex] == nullptr) {
1802                 this->boot[bootIndex] = curElem;
1803             } else {
1804                 if (prevElem != nullptr) {
1805                     prevElem->next.adoptInstead(curElem);
1806                 } else {
1807                     U_ASSERT(false);
1808                 }
1809             }
1810             prevElem = curElem;
1811             otherElem = otherElem->next.getAlias();
1812         }
1813 
1814     }
1815 }
1816 
1817 PtnElem*
getHeader(UChar baseChar) const1818 PatternMap::getHeader(UChar baseChar) const {
1819     PtnElem* curElem;
1820 
1821     if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1822          curElem = boot[baseChar-CAP_A];
1823     }
1824     else {
1825         if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1826             curElem = boot[26+baseChar-LOW_A];
1827         }
1828         else {
1829             return nullptr;
1830         }
1831     }
1832     return curElem;
1833 }
1834 
~PatternMap()1835 PatternMap::~PatternMap() {
1836    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1837        if (boot[i] != nullptr ) {
1838            delete boot[i];
1839            boot[i] = nullptr;
1840        }
1841    }
1842 }  // PatternMap destructor
1843 
1844 void
add(const UnicodeString & basePattern,const PtnSkeleton & skeleton,const UnicodeString & value,UBool skeletonWasSpecified,UErrorCode & status)1845 PatternMap::add(const UnicodeString& basePattern,
1846                 const PtnSkeleton& skeleton,
1847                 const UnicodeString& value,// mapped pattern value
1848                 UBool skeletonWasSpecified,
1849                 UErrorCode &status) {
1850     UChar baseChar = basePattern.charAt(0);
1851     PtnElem *curElem, *baseElem;
1852     status = U_ZERO_ERROR;
1853 
1854     // the baseChar must be A-Z or a-z
1855     if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1856         baseElem = boot[baseChar-CAP_A];
1857     }
1858     else {
1859         if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1860             baseElem = boot[26+baseChar-LOW_A];
1861          }
1862          else {
1863              status = U_ILLEGAL_CHARACTER;
1864              return;
1865          }
1866     }
1867 
1868     if (baseElem == nullptr) {
1869         LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1870         if (U_FAILURE(status)) {
1871             return; // out of memory
1872         }
1873         newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1874         if (U_FAILURE(status)) {
1875             return; // out of memory
1876         }
1877         newElem->skeletonWasSpecified = skeletonWasSpecified;
1878         if (baseChar >= LOW_A) {
1879             boot[26 + (baseChar - LOW_A)] = newElem.orphan(); // the boot array now owns the PtnElem.
1880         }
1881         else {
1882             boot[baseChar - CAP_A] = newElem.orphan(); // the boot array now owns the PtnElem.
1883         }
1884     }
1885     if ( baseElem != nullptr ) {
1886         curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1887 
1888         if (curElem == nullptr) {
1889             // add new element to the list.
1890             curElem = baseElem;
1891             while( curElem -> next != nullptr )
1892             {
1893                 curElem = curElem->next.getAlias();
1894             }
1895 
1896             LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1897             if (U_FAILURE(status)) {
1898                 return; // out of memory
1899             }
1900             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1901             if (U_FAILURE(status)) {
1902                 return; // out of memory
1903             }
1904             newElem->skeletonWasSpecified = skeletonWasSpecified;
1905             curElem->next.adoptInstead(newElem.orphan());
1906             curElem = curElem->next.getAlias();
1907         }
1908         else {
1909             // Pattern exists in the list already.
1910             if ( !isDupAllowed ) {
1911                 return;
1912             }
1913             // Overwrite the value.
1914             curElem->pattern = value;
1915             // It was a bug that we were not doing the following previously,
1916             // though that bug hid other problems by making things partly work.
1917             curElem->skeletonWasSpecified = skeletonWasSpecified;
1918         }
1919     }
1920 }  // PatternMap::add
1921 
1922 // Find the pattern from the given basePattern string.
1923 const UnicodeString *
getPatternFromBasePattern(const UnicodeString & basePattern,UBool & skeletonWasSpecified) const1924 PatternMap::getPatternFromBasePattern(const UnicodeString& basePattern, UBool& skeletonWasSpecified) const { // key to search for
1925    PtnElem *curElem;
1926 
1927    if ((curElem=getHeader(basePattern.charAt(0)))==nullptr) {
1928        return nullptr;  // no match
1929    }
1930 
1931    do  {
1932        if ( basePattern.compare(curElem->basePattern)==0 ) {
1933           skeletonWasSpecified = curElem->skeletonWasSpecified;
1934           return &(curElem->pattern);
1935        }
1936        curElem = curElem->next.getAlias();
1937    } while (curElem != nullptr);
1938 
1939    return nullptr;
1940 }  // PatternMap::getFromBasePattern
1941 
1942 
1943 // Find the pattern from the given skeleton.
1944 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
1945 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
1946 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
1947 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
1948 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
1949 const UnicodeString *
getPatternFromSkeleton(const PtnSkeleton & skeleton,const PtnSkeleton ** specifiedSkeletonPtr) const1950 PatternMap::getPatternFromSkeleton(const PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) const { // key to search for
1951    PtnElem *curElem;
1952 
1953    if (specifiedSkeletonPtr) {
1954        *specifiedSkeletonPtr = nullptr;
1955    }
1956 
1957    // find boot entry
1958    UChar baseChar = skeleton.getFirstChar();
1959    if ((curElem=getHeader(baseChar))==nullptr) {
1960        return nullptr;  // no match
1961    }
1962 
1963    do  {
1964        UBool equal;
1965        if (specifiedSkeletonPtr != nullptr) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
1966            equal = curElem->skeleton->original == skeleton.original;
1967        } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
1968            equal = curElem->skeleton->baseOriginal == skeleton.baseOriginal;
1969        }
1970        if (equal) {
1971            if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
1972                *specifiedSkeletonPtr = curElem->skeleton.getAlias();
1973            }
1974            return &(curElem->pattern);
1975        }
1976        curElem = curElem->next.getAlias();
1977    } while (curElem != nullptr);
1978 
1979    return nullptr;
1980 }
1981 
1982 UBool
equals(const PatternMap & other) const1983 PatternMap::equals(const PatternMap& other) const {
1984     if ( this==&other ) {
1985         return TRUE;
1986     }
1987     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
1988         if (boot[bootIndex] == other.boot[bootIndex]) {
1989             continue;
1990         }
1991         if ((boot[bootIndex] == nullptr) || (other.boot[bootIndex] == nullptr)) {
1992             return FALSE;
1993         }
1994         PtnElem *otherElem = other.boot[bootIndex];
1995         PtnElem *myElem = boot[bootIndex];
1996         while ((otherElem != nullptr) || (myElem != nullptr)) {
1997             if ( myElem == otherElem ) {
1998                 break;
1999             }
2000             if ((otherElem == nullptr) || (myElem == nullptr)) {
2001                 return FALSE;
2002             }
2003             if ( (myElem->basePattern != otherElem->basePattern) ||
2004                  (myElem->pattern != otherElem->pattern) ) {
2005                 return FALSE;
2006             }
2007             if ((myElem->skeleton.getAlias() != otherElem->skeleton.getAlias()) &&
2008                 !myElem->skeleton->equals(*(otherElem->skeleton))) {
2009                 return FALSE;
2010             }
2011             myElem = myElem->next.getAlias();
2012             otherElem = otherElem->next.getAlias();
2013         }
2014     }
2015     return TRUE;
2016 }
2017 
2018 // find any key existing in the mapping table already.
2019 // return TRUE if there is an existing key, otherwise return FALSE.
2020 PtnElem*
getDuplicateElem(const UnicodeString & basePattern,const PtnSkeleton & skeleton,PtnElem * baseElem)2021 PatternMap::getDuplicateElem(
2022             const UnicodeString &basePattern,
2023             const PtnSkeleton &skeleton,
2024             PtnElem *baseElem) {
2025    PtnElem *curElem;
2026 
2027    if ( baseElem == nullptr ) {
2028          return nullptr;
2029    }
2030    else {
2031          curElem = baseElem;
2032    }
2033    do {
2034      if ( basePattern.compare(curElem->basePattern)==0 ) {
2035          UBool isEqual = TRUE;
2036          for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2037             if (curElem->skeleton->type[i] != skeleton.type[i] ) {
2038                 isEqual = FALSE;
2039                 break;
2040             }
2041         }
2042         if (isEqual) {
2043             return curElem;
2044         }
2045      }
2046      curElem = curElem->next.getAlias();
2047    } while( curElem != nullptr );
2048 
2049    // end of the list
2050    return nullptr;
2051 
2052 }  // PatternMap::getDuplicateElem
2053 
DateTimeMatcher(void)2054 DateTimeMatcher::DateTimeMatcher(void) {
2055 }
2056 
~DateTimeMatcher()2057 DateTimeMatcher::~DateTimeMatcher() {}
2058 
DateTimeMatcher(const DateTimeMatcher & other)2059 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
2060     copyFrom(other.skeleton);
2061 }
2062 
2063 
2064 void
set(const UnicodeString & pattern,FormatParser * fp)2065 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
2066     PtnSkeleton localSkeleton;
2067     return set(pattern, fp, localSkeleton);
2068 }
2069 
2070 void
set(const UnicodeString & pattern,FormatParser * fp,PtnSkeleton & skeletonResult)2071 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
2072     int32_t i;
2073     for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
2074         skeletonResult.type[i] = NONE;
2075     }
2076     skeletonResult.original.clear();
2077     skeletonResult.baseOriginal.clear();
2078     skeletonResult.addedDefaultDayPeriod = FALSE;
2079 
2080     fp->set(pattern);
2081     for (i=0; i < fp->itemNumber; i++) {
2082         const UnicodeString& value = fp->items[i];
2083         // don't skip 'a' anymore, dayPeriod handled specially below
2084 
2085         if ( fp->isQuoteLiteral(value) ) {
2086             UnicodeString quoteLiteral;
2087             fp->getQuoteLiteral(quoteLiteral, &i);
2088             continue;
2089         }
2090         int32_t canonicalIndex = fp->getCanonicalIndex(value);
2091         if (canonicalIndex < 0) {
2092             continue;
2093         }
2094         const dtTypeElem *row = &dtTypes[canonicalIndex];
2095         int32_t field = row->field;
2096         skeletonResult.original.populate(field, value);
2097         UChar repeatChar = row->patternChar;
2098         int32_t repeatCount = row->minLen;
2099         skeletonResult.baseOriginal.populate(field, repeatChar, repeatCount);
2100         int16_t subField = row->type;
2101         if (row->type > 0) {
2102             U_ASSERT(value.length() < INT16_MAX);
2103             subField += static_cast<int16_t>(value.length());
2104         }
2105         skeletonResult.type[field] = subField;
2106     }
2107     // #13183, handle special behavior for day period characters (a, b, B)
2108     if (!skeletonResult.original.isFieldEmpty(UDATPG_HOUR_FIELD)) {
2109         if (skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==LOW_H || skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==CAP_K) {
2110             // We have a skeleton with 12-hour-cycle format
2111             if (skeletonResult.original.isFieldEmpty(UDATPG_DAYPERIOD_FIELD)) {
2112                 // But we do not have a day period in the skeleton; add the default DAYPERIOD (currently "a")
2113                 for (i = 0; dtTypes[i].patternChar != 0; i++) {
2114                     if ( dtTypes[i].field == UDATPG_DAYPERIOD_FIELD ) {
2115                         // first entry for UDATPG_DAYPERIOD_FIELD
2116                         skeletonResult.original.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2117                         skeletonResult.baseOriginal.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2118                         skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = dtTypes[i].type;
2119                         skeletonResult.addedDefaultDayPeriod = TRUE;
2120                         break;
2121                     }
2122                 }
2123             }
2124         } else {
2125             // Skeleton has 24-hour-cycle hour format and has dayPeriod, delete dayPeriod (i.e. ignore it)
2126             skeletonResult.original.clearField(UDATPG_DAYPERIOD_FIELD);
2127             skeletonResult.baseOriginal.clearField(UDATPG_DAYPERIOD_FIELD);
2128             skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = NONE;
2129         }
2130     }
2131     copyFrom(skeletonResult);
2132 }
2133 
2134 void
getBasePattern(UnicodeString & result)2135 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
2136     result.remove(); // Reset the result first.
2137     skeleton.baseOriginal.appendTo(result);
2138 }
2139 
2140 UnicodeString
getPattern()2141 DateTimeMatcher::getPattern() {
2142     UnicodeString result;
2143     return skeleton.original.appendTo(result);
2144 }
2145 
2146 int32_t
getDistance(const DateTimeMatcher & other,int32_t includeMask,DistanceInfo & distanceInfo) const2147 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) const {
2148     int32_t result = 0;
2149     distanceInfo.clear();
2150     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
2151         int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
2152         int32_t otherType = other.skeleton.type[i];
2153         if (myType==otherType) {
2154             continue;
2155         }
2156         if (myType==0) {// and other is not
2157             result += EXTRA_FIELD;
2158             distanceInfo.addExtra(i);
2159         }
2160         else {
2161             if (otherType==0) {
2162                 result += MISSING_FIELD;
2163                 distanceInfo.addMissing(i);
2164             }
2165             else {
2166                 result += abs(myType - otherType);
2167             }
2168         }
2169 
2170     }
2171     return result;
2172 }
2173 
2174 void
copyFrom(const PtnSkeleton & newSkeleton)2175 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
2176     skeleton.copyFrom(newSkeleton);
2177 }
2178 
2179 void
copyFrom()2180 DateTimeMatcher::copyFrom() {
2181     // same as clear
2182     skeleton.clear();
2183 }
2184 
2185 UBool
equals(const DateTimeMatcher * other) const2186 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
2187     if (other==nullptr) { return FALSE; }
2188     return skeleton.original == other->skeleton.original;
2189 }
2190 
2191 int32_t
getFieldMask() const2192 DateTimeMatcher::getFieldMask() const {
2193     int32_t result = 0;
2194 
2195     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2196         if (skeleton.type[i]!=0) {
2197             result |= (1<<i);
2198         }
2199     }
2200     return result;
2201 }
2202 
2203 PtnSkeleton*
getSkeletonPtr()2204 DateTimeMatcher::getSkeletonPtr() {
2205     return &skeleton;
2206 }
2207 
FormatParser()2208 FormatParser::FormatParser () {
2209     status = START;
2210     itemNumber = 0;
2211 }
2212 
2213 
~FormatParser()2214 FormatParser::~FormatParser () {
2215 }
2216 
2217 
2218 // Find the next token with the starting position and length
2219 // Note: the startPos may
2220 FormatParser::TokenStatus
setTokens(const UnicodeString & pattern,int32_t startPos,int32_t * len)2221 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
2222     int32_t curLoc = startPos;
2223     if ( curLoc >= pattern.length()) {
2224         return DONE;
2225     }
2226     // check the current char is between A-Z or a-z
2227     do {
2228         UChar c=pattern.charAt(curLoc);
2229         if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
2230            curLoc++;
2231         }
2232         else {
2233                startPos = curLoc;
2234                *len=1;
2235                return ADD_TOKEN;
2236         }
2237 
2238         if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
2239             break;  // not the same token
2240         }
2241     } while(curLoc <= pattern.length());
2242     *len = curLoc-startPos;
2243     return ADD_TOKEN;
2244 }
2245 
2246 void
set(const UnicodeString & pattern)2247 FormatParser::set(const UnicodeString& pattern) {
2248     int32_t startPos = 0;
2249     TokenStatus result = START;
2250     int32_t len = 0;
2251     itemNumber = 0;
2252 
2253     do {
2254         result = setTokens( pattern, startPos, &len );
2255         if ( result == ADD_TOKEN )
2256         {
2257             items[itemNumber++] = UnicodeString(pattern, startPos, len );
2258             startPos += len;
2259         }
2260         else {
2261             break;
2262         }
2263     } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
2264 }
2265 
2266 int32_t
getCanonicalIndex(const UnicodeString & s,UBool strict)2267 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
2268     int32_t len = s.length();
2269     if (len == 0) {
2270         return -1;
2271     }
2272     UChar ch = s.charAt(0);
2273 
2274     // Verify that all are the same character.
2275     for (int32_t l = 1; l < len; l++) {
2276         if (ch != s.charAt(l)) {
2277             return -1;
2278         }
2279     }
2280     int32_t i = 0;
2281     int32_t bestRow = -1;
2282     while (dtTypes[i].patternChar != 0x0000) {
2283         if ( dtTypes[i].patternChar != ch ) {
2284             ++i;
2285             continue;
2286         }
2287         bestRow = i;
2288         if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
2289             return i;
2290         }
2291         if (dtTypes[i+1].minLen <= len) {
2292             ++i;
2293             continue;
2294         }
2295         return i;
2296     }
2297     return strict ? -1 : bestRow;
2298 }
2299 
2300 UBool
isQuoteLiteral(const UnicodeString & s)2301 FormatParser::isQuoteLiteral(const UnicodeString& s) {
2302     return (UBool)(s.charAt(0) == SINGLE_QUOTE);
2303 }
2304 
2305 // This function assumes the current itemIndex points to the quote literal.
2306 // Please call isQuoteLiteral prior to this function.
2307 void
getQuoteLiteral(UnicodeString & quote,int32_t * itemIndex)2308 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
2309     int32_t i = *itemIndex;
2310 
2311     quote.remove();
2312     if (items[i].charAt(0)==SINGLE_QUOTE) {
2313         quote += items[i];
2314         ++i;
2315     }
2316     while ( i < itemNumber ) {
2317         if ( items[i].charAt(0)==SINGLE_QUOTE ) {
2318             if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
2319                 // two single quotes e.g. 'o''clock'
2320                 quote += items[i++];
2321                 quote += items[i++];
2322                 continue;
2323             }
2324             else {
2325                 quote += items[i];
2326                 break;
2327             }
2328         }
2329         else {
2330             quote += items[i];
2331         }
2332         ++i;
2333     }
2334     *itemIndex=i;
2335 }
2336 
2337 UBool
isPatternSeparator(const UnicodeString & field) const2338 FormatParser::isPatternSeparator(const UnicodeString& field) const {
2339     for (int32_t i=0; i<field.length(); ++i ) {
2340         UChar c= field.charAt(i);
2341         if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
2342              (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
2343             continue;
2344         }
2345         else {
2346             return FALSE;
2347         }
2348     }
2349     return TRUE;
2350 }
2351 
~DistanceInfo()2352 DistanceInfo::~DistanceInfo() {}
2353 
2354 void
setTo(const DistanceInfo & other)2355 DistanceInfo::setTo(const DistanceInfo& other) {
2356     missingFieldMask = other.missingFieldMask;
2357     extraFieldMask= other.extraFieldMask;
2358 }
2359 
PatternMapIterator(UErrorCode & status)2360 PatternMapIterator::PatternMapIterator(UErrorCode& status) :
2361     bootIndex(0), nodePtr(nullptr), matcher(nullptr), patternMap(nullptr)
2362 {
2363     if (U_FAILURE(status)) { return; }
2364     matcher.adoptInsteadAndCheckErrorCode(new DateTimeMatcher(), status);
2365 }
2366 
~PatternMapIterator()2367 PatternMapIterator::~PatternMapIterator() {
2368 }
2369 
2370 void
set(PatternMap & newPatternMap)2371 PatternMapIterator::set(PatternMap& newPatternMap) {
2372     this->patternMap=&newPatternMap;
2373 }
2374 
2375 PtnSkeleton*
getSkeleton() const2376 PatternMapIterator::getSkeleton() const {
2377     if ( nodePtr == nullptr ) {
2378         return nullptr;
2379     }
2380     else {
2381         return nodePtr->skeleton.getAlias();
2382     }
2383 }
2384 
2385 UBool
hasNext() const2386 PatternMapIterator::hasNext() const {
2387     int32_t headIndex = bootIndex;
2388     PtnElem *curPtr = nodePtr;
2389 
2390     if (patternMap==nullptr) {
2391         return FALSE;
2392     }
2393     while ( headIndex < MAX_PATTERN_ENTRIES ) {
2394         if ( curPtr != nullptr ) {
2395             if ( curPtr->next != nullptr ) {
2396                 return TRUE;
2397             }
2398             else {
2399                 headIndex++;
2400                 curPtr=nullptr;
2401                 continue;
2402             }
2403         }
2404         else {
2405             if ( patternMap->boot[headIndex] != nullptr ) {
2406                 return TRUE;
2407             }
2408             else {
2409                 headIndex++;
2410                 continue;
2411             }
2412         }
2413     }
2414     return FALSE;
2415 }
2416 
2417 DateTimeMatcher&
next()2418 PatternMapIterator::next() {
2419     while ( bootIndex < MAX_PATTERN_ENTRIES ) {
2420         if ( nodePtr != nullptr ) {
2421             if ( nodePtr->next != nullptr ) {
2422                 nodePtr = nodePtr->next.getAlias();
2423                 break;
2424             }
2425             else {
2426                 bootIndex++;
2427                 nodePtr=nullptr;
2428                 continue;
2429             }
2430         }
2431         else {
2432             if ( patternMap->boot[bootIndex] != nullptr ) {
2433                 nodePtr = patternMap->boot[bootIndex];
2434                 break;
2435             }
2436             else {
2437                 bootIndex++;
2438                 continue;
2439             }
2440         }
2441     }
2442     if (nodePtr!=nullptr) {
2443         matcher->copyFrom(*nodePtr->skeleton);
2444     }
2445     else {
2446         matcher->copyFrom();
2447     }
2448     return *matcher;
2449 }
2450 
2451 
SkeletonFields()2452 SkeletonFields::SkeletonFields() {
2453     // Set initial values to zero
2454     clear();
2455 }
2456 
clear()2457 void SkeletonFields::clear() {
2458     uprv_memset(chars, 0, sizeof(chars));
2459     uprv_memset(lengths, 0, sizeof(lengths));
2460 }
2461 
copyFrom(const SkeletonFields & other)2462 void SkeletonFields::copyFrom(const SkeletonFields& other) {
2463     uprv_memcpy(chars, other.chars, sizeof(chars));
2464     uprv_memcpy(lengths, other.lengths, sizeof(lengths));
2465 }
2466 
clearField(int32_t field)2467 void SkeletonFields::clearField(int32_t field) {
2468     chars[field] = 0;
2469     lengths[field] = 0;
2470 }
2471 
getFieldChar(int32_t field) const2472 UChar SkeletonFields::getFieldChar(int32_t field) const {
2473     return chars[field];
2474 }
2475 
getFieldLength(int32_t field) const2476 int32_t SkeletonFields::getFieldLength(int32_t field) const {
2477     return lengths[field];
2478 }
2479 
populate(int32_t field,const UnicodeString & value)2480 void SkeletonFields::populate(int32_t field, const UnicodeString& value) {
2481     populate(field, value.charAt(0), value.length());
2482 }
2483 
populate(int32_t field,UChar ch,int32_t length)2484 void SkeletonFields::populate(int32_t field, UChar ch, int32_t length) {
2485     chars[field] = (int8_t) ch;
2486     lengths[field] = (int8_t) length;
2487 }
2488 
isFieldEmpty(int32_t field) const2489 UBool SkeletonFields::isFieldEmpty(int32_t field) const {
2490     return lengths[field] == 0;
2491 }
2492 
appendTo(UnicodeString & string) const2493 UnicodeString& SkeletonFields::appendTo(UnicodeString& string) const {
2494     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2495         appendFieldTo(i, string);
2496     }
2497     return string;
2498 }
2499 
appendFieldTo(int32_t field,UnicodeString & string) const2500 UnicodeString& SkeletonFields::appendFieldTo(int32_t field, UnicodeString& string) const {
2501     UChar ch(chars[field]);
2502     int32_t length = (int32_t) lengths[field];
2503 
2504     for (int32_t i=0; i<length; i++) {
2505         string += ch;
2506     }
2507     return string;
2508 }
2509 
getFirstChar() const2510 UChar SkeletonFields::getFirstChar() const {
2511     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2512         if (lengths[i] != 0) {
2513             return chars[i];
2514         }
2515     }
2516     return '\0';
2517 }
2518 
2519 
PtnSkeleton()2520 PtnSkeleton::PtnSkeleton() {
2521 }
2522 
PtnSkeleton(const PtnSkeleton & other)2523 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
2524     copyFrom(other);
2525 }
2526 
copyFrom(const PtnSkeleton & other)2527 void PtnSkeleton::copyFrom(const PtnSkeleton& other) {
2528     uprv_memcpy(type, other.type, sizeof(type));
2529     original.copyFrom(other.original);
2530     baseOriginal.copyFrom(other.baseOriginal);
2531 }
2532 
clear()2533 void PtnSkeleton::clear() {
2534     uprv_memset(type, 0, sizeof(type));
2535     original.clear();
2536     baseOriginal.clear();
2537 }
2538 
2539 UBool
equals(const PtnSkeleton & other) const2540 PtnSkeleton::equals(const PtnSkeleton& other) const  {
2541     return (original == other.original)
2542         && (baseOriginal == other.baseOriginal)
2543         && (uprv_memcmp(type, other.type, sizeof(type)) == 0);
2544 }
2545 
2546 UnicodeString
getSkeleton() const2547 PtnSkeleton::getSkeleton() const {
2548     UnicodeString result;
2549     result = original.appendTo(result);
2550     int32_t pos;
2551     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2552         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2553         // was not in the provided skeleton, remove it here before returning skeleton.
2554         result.remove(pos, 1);
2555     }
2556     return result;
2557 }
2558 
2559 UnicodeString
getBaseSkeleton() const2560 PtnSkeleton::getBaseSkeleton() const {
2561     UnicodeString result;
2562     result = baseOriginal.appendTo(result);
2563     int32_t pos;
2564     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2565         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2566         // was not in the provided skeleton, remove it here before returning skeleton.
2567         result.remove(pos, 1);
2568     }
2569     return result;
2570 }
2571 
2572 UChar
getFirstChar() const2573 PtnSkeleton::getFirstChar() const {
2574     return baseOriginal.getFirstChar();
2575 }
2576 
~PtnSkeleton()2577 PtnSkeleton::~PtnSkeleton() {
2578 }
2579 
PtnElem(const UnicodeString & basePat,const UnicodeString & pat)2580 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
2581     basePattern(basePat), skeleton(nullptr), pattern(pat), next(nullptr)
2582 {
2583 }
2584 
~PtnElem()2585 PtnElem::~PtnElem() {
2586 }
2587 
DTSkeletonEnumeration(PatternMap & patternMap,dtStrEnum type,UErrorCode & status)2588 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap& patternMap, dtStrEnum type, UErrorCode& status) : fSkeletons(nullptr) {
2589     PtnElem  *curElem;
2590     PtnSkeleton *curSkeleton;
2591     UnicodeString s;
2592     int32_t bootIndex;
2593 
2594     pos=0;
2595     fSkeletons.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2596     if (U_FAILURE(status)) {
2597         return;
2598     }
2599 
2600     for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2601         curElem = patternMap.boot[bootIndex];
2602         while (curElem!=nullptr) {
2603             switch(type) {
2604                 case DT_BASESKELETON:
2605                     s=curElem->basePattern;
2606                     break;
2607                 case DT_PATTERN:
2608                     s=curElem->pattern;
2609                     break;
2610                 case DT_SKELETON:
2611                     curSkeleton=curElem->skeleton.getAlias();
2612                     s=curSkeleton->getSkeleton();
2613                     break;
2614             }
2615             if ( !isCanonicalItem(s) ) {
2616                 LocalPointer<UnicodeString> newElem(new UnicodeString(s), status);
2617                 if (U_FAILURE(status)) {
2618                     return;
2619                 }
2620                 fSkeletons->addElement(newElem.getAlias(), status);
2621                 if (U_FAILURE(status)) {
2622                     fSkeletons.adoptInstead(nullptr);
2623                     return;
2624                 }
2625                 newElem.orphan(); // fSkeletons vector now owns the UnicodeString.
2626             }
2627             curElem = curElem->next.getAlias();
2628         }
2629     }
2630     if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=nullptr) ) {
2631         status = U_BUFFER_OVERFLOW_ERROR;
2632     }
2633 }
2634 
2635 const UnicodeString*
snext(UErrorCode & status)2636 DTSkeletonEnumeration::snext(UErrorCode& status) {
2637     if (U_SUCCESS(status) && fSkeletons.isValid() && pos < fSkeletons->size()) {
2638         return (const UnicodeString*)fSkeletons->elementAt(pos++);
2639     }
2640     return nullptr;
2641 }
2642 
2643 void
reset(UErrorCode &)2644 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2645     pos=0;
2646 }
2647 
2648 int32_t
count(UErrorCode &) const2649 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2650    return (fSkeletons.isNull()) ? 0 : fSkeletons->size();
2651 }
2652 
2653 UBool
isCanonicalItem(const UnicodeString & item)2654 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2655     if ( item.length() != 1 ) {
2656         return FALSE;
2657     }
2658     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2659         if (item.charAt(0)==Canonical_Items[i]) {
2660             return TRUE;
2661         }
2662     }
2663     return FALSE;
2664 }
2665 
~DTSkeletonEnumeration()2666 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2667     UnicodeString *s;
2668     if (fSkeletons.isValid()) {
2669         for (int32_t i = 0; i < fSkeletons->size(); ++i) {
2670             if ((s = (UnicodeString *)fSkeletons->elementAt(i)) != nullptr) {
2671                 delete s;
2672             }
2673         }
2674     }
2675 }
2676 
DTRedundantEnumeration()2677 DTRedundantEnumeration::DTRedundantEnumeration() : pos(0), fPatterns(nullptr) {
2678 }
2679 
2680 void
add(const UnicodeString & pattern,UErrorCode & status)2681 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2682     if (U_FAILURE(status)) { return; }
2683     if (fPatterns.isNull())  {
2684         fPatterns.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2685         if (U_FAILURE(status)) {
2686             return;
2687        }
2688     }
2689     LocalPointer<UnicodeString> newElem(new UnicodeString(pattern), status);
2690     if (U_FAILURE(status)) {
2691         return;
2692     }
2693     fPatterns->addElement(newElem.getAlias(), status);
2694     if (U_FAILURE(status)) {
2695         fPatterns.adoptInstead(nullptr);
2696         return;
2697     }
2698     newElem.orphan(); // fPatterns now owns the string.
2699 }
2700 
2701 const UnicodeString*
snext(UErrorCode & status)2702 DTRedundantEnumeration::snext(UErrorCode& status) {
2703     if (U_SUCCESS(status) && fPatterns.isValid() && pos < fPatterns->size()) {
2704         return (const UnicodeString*)fPatterns->elementAt(pos++);
2705     }
2706     return nullptr;
2707 }
2708 
2709 void
reset(UErrorCode &)2710 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2711     pos=0;
2712 }
2713 
2714 int32_t
count(UErrorCode &) const2715 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2716     return (fPatterns.isNull()) ? 0 : fPatterns->size();
2717 }
2718 
2719 UBool
isCanonicalItem(const UnicodeString & item) const2720 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) const {
2721     if ( item.length() != 1 ) {
2722         return FALSE;
2723     }
2724     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2725         if (item.charAt(0)==Canonical_Items[i]) {
2726             return TRUE;
2727         }
2728     }
2729     return FALSE;
2730 }
2731 
~DTRedundantEnumeration()2732 DTRedundantEnumeration::~DTRedundantEnumeration() {
2733     UnicodeString *s;
2734     if (fPatterns.isValid()) {
2735         for (int32_t i = 0; i < fPatterns->size(); ++i) {
2736             if ((s = (UnicodeString *)fPatterns->elementAt(i)) != nullptr) {
2737                 delete s;
2738             }
2739         }
2740     }
2741 }
2742 
2743 U_NAMESPACE_END
2744 
2745 
2746 #endif /* #if !UCONFIG_NO_FORMATTING */
2747 
2748 //eof
2749