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_K,  // Added ICU-20383, used by JP
475     ALLOWED_HOUR_FORMAT_k,  // Added ICU-20383, not currently used
476     ALLOWED_HOUR_FORMAT_hb,
477     ALLOWED_HOUR_FORMAT_hB,
478     ALLOWED_HOUR_FORMAT_Kb, // Added ICU-20383, not currently used
479     ALLOWED_HOUR_FORMAT_KB, // Added ICU-20383, not currently used
480     // ICU-20383 The following are unlikely and not currently used
481     ALLOWED_HOUR_FORMAT_Hb,
482     ALLOWED_HOUR_FORMAT_HB
483 };
484 
485 }  // namespace
486 
487 void
initData(const Locale & locale,UErrorCode & status)488 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status) {
489     //const char *baseLangName = locale.getBaseName(); // unused
490 
491     skipMatcher = nullptr;
492     fAvailableFormatKeyHash=nullptr;
493     addCanonicalItems(status);
494     addICUPatterns(locale, status);
495     addCLDRData(locale, status);
496     setDateTimeFromCalendar(locale, status);
497     setDecimalSymbols(locale, status);
498     umtx_initOnce(initOnce, loadAllowedHourFormatsData, status);
499     getAllowedHourFormats(locale, status);
500     // If any of the above methods failed then the object is in an invalid state.
501     internalErrorCode = status;
502 } // DateTimePatternGenerator::initData
503 
504 namespace {
505 
506 struct AllowedHourFormatsSink : public ResourceSink {
507     // Initialize sub-sinks.
AllowedHourFormatsSink__anon8479db040211::AllowedHourFormatsSink508     AllowedHourFormatsSink() {}
509     virtual ~AllowedHourFormatsSink();
510 
put__anon8479db040211::AllowedHourFormatsSink511     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
512                      UErrorCode &errorCode) {
513         ResourceTable timeData = value.getTable(errorCode);
514         if (U_FAILURE(errorCode)) { return; }
515         for (int32_t i = 0; timeData.getKeyAndValue(i, key, value); ++i) {
516             const char *regionOrLocale = key;
517             ResourceTable formatList = value.getTable(errorCode);
518             if (U_FAILURE(errorCode)) { return; }
519             // below we construct a list[] that has an entry for the "preferred" value at [0],
520             // followed by 1 or more entries for the "allowed" values, terminated with an
521             // entry for ALLOWED_HOUR_FORMAT_UNKNOWN (not included in length below)
522             LocalMemory<int32_t> list;
523             int32_t length = 0;
524             int32_t preferredFormat = ALLOWED_HOUR_FORMAT_UNKNOWN;
525             for (int32_t j = 0; formatList.getKeyAndValue(j, key, value); ++j) {
526                 if (uprv_strcmp(key, "allowed") == 0) {
527                     if (value.getType() == URES_STRING) {
528                         length = 2; // 1 preferred to add later, 1 allowed to add now
529                         if (list.allocateInsteadAndReset(length + 1) == nullptr) {
530                             errorCode = U_MEMORY_ALLOCATION_ERROR;
531                             return;
532                         }
533                         list[1] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
534                     }
535                     else {
536                         ResourceArray allowedFormats = value.getArray(errorCode);
537                         length = allowedFormats.getSize() + 1; // 1 preferred, getSize allowed
538                         if (list.allocateInsteadAndReset(length + 1) == nullptr) {
539                             errorCode = U_MEMORY_ALLOCATION_ERROR;
540                             return;
541                         }
542                         for (int32_t k = 1; k < length; ++k) {
543                             allowedFormats.getValue(k-1, value);
544                             list[k] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
545                         }
546                     }
547                 } else if (uprv_strcmp(key, "preferred") == 0) {
548                     preferredFormat = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
549                 }
550             }
551             if (length > 1) {
552                 list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: list[1];
553             } else {
554                 // fallback handling for missing data
555                 length = 2; // 1 preferred, 1 allowed
556                 if (list.allocateInsteadAndReset(length + 1) == nullptr) {
557                     errorCode = U_MEMORY_ALLOCATION_ERROR;
558                     return;
559                 }
560                 list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: ALLOWED_HOUR_FORMAT_H;
561                 list[1] = list[0];
562             }
563             list[length] = ALLOWED_HOUR_FORMAT_UNKNOWN;
564             // At this point list[] will have at least two non-ALLOWED_HOUR_FORMAT_UNKNOWN entries,
565             // followed by ALLOWED_HOUR_FORMAT_UNKNOWN.
566             uhash_put(localeToAllowedHourFormatsMap, const_cast<char *>(regionOrLocale), list.orphan(), &errorCode);
567             if (U_FAILURE(errorCode)) { return; }
568         }
569     }
570 
getHourFormatFromUnicodeString__anon8479db040211::AllowedHourFormatsSink571     AllowedHourFormat getHourFormatFromUnicodeString(const UnicodeString &s) {
572         if (s.length() == 1) {
573             if (s[0] == LOW_H) { return ALLOWED_HOUR_FORMAT_h; }
574             if (s[0] == CAP_H) { return ALLOWED_HOUR_FORMAT_H; }
575             if (s[0] == CAP_K) { return ALLOWED_HOUR_FORMAT_K; }
576             if (s[0] == LOW_K) { return ALLOWED_HOUR_FORMAT_k; }
577         } else if (s.length() == 2) {
578             if (s[0] == LOW_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_hb; }
579             if (s[0] == LOW_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_hB; }
580             if (s[0] == CAP_K && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Kb; }
581             if (s[0] == CAP_K && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_KB; }
582             if (s[0] == CAP_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Hb; }
583             if (s[0] == CAP_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_HB; }
584         }
585 
586         return ALLOWED_HOUR_FORMAT_UNKNOWN;
587     }
588 };
589 
590 }  // namespace
591 
~AllowedHourFormatsSink()592 AllowedHourFormatsSink::~AllowedHourFormatsSink() {}
593 
loadAllowedHourFormatsData(UErrorCode & status)594 U_CFUNC void U_CALLCONV DateTimePatternGenerator::loadAllowedHourFormatsData(UErrorCode &status) {
595     if (U_FAILURE(status)) { return; }
596     localeToAllowedHourFormatsMap = uhash_open(
597         uhash_hashChars, uhash_compareChars, nullptr, &status);
598     if (U_FAILURE(status)) { return; }
599 
600     uhash_setValueDeleter(localeToAllowedHourFormatsMap, deleteAllowedHourFormats);
601     ucln_i18n_registerCleanup(UCLN_I18N_ALLOWED_HOUR_FORMATS, allowedHourFormatsCleanup);
602 
603     LocalUResourceBundlePointer rb(ures_openDirect(nullptr, "supplementalData", &status));
604     if (U_FAILURE(status)) { return; }
605 
606     AllowedHourFormatsSink sink;
607     // TODO: Currently in the enumeration each table allocates a new array.
608     // Try to reduce the number of memory allocations. Consider storing a
609     // UVector32 with the concatenation of all of the sub-arrays, put the start index
610     // into the hashmap, store 6 single-value sub-arrays right at the beginning of the
611     // vector (at index enum*2) for easy data sharing, copy sub-arrays into runtime
612     // object. Remember to clean up the vector, too.
613     ures_getAllItemsWithFallback(rb.getAlias(), "timeData", sink, status);
614 }
615 
getAllowedHourFormats(const Locale & locale,UErrorCode & status)616 void DateTimePatternGenerator::getAllowedHourFormats(const Locale &locale, UErrorCode &status) {
617     if (U_FAILURE(status)) { return; }
618     Locale maxLocale(locale);
619     maxLocale.addLikelySubtags(status);
620     if (U_FAILURE(status)) {
621         return;
622     }
623 
624     const char *country = maxLocale.getCountry();
625     if (*country == '\0') { country = "001"; }
626     const char *language = maxLocale.getLanguage();
627 
628     CharString langCountry;
629     langCountry.append(language, static_cast<int32_t>(uprv_strlen(language)), status);
630     langCountry.append('_', status);
631     langCountry.append(country, static_cast<int32_t>(uprv_strlen(country)), status);
632 
633     int32_t *allowedFormats;
634     allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, langCountry.data());
635     if (allowedFormats == nullptr) {
636         allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, const_cast<char *>(country));
637     }
638 
639     if (allowedFormats != nullptr) {  // Lookup is successful
640         // Here allowedFormats points to a list consisting of key for preferredFormat,
641         // followed by one or more keys for allowedFormats, then followed by ALLOWED_HOUR_FORMAT_UNKNOWN.
642         switch (allowedFormats[0]) {
643             case ALLOWED_HOUR_FORMAT_h: fDefaultHourFormatChar = LOW_H; break;
644             case ALLOWED_HOUR_FORMAT_H: fDefaultHourFormatChar = CAP_H; break;
645             case ALLOWED_HOUR_FORMAT_K: fDefaultHourFormatChar = CAP_K; break;
646             case ALLOWED_HOUR_FORMAT_k: fDefaultHourFormatChar = LOW_K; break;
647             default: fDefaultHourFormatChar = CAP_H; break;
648         }
649         for (int32_t i = 0; i < UPRV_LENGTHOF(fAllowedHourFormats); ++i) {
650             fAllowedHourFormats[i] = allowedFormats[i + 1];
651             if (fAllowedHourFormats[i] == ALLOWED_HOUR_FORMAT_UNKNOWN) {
652                 break;
653             }
654         }
655     } else {  // Lookup failed, twice
656         fDefaultHourFormatChar = CAP_H;
657         fAllowedHourFormats[0] = ALLOWED_HOUR_FORMAT_H;
658         fAllowedHourFormats[1] = ALLOWED_HOUR_FORMAT_UNKNOWN;
659     }
660 }
661 
662 UnicodeString
getSkeleton(const UnicodeString & pattern,UErrorCode &)663 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode&
664 /*status*/) {
665     FormatParser fp2;
666     DateTimeMatcher matcher;
667     PtnSkeleton localSkeleton;
668     matcher.set(pattern, &fp2, localSkeleton);
669     return localSkeleton.getSkeleton();
670 }
671 
672 UnicodeString
staticGetSkeleton(const UnicodeString & pattern,UErrorCode &)673 DateTimePatternGenerator::staticGetSkeleton(
674         const UnicodeString& pattern, UErrorCode& /*status*/) {
675     FormatParser fp;
676     DateTimeMatcher matcher;
677     PtnSkeleton localSkeleton;
678     matcher.set(pattern, &fp, localSkeleton);
679     return localSkeleton.getSkeleton();
680 }
681 
682 UnicodeString
getBaseSkeleton(const UnicodeString & pattern,UErrorCode &)683 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) {
684     FormatParser fp2;
685     DateTimeMatcher matcher;
686     PtnSkeleton localSkeleton;
687     matcher.set(pattern, &fp2, localSkeleton);
688     return localSkeleton.getBaseSkeleton();
689 }
690 
691 UnicodeString
staticGetBaseSkeleton(const UnicodeString & pattern,UErrorCode &)692 DateTimePatternGenerator::staticGetBaseSkeleton(
693         const UnicodeString& pattern, UErrorCode& /*status*/) {
694     FormatParser fp;
695     DateTimeMatcher matcher;
696     PtnSkeleton localSkeleton;
697     matcher.set(pattern, &fp, localSkeleton);
698     return localSkeleton.getBaseSkeleton();
699 }
700 
701 void
addICUPatterns(const Locale & locale,UErrorCode & status)702 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) {
703     if (U_FAILURE(status)) { return; }
704     UnicodeString dfPattern;
705     UnicodeString conflictingString;
706     DateFormat* df;
707 
708     // Load with ICU patterns
709     for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) {
710         DateFormat::EStyle style = (DateFormat::EStyle)i;
711         df = DateFormat::createDateInstance(style, locale);
712         SimpleDateFormat* sdf;
713         if (df != nullptr && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != nullptr) {
714             sdf->toPattern(dfPattern);
715             addPattern(dfPattern, FALSE, conflictingString, status);
716         }
717         // TODO Maybe we should return an error when the date format isn't simple.
718         delete df;
719         if (U_FAILURE(status)) { return; }
720 
721         df = DateFormat::createTimeInstance(style, locale);
722         if (df != nullptr && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != nullptr) {
723             sdf->toPattern(dfPattern);
724             addPattern(dfPattern, FALSE, conflictingString, status);
725 
726             // TODO: C++ and Java are inconsistent (see #12568).
727             // C++ uses MEDIUM, but Java uses SHORT.
728             if ( i==DateFormat::kShort && !dfPattern.isEmpty() ) {
729                 consumeShortTimePattern(dfPattern, status);
730             }
731         }
732         // TODO Maybe we should return an error when the date format isn't simple.
733         delete df;
734         if (U_FAILURE(status)) { return; }
735     }
736 }
737 
738 void
hackTimes(const UnicodeString & hackPattern,UErrorCode & status)739 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status)  {
740     UnicodeString conflictingString;
741 
742     fp->set(hackPattern);
743     UnicodeString mmss;
744     UBool gotMm=FALSE;
745     for (int32_t i=0; i<fp->itemNumber; ++i) {
746         UnicodeString field = fp->items[i];
747         if ( fp->isQuoteLiteral(field) ) {
748             if ( gotMm ) {
749                UnicodeString quoteLiteral;
750                fp->getQuoteLiteral(quoteLiteral, &i);
751                mmss += quoteLiteral;
752             }
753         }
754         else {
755             if (fp->isPatternSeparator(field) && gotMm) {
756                 mmss+=field;
757             }
758             else {
759                 UChar ch=field.charAt(0);
760                 if (ch==LOW_M) {
761                     gotMm=TRUE;
762                     mmss+=field;
763                 }
764                 else {
765                     if (ch==LOW_S) {
766                         if (!gotMm) {
767                             break;
768                         }
769                         mmss+= field;
770                         addPattern(mmss, FALSE, conflictingString, status);
771                         break;
772                     }
773                     else {
774                         if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) {
775                             break;
776                         }
777                     }
778                 }
779             }
780         }
781     }
782 }
783 
784 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY)
785 
786 void
getCalendarTypeToUse(const Locale & locale,CharString & destination,UErrorCode & err)787 DateTimePatternGenerator::getCalendarTypeToUse(const Locale& locale, CharString& destination, UErrorCode& err) {
788     destination.clear().append(DT_DateTimeGregorianTag, -1, err); // initial default
789     if ( U_SUCCESS(err) ) {
790         UErrorCode localStatus = U_ZERO_ERROR;
791         char localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
792         // obtain a locale that always has the calendar key value that should be used
793         ures_getFunctionalEquivalent(
794             localeWithCalendarKey,
795             ULOC_LOCALE_IDENTIFIER_CAPACITY,
796             nullptr,
797             "calendar",
798             "calendar",
799             locale.getName(),
800             nullptr,
801             FALSE,
802             &localStatus);
803         localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination
804         // now get the calendar key value from that locale
805         char calendarType[ULOC_KEYWORDS_CAPACITY];
806         int32_t calendarTypeLen = uloc_getKeywordValue(
807             localeWithCalendarKey,
808             "calendar",
809             calendarType,
810             ULOC_KEYWORDS_CAPACITY,
811             &localStatus);
812         // If the input locale was invalid, don't fail with missing resource error, instead
813         // continue with default of Gregorian.
814         if (U_FAILURE(localStatus) && localStatus != U_MISSING_RESOURCE_ERROR) {
815             err = localStatus;
816             return;
817         }
818         if (calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
819             destination.clear().append(calendarType, -1, err);
820             if (U_FAILURE(err)) { return; }
821         }
822     }
823 }
824 
825 void
consumeShortTimePattern(const UnicodeString & shortTimePattern,UErrorCode & status)826 DateTimePatternGenerator::consumeShortTimePattern(const UnicodeString& shortTimePattern,
827         UErrorCode& status) {
828     if (U_FAILURE(status)) { return; }
829     // ICU-20383 No longer set fDefaultHourFormatChar to the hour format character from
830     // this pattern; instead it is set from localeToAllowedHourFormatsMap which now
831     // includes entries for both preferred and allowed formats.
832 
833     // HACK for hh:ss
834     hackTimes(shortTimePattern, status);
835 }
836 
837 struct DateTimePatternGenerator::AppendItemFormatsSink : public ResourceSink {
838 
839     // Destination for data, modified via setters.
840     DateTimePatternGenerator& dtpg;
841 
AppendItemFormatsSinkDateTimePatternGenerator::AppendItemFormatsSink842     AppendItemFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
843     virtual ~AppendItemFormatsSink();
844 
putDateTimePatternGenerator::AppendItemFormatsSink845     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
846             UErrorCode &errorCode) {
847         ResourceTable itemsTable = value.getTable(errorCode);
848         if (U_FAILURE(errorCode)) { return; }
849         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
850             UDateTimePatternField field = dtpg.getAppendFormatNumber(key);
851             if (field == UDATPG_FIELD_COUNT) { continue; }
852             const UnicodeString& valueStr = value.getUnicodeString(errorCode);
853             if (dtpg.getAppendItemFormat(field).isEmpty() && !valueStr.isEmpty()) {
854                 dtpg.setAppendItemFormat(field, valueStr);
855             }
856         }
857     }
858 
fillInMissingDateTimePatternGenerator::AppendItemFormatsSink859     void fillInMissing() {
860         UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, UPRV_LENGTHOF(UDATPG_ItemFormat)-1);  // Read-only alias.
861         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
862             UDateTimePatternField field = (UDateTimePatternField)i;
863             if (dtpg.getAppendItemFormat(field).isEmpty()) {
864                 dtpg.setAppendItemFormat(field, defaultItemFormat);
865             }
866         }
867     }
868 };
869 
870 struct DateTimePatternGenerator::AppendItemNamesSink : public ResourceSink {
871 
872     // Destination for data, modified via setters.
873     DateTimePatternGenerator& dtpg;
874 
AppendItemNamesSinkDateTimePatternGenerator::AppendItemNamesSink875     AppendItemNamesSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
876     virtual ~AppendItemNamesSink();
877 
putDateTimePatternGenerator::AppendItemNamesSink878     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
879             UErrorCode &errorCode) {
880         ResourceTable itemsTable = value.getTable(errorCode);
881         if (U_FAILURE(errorCode)) { return; }
882         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
883             UDateTimePGDisplayWidth width;
884             UDateTimePatternField field = dtpg.getFieldAndWidthIndices(key, &width);
885             if (field == UDATPG_FIELD_COUNT) { continue; }
886             ResourceTable detailsTable = value.getTable(errorCode);
887             if (U_FAILURE(errorCode)) { return; }
888             for (int32_t j = 0; detailsTable.getKeyAndValue(j, key, value); ++j) {
889                 if (uprv_strcmp(key, "dn") != 0) { continue; }
890                 const UnicodeString& valueStr = value.getUnicodeString(errorCode);
891                 if (dtpg.getFieldDisplayName(field,width).isEmpty() && !valueStr.isEmpty()) {
892                     dtpg.setFieldDisplayName(field,width,valueStr);
893                 }
894                 break;
895             }
896         }
897     }
898 
fillInMissingDateTimePatternGenerator::AppendItemNamesSink899     void fillInMissing() {
900         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
901             UnicodeString& valueStr = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, UDATPG_WIDE);
902             if (valueStr.isEmpty()) {
903                 valueStr = CAP_F;
904                 U_ASSERT(i < 20);
905                 if (i < 10) {
906                     // F0, F1, ..., F9
907                     valueStr += (UChar)(i+0x30);
908                 } else {
909                     // F10, F11, ...
910                     valueStr += (UChar)0x31;
911                     valueStr += (UChar)(i-10 + 0x30);
912                 }
913                 // NUL-terminate for the C API.
914                 valueStr.getTerminatedBuffer();
915             }
916             for (int32_t j = 1; j < UDATPG_WIDTH_COUNT; j++) {
917                 UnicodeString& valueStr2 = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)j);
918                 if (valueStr2.isEmpty()) {
919                     valueStr2 = dtpg.getFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)(j-1));
920                 }
921             }
922         }
923     }
924 };
925 
926 struct DateTimePatternGenerator::AvailableFormatsSink : public ResourceSink {
927 
928     // Destination for data, modified via setters.
929     DateTimePatternGenerator& dtpg;
930 
931     // Temporary variable, required for calling addPatternWithSkeleton.
932     UnicodeString conflictingPattern;
933 
AvailableFormatsSinkDateTimePatternGenerator::AvailableFormatsSink934     AvailableFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
935     virtual ~AvailableFormatsSink();
936 
putDateTimePatternGenerator::AvailableFormatsSink937     virtual void put(const char *key, ResourceValue &value, UBool isRoot,
938             UErrorCode &errorCode) {
939         ResourceTable itemsTable = value.getTable(errorCode);
940         if (U_FAILURE(errorCode)) { return; }
941         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
942             const UnicodeString formatKey(key, -1, US_INV);
943             if (!dtpg.isAvailableFormatSet(formatKey) ) {
944                 dtpg.setAvailableFormat(formatKey, errorCode);
945                 // Add pattern with its associated skeleton. Override any duplicate
946                 // derived from std patterns, but not a previous availableFormats entry:
947                 const UnicodeString& formatValue = value.getUnicodeString(errorCode);
948                 conflictingPattern.remove();
949                 dtpg.addPatternWithSkeleton(formatValue, &formatKey, !isRoot, conflictingPattern, errorCode);
950             }
951         }
952     }
953 };
954 
955 // Virtual destructors must be defined out of line.
~AppendItemFormatsSink()956 DateTimePatternGenerator::AppendItemFormatsSink::~AppendItemFormatsSink() {}
~AppendItemNamesSink()957 DateTimePatternGenerator::AppendItemNamesSink::~AppendItemNamesSink() {}
~AvailableFormatsSink()958 DateTimePatternGenerator::AvailableFormatsSink::~AvailableFormatsSink() {}
959 
960 void
addCLDRData(const Locale & locale,UErrorCode & errorCode)961 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& errorCode) {
962     if (U_FAILURE(errorCode)) { return; }
963     UnicodeString rbPattern, value, field;
964     CharString path;
965 
966     LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &errorCode));
967     if (U_FAILURE(errorCode)) { return; }
968 
969     CharString calendarTypeToUse; // to be filled in with the type to use, if all goes well
970     getCalendarTypeToUse(locale, calendarTypeToUse, errorCode);
971     if (U_FAILURE(errorCode)) { return; }
972 
973     // Local err to ignore resource not found exceptions
974     UErrorCode err = U_ZERO_ERROR;
975 
976     // Load append item formats.
977     AppendItemFormatsSink appendItemFormatsSink(*this);
978     path.clear()
979         .append(DT_DateTimeCalendarTag, errorCode)
980         .append('/', errorCode)
981         .append(calendarTypeToUse, errorCode)
982         .append('/', errorCode)
983         .append(DT_DateTimeAppendItemsTag, errorCode); // i.e., calendar/xxx/appendItems
984     if (U_FAILURE(errorCode)) { return; }
985     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), appendItemFormatsSink, err);
986     appendItemFormatsSink.fillInMissing();
987 
988     // Load CLDR item names.
989     err = U_ZERO_ERROR;
990     AppendItemNamesSink appendItemNamesSink(*this);
991     ures_getAllItemsWithFallback(rb.getAlias(), DT_DateTimeFieldsTag, appendItemNamesSink, err);
992     appendItemNamesSink.fillInMissing();
993 
994     // Load the available formats from CLDR.
995     err = U_ZERO_ERROR;
996     initHashtable(errorCode);
997     if (U_FAILURE(errorCode)) { return; }
998     AvailableFormatsSink availableFormatsSink(*this);
999     path.clear()
1000         .append(DT_DateTimeCalendarTag, errorCode)
1001         .append('/', errorCode)
1002         .append(calendarTypeToUse, errorCode)
1003         .append('/', errorCode)
1004         .append(DT_DateTimeAvailableFormatsTag, errorCode); // i.e., calendar/xxx/availableFormats
1005     if (U_FAILURE(errorCode)) { return; }
1006     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), availableFormatsSink, err);
1007 }
1008 
1009 void
initHashtable(UErrorCode & err)1010 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
1011     if (U_FAILURE(err)) { return; }
1012     if (fAvailableFormatKeyHash!=nullptr) {
1013         return;
1014     }
1015     LocalPointer<Hashtable> hash(new Hashtable(FALSE, err), err);
1016     if (U_SUCCESS(err)) {
1017         fAvailableFormatKeyHash = hash.orphan();
1018     }
1019 }
1020 
1021 void
setAppendItemFormat(UDateTimePatternField field,const UnicodeString & value)1022 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
1023     appendItemFormats[field] = value;
1024     // NUL-terminate for the C API.
1025     appendItemFormats[field].getTerminatedBuffer();
1026 }
1027 
1028 const UnicodeString&
getAppendItemFormat(UDateTimePatternField field) const1029 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
1030     return appendItemFormats[field];
1031 }
1032 
1033 void
setAppendItemName(UDateTimePatternField field,const UnicodeString & value)1034 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
1035     setFieldDisplayName(field, UDATPG_WIDTH_APPENDITEM, value);
1036 }
1037 
1038 const UnicodeString&
getAppendItemName(UDateTimePatternField field) const1039 DateTimePatternGenerator::getAppendItemName(UDateTimePatternField field) const {
1040     return fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1041 }
1042 
1043 void
setFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width,const UnicodeString & value)1044 DateTimePatternGenerator::setFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width, const UnicodeString& value) {
1045     fieldDisplayNames[field][width] = value;
1046     // NUL-terminate for the C API.
1047     fieldDisplayNames[field][width].getTerminatedBuffer();
1048 }
1049 
1050 UnicodeString
getFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width) const1051 DateTimePatternGenerator::getFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) const {
1052     return fieldDisplayNames[field][width];
1053 }
1054 
1055 UnicodeString&
getMutableFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width)1056 DateTimePatternGenerator::getMutableFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) {
1057     return fieldDisplayNames[field][width];
1058 }
1059 
1060 void
getAppendName(UDateTimePatternField field,UnicodeString & value)1061 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
1062     value = SINGLE_QUOTE;
1063     value += fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1064     value += SINGLE_QUOTE;
1065 }
1066 
1067 UnicodeString
getBestPattern(const UnicodeString & patternForm,UErrorCode & status)1068 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
1069     return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
1070 }
1071 
1072 UnicodeString
getBestPattern(const UnicodeString & patternForm,UDateTimePatternMatchOptions options,UErrorCode & status)1073 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
1074     if (U_FAILURE(status)) {
1075         return UnicodeString();
1076     }
1077     if (U_FAILURE(internalErrorCode)) {
1078         status = internalErrorCode;
1079         return UnicodeString();
1080     }
1081     const UnicodeString *bestPattern = nullptr;
1082     UnicodeString dtFormat;
1083     UnicodeString resultPattern;
1084     int32_t flags = kDTPGNoFlags;
1085 
1086     int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
1087     int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
1088 
1089     // Replace hour metacharacters 'j', 'C' and 'J', set flags as necessary
1090     UnicodeString patternFormMapped = mapSkeletonMetacharacters(patternForm, &flags, status);
1091     if (U_FAILURE(status)) {
1092         return UnicodeString();
1093     }
1094 
1095     resultPattern.remove();
1096     dtMatcher->set(patternFormMapped, fp);
1097     const PtnSkeleton* specifiedSkeleton = nullptr;
1098     bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, status, &specifiedSkeleton);
1099     if (U_FAILURE(status)) {
1100         return UnicodeString();
1101     }
1102 
1103     if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
1104         resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, flags, options);
1105 
1106         return resultPattern;
1107     }
1108     int32_t neededFields = dtMatcher->getFieldMask();
1109     UnicodeString datePattern=getBestAppending(neededFields & dateMask, flags, status, options);
1110     UnicodeString timePattern=getBestAppending(neededFields & timeMask, flags, status, options);
1111     if (U_FAILURE(status)) {
1112         return UnicodeString();
1113     }
1114     if (datePattern.length()==0) {
1115         if (timePattern.length()==0) {
1116             resultPattern.remove();
1117         }
1118         else {
1119             return timePattern;
1120         }
1121     }
1122     if (timePattern.length()==0) {
1123         return datePattern;
1124     }
1125     resultPattern.remove();
1126     status = U_ZERO_ERROR;
1127     dtFormat=getDateTimeFormat();
1128     SimpleFormatter(dtFormat, 2, 2, status).format(timePattern, datePattern, resultPattern, status);
1129     return resultPattern;
1130 }
1131 
1132 /*
1133  * Map a skeleton that may have metacharacters jJC to one without, by replacing
1134  * the metacharacters with locale-appropriate fields of h/H/k/K and of a/b/B
1135  * (depends on fDefaultHourFormatChar and fAllowedHourFormats being set, which in
1136  * turn depends on initData having been run). This method also updates the flags
1137  * as necessary. Returns the updated skeleton.
1138  */
1139 UnicodeString
mapSkeletonMetacharacters(const UnicodeString & patternForm,int32_t * flags,UErrorCode & status)1140 DateTimePatternGenerator::mapSkeletonMetacharacters(const UnicodeString& patternForm, int32_t* flags, UErrorCode& status) {
1141     UnicodeString patternFormMapped;
1142     patternFormMapped.remove();
1143     UBool inQuoted = FALSE;
1144     int32_t patPos, patLen = patternForm.length();
1145     for (patPos = 0; patPos < patLen; patPos++) {
1146         UChar patChr = patternForm.charAt(patPos);
1147         if (patChr == SINGLE_QUOTE) {
1148             inQuoted = !inQuoted;
1149         } else if (!inQuoted) {
1150             // Handle special mappings for 'j' and 'C' in which fields lengths
1151             // 1,3,5 => hour field length 1
1152             // 2,4,6 => hour field length 2
1153             // 1,2 => abbreviated dayPeriod (field length 1..3)
1154             // 3,4 => long dayPeriod (field length 4)
1155             // 5,6 => narrow dayPeriod (field length 5)
1156             if (patChr == LOW_J || patChr == CAP_C) {
1157                 int32_t extraLen = 0; // 1 less than total field length
1158                 while (patPos+1 < patLen && patternForm.charAt(patPos+1)==patChr) {
1159                     extraLen++;
1160                     patPos++;
1161                 }
1162                 int32_t hourLen = 1 + (extraLen & 1);
1163                 int32_t dayPeriodLen = (extraLen < 2)? 1: 3 + (extraLen >> 1);
1164                 UChar hourChar = LOW_H;
1165                 UChar dayPeriodChar = LOW_A;
1166                 if (patChr == LOW_J) {
1167                     hourChar = fDefaultHourFormatChar;
1168                 } else {
1169                     AllowedHourFormat bestAllowed;
1170                     if (fAllowedHourFormats[0] != ALLOWED_HOUR_FORMAT_UNKNOWN) {
1171                         bestAllowed = (AllowedHourFormat)fAllowedHourFormats[0];
1172                     } else {
1173                         status = U_INVALID_FORMAT_ERROR;
1174                         return UnicodeString();
1175                     }
1176                     if (bestAllowed == ALLOWED_HOUR_FORMAT_H || bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_Hb) {
1177                         hourChar = CAP_H;
1178                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_K || bestAllowed == ALLOWED_HOUR_FORMAT_KB || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) {
1179                         hourChar = CAP_K;
1180                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_k) {
1181                         hourChar = LOW_K;
1182                     }
1183                     // in #13183 just add b/B to skeleton, no longer need to set special flags
1184                     if (bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_hB || bestAllowed == ALLOWED_HOUR_FORMAT_KB) {
1185                         dayPeriodChar = CAP_B;
1186                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_Hb || bestAllowed == ALLOWED_HOUR_FORMAT_hb || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) {
1187                         dayPeriodChar = LOW_B;
1188                     }
1189                 }
1190                 if (hourChar==CAP_H || hourChar==LOW_K) {
1191                     dayPeriodLen = 0;
1192                 }
1193                 while (dayPeriodLen-- > 0) {
1194                     patternFormMapped.append(dayPeriodChar);
1195                 }
1196                 while (hourLen-- > 0) {
1197                     patternFormMapped.append(hourChar);
1198                 }
1199             } else if (patChr == CAP_J) {
1200                 // Get pattern for skeleton with H, then replace H or k
1201                 // with fDefaultHourFormatChar (if different)
1202                 patternFormMapped.append(CAP_H);
1203                 *flags |= kDTPGSkeletonUsesCapJ;
1204             } else {
1205                 patternFormMapped.append(patChr);
1206             }
1207         }
1208     }
1209     return patternFormMapped;
1210 }
1211 
1212 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UErrorCode & status)1213 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1214                                             const UnicodeString& skeleton,
1215                                             UErrorCode& status) {
1216     return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
1217 }
1218 
1219 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UDateTimePatternMatchOptions options,UErrorCode & status)1220 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1221                                             const UnicodeString& skeleton,
1222                                             UDateTimePatternMatchOptions options,
1223                                             UErrorCode& status) {
1224     if (U_FAILURE(status)) {
1225         return UnicodeString();
1226     }
1227     if (U_FAILURE(internalErrorCode)) {
1228         status = internalErrorCode;
1229         return UnicodeString();
1230     }
1231     dtMatcher->set(skeleton, fp);
1232     UnicodeString result = adjustFieldTypes(pattern, nullptr, kDTPGNoFlags, options);
1233     return result;
1234 }
1235 
1236 void
setDecimal(const UnicodeString & newDecimal)1237 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
1238     this->decimal = newDecimal;
1239     // NUL-terminate for the C API.
1240     this->decimal.getTerminatedBuffer();
1241 }
1242 
1243 const UnicodeString&
getDecimal() const1244 DateTimePatternGenerator::getDecimal() const {
1245     return decimal;
1246 }
1247 
1248 void
addCanonicalItems(UErrorCode & status)1249 DateTimePatternGenerator::addCanonicalItems(UErrorCode& status) {
1250     if (U_FAILURE(status)) { return; }
1251     UnicodeString  conflictingPattern;
1252 
1253     for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
1254         if (Canonical_Items[i] > 0) {
1255             addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
1256         }
1257         if (U_FAILURE(status)) { return; }
1258     }
1259 }
1260 
1261 void
setDateTimeFormat(const UnicodeString & dtFormat)1262 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
1263     dateTimeFormat = dtFormat;
1264     // NUL-terminate for the C API.
1265     dateTimeFormat.getTerminatedBuffer();
1266 }
1267 
1268 const UnicodeString&
getDateTimeFormat() const1269 DateTimePatternGenerator::getDateTimeFormat() const {
1270     return dateTimeFormat;
1271 }
1272 
1273 void
setDateTimeFromCalendar(const Locale & locale,UErrorCode & status)1274 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
1275     if (U_FAILURE(status)) { return; }
1276 
1277     const UChar *resStr;
1278     int32_t resStrLen = 0;
1279 
1280     LocalPointer<Calendar> fCalendar(Calendar::createInstance(locale, status), status);
1281     if (U_FAILURE(status)) { return; }
1282 
1283     LocalUResourceBundlePointer calData(ures_open(nullptr, locale.getBaseName(), &status));
1284     if (U_FAILURE(status)) { return; }
1285     ures_getByKey(calData.getAlias(), DT_DateTimeCalendarTag, calData.getAlias(), &status);
1286     if (U_FAILURE(status)) { return; }
1287 
1288     LocalUResourceBundlePointer dateTimePatterns;
1289     if (fCalendar->getType() != nullptr && *fCalendar->getType() != '\0'
1290             && uprv_strcmp(fCalendar->getType(), DT_DateTimeGregorianTag) != 0) {
1291         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), fCalendar->getType(),
1292                                                                 nullptr, &status));
1293         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1294                                   dateTimePatterns.getAlias(), &status);
1295     }
1296 
1297     if (dateTimePatterns.isNull() || status == U_MISSING_RESOURCE_ERROR) {
1298         status = U_ZERO_ERROR;
1299         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), DT_DateTimeGregorianTag,
1300                                                                 dateTimePatterns.orphan(), &status));
1301         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1302                                   dateTimePatterns.getAlias(), &status);
1303     }
1304     if (U_FAILURE(status)) { return; }
1305 
1306     if (ures_getSize(dateTimePatterns.getAlias()) <= DateFormat::kDateTime)
1307     {
1308         status = U_INVALID_FORMAT_ERROR;
1309         return;
1310     }
1311     resStr = ures_getStringByIndex(dateTimePatterns.getAlias(), (int32_t)DateFormat::kDateTime, &resStrLen, &status);
1312     setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
1313 }
1314 
1315 void
setDecimalSymbols(const Locale & locale,UErrorCode & status)1316 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
1317     DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
1318     if(U_SUCCESS(status)) {
1319         decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
1320         // NUL-terminate for the C API.
1321         decimal.getTerminatedBuffer();
1322     }
1323 }
1324 
1325 UDateTimePatternConflict
addPattern(const UnicodeString & pattern,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1326 DateTimePatternGenerator::addPattern(
1327     const UnicodeString& pattern,
1328     UBool override,
1329     UnicodeString &conflictingPattern,
1330     UErrorCode& status)
1331 {
1332     if (U_FAILURE(internalErrorCode)) {
1333         status = internalErrorCode;
1334         return UDATPG_NO_CONFLICT;
1335     }
1336 
1337     return addPatternWithSkeleton(pattern, nullptr, override, conflictingPattern, status);
1338 }
1339 
1340 // For DateTimePatternGenerator::addPatternWithSkeleton -
1341 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
1342 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
1343 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
1344 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
1345 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
1346 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
1347 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
1348 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
1349 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
1350 UDateTimePatternConflict
addPatternWithSkeleton(const UnicodeString & pattern,const UnicodeString * skeletonToUse,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1351 DateTimePatternGenerator::addPatternWithSkeleton(
1352     const UnicodeString& pattern,
1353     const UnicodeString* skeletonToUse,
1354     UBool override,
1355     UnicodeString& conflictingPattern,
1356     UErrorCode& status)
1357 {
1358     if (U_FAILURE(internalErrorCode)) {
1359         status = internalErrorCode;
1360         return UDATPG_NO_CONFLICT;
1361     }
1362 
1363     UnicodeString basePattern;
1364     PtnSkeleton   skeleton;
1365     UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
1366 
1367     DateTimeMatcher matcher;
1368     if ( skeletonToUse == nullptr ) {
1369         matcher.set(pattern, fp, skeleton);
1370         matcher.getBasePattern(basePattern);
1371     } else {
1372         matcher.set(*skeletonToUse, fp, skeleton); // no longer trims skeleton fields to max len 3, per #7930
1373         matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
1374     }
1375     // We only care about base conflicts - and replacing the pattern associated with a base - if:
1376     // 1. the conflicting previous base pattern did *not* have an explicit skeleton; in that case the previous
1377     // base + pattern combination was derived from either (a) a canonical item, (b) a standard format, or
1378     // (c) a pattern specified programmatically with a previous call to addPattern (which would only happen
1379     // if we are getting here from a subsequent call to addPattern).
1380     // 2. a skeleton is specified for the current pattern, but override=false; in that case we are checking
1381     // availableFormats items from root, which should not override any previous entry with the same base.
1382     UBool entryHadSpecifiedSkeleton;
1383     const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
1384     if (duplicatePattern != nullptr && (!entryHadSpecifiedSkeleton || (skeletonToUse != nullptr && !override))) {
1385         conflictingStatus = UDATPG_BASE_CONFLICT;
1386         conflictingPattern = *duplicatePattern;
1387         if (!override) {
1388             return conflictingStatus;
1389         }
1390     }
1391     // The only time we get here with override=true and skeletonToUse!=null is when adding availableFormats
1392     // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with
1393     // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for
1394     // the previously-specified conflicting item.
1395     const PtnSkeleton* entrySpecifiedSkeleton = nullptr;
1396     duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
1397     if (duplicatePattern != nullptr ) {
1398         conflictingStatus = UDATPG_CONFLICT;
1399         conflictingPattern = *duplicatePattern;
1400         if (!override || (skeletonToUse != nullptr && entrySpecifiedSkeleton != nullptr)) {
1401             return conflictingStatus;
1402         }
1403     }
1404     patternMap->add(basePattern, skeleton, pattern, skeletonToUse != nullptr, status);
1405     if(U_FAILURE(status)) {
1406         return conflictingStatus;
1407     }
1408 
1409     return UDATPG_NO_CONFLICT;
1410 }
1411 
1412 
1413 UDateTimePatternField
getAppendFormatNumber(const char * field) const1414 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
1415     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1416         if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
1417             return (UDateTimePatternField)i;
1418         }
1419     }
1420     return UDATPG_FIELD_COUNT;
1421 }
1422 
1423 UDateTimePatternField
getFieldAndWidthIndices(const char * key,UDateTimePGDisplayWidth * widthP) const1424 DateTimePatternGenerator::getFieldAndWidthIndices(const char* key, UDateTimePGDisplayWidth* widthP) const {
1425     char cldrFieldKey[UDATPG_FIELD_KEY_MAX + 1];
1426     uprv_strncpy(cldrFieldKey, key, UDATPG_FIELD_KEY_MAX);
1427     cldrFieldKey[UDATPG_FIELD_KEY_MAX]=0; // ensure termination
1428     *widthP = UDATPG_WIDE;
1429     char* hyphenPtr = uprv_strchr(cldrFieldKey, '-');
1430     if (hyphenPtr) {
1431         for (int32_t i=UDATPG_WIDTH_COUNT-1; i>0; --i) {
1432             if (uprv_strcmp(CLDR_FIELD_WIDTH[i], hyphenPtr)==0) {
1433                 *widthP=(UDateTimePGDisplayWidth)i;
1434                 break;
1435             }
1436         }
1437         *hyphenPtr = 0; // now delete width portion of key
1438     }
1439     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1440         if (uprv_strcmp(CLDR_FIELD_NAME[i],cldrFieldKey)==0) {
1441             return (UDateTimePatternField)i;
1442         }
1443     }
1444     return UDATPG_FIELD_COUNT;
1445 }
1446 
1447 const UnicodeString*
getBestRaw(DateTimeMatcher & source,int32_t includeMask,DistanceInfo * missingFields,UErrorCode & status,const PtnSkeleton ** specifiedSkeletonPtr)1448 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
1449                                      int32_t includeMask,
1450                                      DistanceInfo* missingFields,
1451                                      UErrorCode &status,
1452                                      const PtnSkeleton** specifiedSkeletonPtr) {
1453     int32_t bestDistance = 0x7fffffff;
1454     DistanceInfo tempInfo;
1455     const UnicodeString *bestPattern=nullptr;
1456     const PtnSkeleton* specifiedSkeleton=nullptr;
1457 
1458     PatternMapIterator it(status);
1459     if (U_FAILURE(status)) { return nullptr; }
1460 
1461     for (it.set(*patternMap); it.hasNext(); ) {
1462         DateTimeMatcher trial = it.next();
1463         if (trial.equals(skipMatcher)) {
1464             continue;
1465         }
1466         int32_t distance=source.getDistance(trial, includeMask, tempInfo);
1467         if (distance<bestDistance) {
1468             bestDistance=distance;
1469             bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
1470             missingFields->setTo(tempInfo);
1471             if (distance==0) {
1472                 break;
1473             }
1474         }
1475     }
1476 
1477     // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
1478     // then return it too. This generally happens when the caller needs to pass that skeleton
1479     // through to adjustFieldTypes so the latter can do a better job.
1480     if (bestPattern && specifiedSkeletonPtr) {
1481         *specifiedSkeletonPtr = specifiedSkeleton;
1482     }
1483     return bestPattern;
1484 }
1485 
1486 UnicodeString
adjustFieldTypes(const UnicodeString & pattern,const PtnSkeleton * specifiedSkeleton,int32_t flags,UDateTimePatternMatchOptions options)1487 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
1488                                            const PtnSkeleton* specifiedSkeleton,
1489                                            int32_t flags,
1490                                            UDateTimePatternMatchOptions options) {
1491     UnicodeString newPattern;
1492     fp->set(pattern);
1493     for (int32_t i=0; i < fp->itemNumber; i++) {
1494         UnicodeString field = fp->items[i];
1495         if ( fp->isQuoteLiteral(field) ) {
1496 
1497             UnicodeString quoteLiteral;
1498             fp->getQuoteLiteral(quoteLiteral, &i);
1499             newPattern += quoteLiteral;
1500         }
1501         else {
1502             if (fp->isPatternSeparator(field)) {
1503                 newPattern+=field;
1504                 continue;
1505             }
1506             int32_t canonicalIndex = fp->getCanonicalIndex(field);
1507             if (canonicalIndex < 0) {
1508                 newPattern+=field;
1509                 continue;  // don't adjust
1510             }
1511             const dtTypeElem *row = &dtTypes[canonicalIndex];
1512             int32_t typeValue = row->field;
1513 
1514             // handle day periods - with #13183, no longer need special handling here, integrated with normal types
1515 
1516             if ((flags & kDTPGFixFractionalSeconds) != 0 && typeValue == UDATPG_SECOND_FIELD) {
1517                 field += decimal;
1518                 dtMatcher->skeleton.original.appendFieldTo(UDATPG_FRACTIONAL_SECOND_FIELD, field);
1519             } else if (dtMatcher->skeleton.type[typeValue]!=0) {
1520                     // Here:
1521                     // - "reqField" is the field from the originally requested skeleton, with length
1522                     // "reqFieldLen".
1523                     // - "field" is the field from the found pattern.
1524                     //
1525                     // The adjusted field should consist of characters from the originally requested
1526                     // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1527                     // UDATPG_WEEKDAY_FIELD or UDATPG_YEAR_FIELD, in which case it should consist
1528                     // of characters from the  found pattern.
1529                     //
1530                     // The length of the adjusted field (adjFieldLen) should match that in the originally
1531                     // requested skeleton, except that in the following cases the length of the adjusted field
1532                     // should match that in the found pattern (i.e. the length of this pattern field should
1533                     // not be adjusted):
1534                     // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1535                     //    not set (ticket #7180). Note, we may want to implement a similar change for other
1536                     //    numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1537                     //    field length, but options bits can be used to override this.
1538                     // 2. There is a specified skeleton for the found pattern and one of the following is true:
1539                     //    a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1540                     //    b) The pattern field is numeric and the skeleton field is not, or vice versa.
1541 
1542                     UChar reqFieldChar = dtMatcher->skeleton.original.getFieldChar(typeValue);
1543                     int32_t reqFieldLen = dtMatcher->skeleton.original.getFieldLength(typeValue);
1544                     if (reqFieldChar == CAP_E && reqFieldLen < 3)
1545                         reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1546                     int32_t adjFieldLen = reqFieldLen;
1547                     if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1548                          (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1549                          (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1550                          adjFieldLen = field.length();
1551                     } else if (specifiedSkeleton) {
1552                         int32_t skelFieldLen = specifiedSkeleton->original.getFieldLength(typeValue);
1553                         UBool patFieldIsNumeric = (row->type > 0);
1554                         UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1555                         if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1556                             // don't adjust the field length in the found pattern
1557                             adjFieldLen = field.length();
1558                         }
1559                     }
1560                     UChar c = (typeValue!= UDATPG_HOUR_FIELD
1561                             && typeValue!= UDATPG_MONTH_FIELD
1562                             && typeValue!= UDATPG_WEEKDAY_FIELD
1563                             && (typeValue!= UDATPG_YEAR_FIELD || reqFieldChar==CAP_Y))
1564                             ? reqFieldChar
1565                             : field.charAt(0);
1566                     if (typeValue == UDATPG_HOUR_FIELD && (flags & kDTPGSkeletonUsesCapJ) != 0) {
1567                         c = fDefaultHourFormatChar;
1568                     }
1569                     field.remove();
1570                     for (int32_t j=adjFieldLen; j>0; --j) {
1571                         field += c;
1572                     }
1573             }
1574             newPattern+=field;
1575         }
1576     }
1577     return newPattern;
1578 }
1579 
1580 UnicodeString
getBestAppending(int32_t missingFields,int32_t flags,UErrorCode & status,UDateTimePatternMatchOptions options)1581 DateTimePatternGenerator::getBestAppending(int32_t missingFields, int32_t flags, UErrorCode &status, UDateTimePatternMatchOptions options) {
1582     if (U_FAILURE(status)) {
1583         return UnicodeString();
1584     }
1585     UnicodeString  resultPattern, tempPattern;
1586     const UnicodeString* tempPatternPtr;
1587     int32_t lastMissingFieldMask=0;
1588     if (missingFields!=0) {
1589         resultPattern=UnicodeString();
1590         const PtnSkeleton* specifiedSkeleton=nullptr;
1591         tempPatternPtr = getBestRaw(*dtMatcher, missingFields, distanceInfo, status, &specifiedSkeleton);
1592         if (U_FAILURE(status)) {
1593             return UnicodeString();
1594         }
1595         tempPattern = *tempPatternPtr;
1596         resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1597         if ( distanceInfo->missingFieldMask==0 ) {
1598             return resultPattern;
1599         }
1600         while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1601             if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1602                 break;  // cannot find the proper missing field
1603             }
1604             if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1605                 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1606                 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, flags | kDTPGFixFractionalSeconds, options);
1607                 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1608                 continue;
1609             }
1610             int32_t startingMask = distanceInfo->missingFieldMask;
1611             tempPatternPtr = getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, status, &specifiedSkeleton);
1612             if (U_FAILURE(status)) {
1613                 return UnicodeString();
1614             }
1615             tempPattern = *tempPatternPtr;
1616             tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1617             int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1618             int32_t topField=getTopBitNumber(foundMask);
1619 
1620             if (appendItemFormats[topField].length() != 0) {
1621                 UnicodeString appendName;
1622                 getAppendName((UDateTimePatternField)topField, appendName);
1623                 const UnicodeString *values[3] = {
1624                     &resultPattern,
1625                     &tempPattern,
1626                     &appendName
1627                 };
1628                 SimpleFormatter(appendItemFormats[topField], 2, 3, status).
1629                     formatAndReplace(values, 3, resultPattern, nullptr, 0, status);
1630             }
1631             lastMissingFieldMask = distanceInfo->missingFieldMask;
1632         }
1633     }
1634     return resultPattern;
1635 }
1636 
1637 int32_t
getTopBitNumber(int32_t foundMask) const1638 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) const {
1639     if ( foundMask==0 ) {
1640         return 0;
1641     }
1642     int32_t i=0;
1643     while (foundMask!=0) {
1644         foundMask >>=1;
1645         ++i;
1646     }
1647     if (i-1 >UDATPG_ZONE_FIELD) {
1648         return UDATPG_ZONE_FIELD;
1649     }
1650     else
1651         return i-1;
1652 }
1653 
1654 void
setAvailableFormat(const UnicodeString & key,UErrorCode & err)1655 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1656 {
1657     fAvailableFormatKeyHash->puti(key, 1, err);
1658 }
1659 
1660 UBool
isAvailableFormatSet(const UnicodeString & key) const1661 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1662     return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1663 }
1664 
1665 void
copyHashtable(Hashtable * other,UErrorCode & status)1666 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1667     if (other == nullptr || U_FAILURE(status)) {
1668         return;
1669     }
1670     if (fAvailableFormatKeyHash != nullptr) {
1671         delete fAvailableFormatKeyHash;
1672         fAvailableFormatKeyHash = nullptr;
1673     }
1674     initHashtable(status);
1675     if(U_FAILURE(status)){
1676         return;
1677     }
1678     int32_t pos = UHASH_FIRST;
1679     const UHashElement* elem = nullptr;
1680     // walk through the hash table and create a deep clone
1681     while((elem = other->nextElement(pos))!= nullptr){
1682         const UHashTok otherKeyTok = elem->key;
1683         UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1684         fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1685         if(U_FAILURE(status)){
1686             return;
1687         }
1688     }
1689 }
1690 
1691 StringEnumeration*
getSkeletons(UErrorCode & status) const1692 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1693     if (U_FAILURE(status)) {
1694         return nullptr;
1695     }
1696     if (U_FAILURE(internalErrorCode)) {
1697         status = internalErrorCode;
1698         return nullptr;
1699     }
1700     LocalPointer<StringEnumeration> skeletonEnumerator(
1701         new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status), status);
1702 
1703     return U_SUCCESS(status) ? skeletonEnumerator.orphan() : nullptr;
1704 }
1705 
1706 const UnicodeString&
getPatternForSkeleton(const UnicodeString & skeleton) const1707 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1708     PtnElem *curElem;
1709 
1710     if (skeleton.length() ==0) {
1711         return emptyString;
1712     }
1713     curElem = patternMap->getHeader(skeleton.charAt(0));
1714     while ( curElem != nullptr ) {
1715         if ( curElem->skeleton->getSkeleton()==skeleton ) {
1716             return curElem->pattern;
1717         }
1718         curElem = curElem->next.getAlias();
1719     }
1720     return emptyString;
1721 }
1722 
1723 StringEnumeration*
getBaseSkeletons(UErrorCode & status) const1724 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1725     if (U_FAILURE(status)) {
1726         return nullptr;
1727     }
1728     if (U_FAILURE(internalErrorCode)) {
1729         status = internalErrorCode;
1730         return nullptr;
1731     }
1732     LocalPointer<StringEnumeration> baseSkeletonEnumerator(
1733         new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status), status);
1734 
1735     return U_SUCCESS(status) ? baseSkeletonEnumerator.orphan() : nullptr;
1736 }
1737 
1738 StringEnumeration*
getRedundants(UErrorCode & status)1739 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1740     if (U_FAILURE(status)) { return nullptr; }
1741     if (U_FAILURE(internalErrorCode)) {
1742         status = internalErrorCode;
1743         return nullptr;
1744     }
1745     LocalPointer<StringEnumeration> output(new DTRedundantEnumeration(), status);
1746     if (U_FAILURE(status)) { return nullptr; }
1747     const UnicodeString *pattern;
1748     PatternMapIterator it(status);
1749     if (U_FAILURE(status)) { return nullptr; }
1750 
1751     for (it.set(*patternMap); it.hasNext(); ) {
1752         DateTimeMatcher current = it.next();
1753         pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1754         if ( isCanonicalItem(*pattern) ) {
1755             continue;
1756         }
1757         if ( skipMatcher == nullptr ) {
1758             skipMatcher = new DateTimeMatcher(current);
1759             if (skipMatcher == nullptr) {
1760                 status = U_MEMORY_ALLOCATION_ERROR;
1761                 return nullptr;
1762             }
1763         }
1764         else {
1765             *skipMatcher = current;
1766         }
1767         UnicodeString trial = getBestPattern(current.getPattern(), status);
1768         if (U_FAILURE(status)) { return nullptr; }
1769         if (trial == *pattern) {
1770             ((DTRedundantEnumeration *)output.getAlias())->add(*pattern, status);
1771             if (U_FAILURE(status)) { return nullptr; }
1772         }
1773         if (current.equals(skipMatcher)) {
1774             continue;
1775         }
1776     }
1777     return output.orphan();
1778 }
1779 
1780 UBool
isCanonicalItem(const UnicodeString & item) const1781 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1782     if ( item.length() != 1 ) {
1783         return FALSE;
1784     }
1785     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1786         if (item.charAt(0)==Canonical_Items[i]) {
1787             return TRUE;
1788         }
1789     }
1790     return FALSE;
1791 }
1792 
1793 
1794 DateTimePatternGenerator*
clone() const1795 DateTimePatternGenerator::clone() const {
1796     return new DateTimePatternGenerator(*this);
1797 }
1798 
PatternMap()1799 PatternMap::PatternMap() {
1800    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1801        boot[i] = nullptr;
1802    }
1803    isDupAllowed = TRUE;
1804 }
1805 
1806 void
copyFrom(const PatternMap & other,UErrorCode & status)1807 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1808     if (U_FAILURE(status)) {
1809         return;
1810     }
1811     this->isDupAllowed = other.isDupAllowed;
1812     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
1813         PtnElem *curElem, *otherElem, *prevElem=nullptr;
1814         otherElem = other.boot[bootIndex];
1815         while (otherElem != nullptr) {
1816             LocalPointer<PtnElem> newElem(new PtnElem(otherElem->basePattern, otherElem->pattern), status);
1817             if (U_FAILURE(status)) {
1818                 return; // out of memory
1819             }
1820             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(*(otherElem->skeleton)), status);
1821             if (U_FAILURE(status)) {
1822                 return; // out of memory
1823             }
1824             newElem->skeletonWasSpecified = otherElem->skeletonWasSpecified;
1825 
1826             // Release ownership from the LocalPointer of the PtnElem object.
1827             // The PtnElem will now be owned by either the boot (for the first entry in the linked-list)
1828             // or owned by the previous PtnElem object in the linked-list.
1829             curElem = newElem.orphan();
1830 
1831             if (this->boot[bootIndex] == nullptr) {
1832                 this->boot[bootIndex] = curElem;
1833             } else {
1834                 if (prevElem != nullptr) {
1835                     prevElem->next.adoptInstead(curElem);
1836                 } else {
1837                     UPRV_UNREACHABLE;
1838                 }
1839             }
1840             prevElem = curElem;
1841             otherElem = otherElem->next.getAlias();
1842         }
1843 
1844     }
1845 }
1846 
1847 PtnElem*
getHeader(UChar baseChar) const1848 PatternMap::getHeader(UChar baseChar) const {
1849     PtnElem* curElem;
1850 
1851     if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1852          curElem = boot[baseChar-CAP_A];
1853     }
1854     else {
1855         if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1856             curElem = boot[26+baseChar-LOW_A];
1857         }
1858         else {
1859             return nullptr;
1860         }
1861     }
1862     return curElem;
1863 }
1864 
~PatternMap()1865 PatternMap::~PatternMap() {
1866    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1867        if (boot[i] != nullptr ) {
1868            delete boot[i];
1869            boot[i] = nullptr;
1870        }
1871    }
1872 }  // PatternMap destructor
1873 
1874 void
add(const UnicodeString & basePattern,const PtnSkeleton & skeleton,const UnicodeString & value,UBool skeletonWasSpecified,UErrorCode & status)1875 PatternMap::add(const UnicodeString& basePattern,
1876                 const PtnSkeleton& skeleton,
1877                 const UnicodeString& value,// mapped pattern value
1878                 UBool skeletonWasSpecified,
1879                 UErrorCode &status) {
1880     UChar baseChar = basePattern.charAt(0);
1881     PtnElem *curElem, *baseElem;
1882     status = U_ZERO_ERROR;
1883 
1884     // the baseChar must be A-Z or a-z
1885     if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1886         baseElem = boot[baseChar-CAP_A];
1887     }
1888     else {
1889         if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1890             baseElem = boot[26+baseChar-LOW_A];
1891          }
1892          else {
1893              status = U_ILLEGAL_CHARACTER;
1894              return;
1895          }
1896     }
1897 
1898     if (baseElem == nullptr) {
1899         LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1900         if (U_FAILURE(status)) {
1901             return; // out of memory
1902         }
1903         newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1904         if (U_FAILURE(status)) {
1905             return; // out of memory
1906         }
1907         newElem->skeletonWasSpecified = skeletonWasSpecified;
1908         if (baseChar >= LOW_A) {
1909             boot[26 + (baseChar - LOW_A)] = newElem.orphan(); // the boot array now owns the PtnElem.
1910         }
1911         else {
1912             boot[baseChar - CAP_A] = newElem.orphan(); // the boot array now owns the PtnElem.
1913         }
1914     }
1915     if ( baseElem != nullptr ) {
1916         curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1917 
1918         if (curElem == nullptr) {
1919             // add new element to the list.
1920             curElem = baseElem;
1921             while( curElem -> next != nullptr )
1922             {
1923                 curElem = curElem->next.getAlias();
1924             }
1925 
1926             LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1927             if (U_FAILURE(status)) {
1928                 return; // out of memory
1929             }
1930             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1931             if (U_FAILURE(status)) {
1932                 return; // out of memory
1933             }
1934             newElem->skeletonWasSpecified = skeletonWasSpecified;
1935             curElem->next.adoptInstead(newElem.orphan());
1936             curElem = curElem->next.getAlias();
1937         }
1938         else {
1939             // Pattern exists in the list already.
1940             if ( !isDupAllowed ) {
1941                 return;
1942             }
1943             // Overwrite the value.
1944             curElem->pattern = value;
1945             // It was a bug that we were not doing the following previously,
1946             // though that bug hid other problems by making things partly work.
1947             curElem->skeletonWasSpecified = skeletonWasSpecified;
1948         }
1949     }
1950 }  // PatternMap::add
1951 
1952 // Find the pattern from the given basePattern string.
1953 const UnicodeString *
getPatternFromBasePattern(const UnicodeString & basePattern,UBool & skeletonWasSpecified) const1954 PatternMap::getPatternFromBasePattern(const UnicodeString& basePattern, UBool& skeletonWasSpecified) const { // key to search for
1955    PtnElem *curElem;
1956 
1957    if ((curElem=getHeader(basePattern.charAt(0)))==nullptr) {
1958        return nullptr;  // no match
1959    }
1960 
1961    do  {
1962        if ( basePattern.compare(curElem->basePattern)==0 ) {
1963           skeletonWasSpecified = curElem->skeletonWasSpecified;
1964           return &(curElem->pattern);
1965        }
1966        curElem = curElem->next.getAlias();
1967    } while (curElem != nullptr);
1968 
1969    return nullptr;
1970 }  // PatternMap::getFromBasePattern
1971 
1972 
1973 // Find the pattern from the given skeleton.
1974 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
1975 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
1976 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
1977 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
1978 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
1979 const UnicodeString *
getPatternFromSkeleton(const PtnSkeleton & skeleton,const PtnSkeleton ** specifiedSkeletonPtr) const1980 PatternMap::getPatternFromSkeleton(const PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) const { // key to search for
1981    PtnElem *curElem;
1982 
1983    if (specifiedSkeletonPtr) {
1984        *specifiedSkeletonPtr = nullptr;
1985    }
1986 
1987    // find boot entry
1988    UChar baseChar = skeleton.getFirstChar();
1989    if ((curElem=getHeader(baseChar))==nullptr) {
1990        return nullptr;  // no match
1991    }
1992 
1993    do  {
1994        UBool equal;
1995        if (specifiedSkeletonPtr != nullptr) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
1996            equal = curElem->skeleton->original == skeleton.original;
1997        } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
1998            equal = curElem->skeleton->baseOriginal == skeleton.baseOriginal;
1999        }
2000        if (equal) {
2001            if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
2002                *specifiedSkeletonPtr = curElem->skeleton.getAlias();
2003            }
2004            return &(curElem->pattern);
2005        }
2006        curElem = curElem->next.getAlias();
2007    } while (curElem != nullptr);
2008 
2009    return nullptr;
2010 }
2011 
2012 UBool
equals(const PatternMap & other) const2013 PatternMap::equals(const PatternMap& other) const {
2014     if ( this==&other ) {
2015         return TRUE;
2016     }
2017     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
2018         if (boot[bootIndex] == other.boot[bootIndex]) {
2019             continue;
2020         }
2021         if ((boot[bootIndex] == nullptr) || (other.boot[bootIndex] == nullptr)) {
2022             return FALSE;
2023         }
2024         PtnElem *otherElem = other.boot[bootIndex];
2025         PtnElem *myElem = boot[bootIndex];
2026         while ((otherElem != nullptr) || (myElem != nullptr)) {
2027             if ( myElem == otherElem ) {
2028                 break;
2029             }
2030             if ((otherElem == nullptr) || (myElem == nullptr)) {
2031                 return FALSE;
2032             }
2033             if ( (myElem->basePattern != otherElem->basePattern) ||
2034                  (myElem->pattern != otherElem->pattern) ) {
2035                 return FALSE;
2036             }
2037             if ((myElem->skeleton.getAlias() != otherElem->skeleton.getAlias()) &&
2038                 !myElem->skeleton->equals(*(otherElem->skeleton))) {
2039                 return FALSE;
2040             }
2041             myElem = myElem->next.getAlias();
2042             otherElem = otherElem->next.getAlias();
2043         }
2044     }
2045     return TRUE;
2046 }
2047 
2048 // find any key existing in the mapping table already.
2049 // return TRUE if there is an existing key, otherwise return FALSE.
2050 PtnElem*
getDuplicateElem(const UnicodeString & basePattern,const PtnSkeleton & skeleton,PtnElem * baseElem)2051 PatternMap::getDuplicateElem(
2052             const UnicodeString &basePattern,
2053             const PtnSkeleton &skeleton,
2054             PtnElem *baseElem) {
2055    PtnElem *curElem;
2056 
2057    if ( baseElem == nullptr ) {
2058          return nullptr;
2059    }
2060    else {
2061          curElem = baseElem;
2062    }
2063    do {
2064      if ( basePattern.compare(curElem->basePattern)==0 ) {
2065          UBool isEqual = TRUE;
2066          for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2067             if (curElem->skeleton->type[i] != skeleton.type[i] ) {
2068                 isEqual = FALSE;
2069                 break;
2070             }
2071         }
2072         if (isEqual) {
2073             return curElem;
2074         }
2075      }
2076      curElem = curElem->next.getAlias();
2077    } while( curElem != nullptr );
2078 
2079    // end of the list
2080    return nullptr;
2081 
2082 }  // PatternMap::getDuplicateElem
2083 
DateTimeMatcher(void)2084 DateTimeMatcher::DateTimeMatcher(void) {
2085 }
2086 
~DateTimeMatcher()2087 DateTimeMatcher::~DateTimeMatcher() {}
2088 
DateTimeMatcher(const DateTimeMatcher & other)2089 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
2090     copyFrom(other.skeleton);
2091 }
2092 
2093 
2094 void
set(const UnicodeString & pattern,FormatParser * fp)2095 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
2096     PtnSkeleton localSkeleton;
2097     return set(pattern, fp, localSkeleton);
2098 }
2099 
2100 void
set(const UnicodeString & pattern,FormatParser * fp,PtnSkeleton & skeletonResult)2101 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
2102     int32_t i;
2103     for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
2104         skeletonResult.type[i] = NONE;
2105     }
2106     skeletonResult.original.clear();
2107     skeletonResult.baseOriginal.clear();
2108     skeletonResult.addedDefaultDayPeriod = FALSE;
2109 
2110     fp->set(pattern);
2111     for (i=0; i < fp->itemNumber; i++) {
2112         const UnicodeString& value = fp->items[i];
2113         // don't skip 'a' anymore, dayPeriod handled specially below
2114 
2115         if ( fp->isQuoteLiteral(value) ) {
2116             UnicodeString quoteLiteral;
2117             fp->getQuoteLiteral(quoteLiteral, &i);
2118             continue;
2119         }
2120         int32_t canonicalIndex = fp->getCanonicalIndex(value);
2121         if (canonicalIndex < 0) {
2122             continue;
2123         }
2124         const dtTypeElem *row = &dtTypes[canonicalIndex];
2125         int32_t field = row->field;
2126         skeletonResult.original.populate(field, value);
2127         UChar repeatChar = row->patternChar;
2128         int32_t repeatCount = row->minLen;
2129         skeletonResult.baseOriginal.populate(field, repeatChar, repeatCount);
2130         int16_t subField = row->type;
2131         if (row->type > 0) {
2132             U_ASSERT(value.length() < INT16_MAX);
2133             subField += static_cast<int16_t>(value.length());
2134         }
2135         skeletonResult.type[field] = subField;
2136     }
2137     // #13183, handle special behavior for day period characters (a, b, B)
2138     if (!skeletonResult.original.isFieldEmpty(UDATPG_HOUR_FIELD)) {
2139         if (skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==LOW_H || skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==CAP_K) {
2140             // We have a skeleton with 12-hour-cycle format
2141             if (skeletonResult.original.isFieldEmpty(UDATPG_DAYPERIOD_FIELD)) {
2142                 // But we do not have a day period in the skeleton; add the default DAYPERIOD (currently "a")
2143                 for (i = 0; dtTypes[i].patternChar != 0; i++) {
2144                     if ( dtTypes[i].field == UDATPG_DAYPERIOD_FIELD ) {
2145                         // first entry for UDATPG_DAYPERIOD_FIELD
2146                         skeletonResult.original.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2147                         skeletonResult.baseOriginal.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2148                         skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = dtTypes[i].type;
2149                         skeletonResult.addedDefaultDayPeriod = TRUE;
2150                         break;
2151                     }
2152                 }
2153             }
2154         } else {
2155             // Skeleton has 24-hour-cycle hour format and has dayPeriod, delete dayPeriod (i.e. ignore it)
2156             skeletonResult.original.clearField(UDATPG_DAYPERIOD_FIELD);
2157             skeletonResult.baseOriginal.clearField(UDATPG_DAYPERIOD_FIELD);
2158             skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = NONE;
2159         }
2160     }
2161     copyFrom(skeletonResult);
2162 }
2163 
2164 void
getBasePattern(UnicodeString & result)2165 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
2166     result.remove(); // Reset the result first.
2167     skeleton.baseOriginal.appendTo(result);
2168 }
2169 
2170 UnicodeString
getPattern()2171 DateTimeMatcher::getPattern() {
2172     UnicodeString result;
2173     return skeleton.original.appendTo(result);
2174 }
2175 
2176 int32_t
getDistance(const DateTimeMatcher & other,int32_t includeMask,DistanceInfo & distanceInfo) const2177 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) const {
2178     int32_t result = 0;
2179     distanceInfo.clear();
2180     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
2181         int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
2182         int32_t otherType = other.skeleton.type[i];
2183         if (myType==otherType) {
2184             continue;
2185         }
2186         if (myType==0) {// and other is not
2187             result += EXTRA_FIELD;
2188             distanceInfo.addExtra(i);
2189         }
2190         else {
2191             if (otherType==0) {
2192                 result += MISSING_FIELD;
2193                 distanceInfo.addMissing(i);
2194             }
2195             else {
2196                 result += abs(myType - otherType);
2197             }
2198         }
2199 
2200     }
2201     return result;
2202 }
2203 
2204 void
copyFrom(const PtnSkeleton & newSkeleton)2205 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
2206     skeleton.copyFrom(newSkeleton);
2207 }
2208 
2209 void
copyFrom()2210 DateTimeMatcher::copyFrom() {
2211     // same as clear
2212     skeleton.clear();
2213 }
2214 
2215 UBool
equals(const DateTimeMatcher * other) const2216 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
2217     if (other==nullptr) { return FALSE; }
2218     return skeleton.original == other->skeleton.original;
2219 }
2220 
2221 int32_t
getFieldMask() const2222 DateTimeMatcher::getFieldMask() const {
2223     int32_t result = 0;
2224 
2225     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2226         if (skeleton.type[i]!=0) {
2227             result |= (1<<i);
2228         }
2229     }
2230     return result;
2231 }
2232 
2233 PtnSkeleton*
getSkeletonPtr()2234 DateTimeMatcher::getSkeletonPtr() {
2235     return &skeleton;
2236 }
2237 
FormatParser()2238 FormatParser::FormatParser () {
2239     status = START;
2240     itemNumber = 0;
2241 }
2242 
2243 
~FormatParser()2244 FormatParser::~FormatParser () {
2245 }
2246 
2247 
2248 // Find the next token with the starting position and length
2249 // Note: the startPos may
2250 FormatParser::TokenStatus
setTokens(const UnicodeString & pattern,int32_t startPos,int32_t * len)2251 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
2252     int32_t curLoc = startPos;
2253     if ( curLoc >= pattern.length()) {
2254         return DONE;
2255     }
2256     // check the current char is between A-Z or a-z
2257     do {
2258         UChar c=pattern.charAt(curLoc);
2259         if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
2260            curLoc++;
2261         }
2262         else {
2263                startPos = curLoc;
2264                *len=1;
2265                return ADD_TOKEN;
2266         }
2267 
2268         if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
2269             break;  // not the same token
2270         }
2271     } while(curLoc <= pattern.length());
2272     *len = curLoc-startPos;
2273     return ADD_TOKEN;
2274 }
2275 
2276 void
set(const UnicodeString & pattern)2277 FormatParser::set(const UnicodeString& pattern) {
2278     int32_t startPos = 0;
2279     TokenStatus result = START;
2280     int32_t len = 0;
2281     itemNumber = 0;
2282 
2283     do {
2284         result = setTokens( pattern, startPos, &len );
2285         if ( result == ADD_TOKEN )
2286         {
2287             items[itemNumber++] = UnicodeString(pattern, startPos, len );
2288             startPos += len;
2289         }
2290         else {
2291             break;
2292         }
2293     } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
2294 }
2295 
2296 int32_t
getCanonicalIndex(const UnicodeString & s,UBool strict)2297 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
2298     int32_t len = s.length();
2299     if (len == 0) {
2300         return -1;
2301     }
2302     UChar ch = s.charAt(0);
2303 
2304     // Verify that all are the same character.
2305     for (int32_t l = 1; l < len; l++) {
2306         if (ch != s.charAt(l)) {
2307             return -1;
2308         }
2309     }
2310     int32_t i = 0;
2311     int32_t bestRow = -1;
2312     while (dtTypes[i].patternChar != 0x0000) {
2313         if ( dtTypes[i].patternChar != ch ) {
2314             ++i;
2315             continue;
2316         }
2317         bestRow = i;
2318         if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
2319             return i;
2320         }
2321         if (dtTypes[i+1].minLen <= len) {
2322             ++i;
2323             continue;
2324         }
2325         return i;
2326     }
2327     return strict ? -1 : bestRow;
2328 }
2329 
2330 UBool
isQuoteLiteral(const UnicodeString & s)2331 FormatParser::isQuoteLiteral(const UnicodeString& s) {
2332     return (UBool)(s.charAt(0) == SINGLE_QUOTE);
2333 }
2334 
2335 // This function assumes the current itemIndex points to the quote literal.
2336 // Please call isQuoteLiteral prior to this function.
2337 void
getQuoteLiteral(UnicodeString & quote,int32_t * itemIndex)2338 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
2339     int32_t i = *itemIndex;
2340 
2341     quote.remove();
2342     if (items[i].charAt(0)==SINGLE_QUOTE) {
2343         quote += items[i];
2344         ++i;
2345     }
2346     while ( i < itemNumber ) {
2347         if ( items[i].charAt(0)==SINGLE_QUOTE ) {
2348             if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
2349                 // two single quotes e.g. 'o''clock'
2350                 quote += items[i++];
2351                 quote += items[i++];
2352                 continue;
2353             }
2354             else {
2355                 quote += items[i];
2356                 break;
2357             }
2358         }
2359         else {
2360             quote += items[i];
2361         }
2362         ++i;
2363     }
2364     *itemIndex=i;
2365 }
2366 
2367 UBool
isPatternSeparator(const UnicodeString & field) const2368 FormatParser::isPatternSeparator(const UnicodeString& field) const {
2369     for (int32_t i=0; i<field.length(); ++i ) {
2370         UChar c= field.charAt(i);
2371         if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
2372              (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
2373             continue;
2374         }
2375         else {
2376             return FALSE;
2377         }
2378     }
2379     return TRUE;
2380 }
2381 
~DistanceInfo()2382 DistanceInfo::~DistanceInfo() {}
2383 
2384 void
setTo(const DistanceInfo & other)2385 DistanceInfo::setTo(const DistanceInfo& other) {
2386     missingFieldMask = other.missingFieldMask;
2387     extraFieldMask= other.extraFieldMask;
2388 }
2389 
PatternMapIterator(UErrorCode & status)2390 PatternMapIterator::PatternMapIterator(UErrorCode& status) :
2391     bootIndex(0), nodePtr(nullptr), matcher(nullptr), patternMap(nullptr)
2392 {
2393     if (U_FAILURE(status)) { return; }
2394     matcher.adoptInsteadAndCheckErrorCode(new DateTimeMatcher(), status);
2395 }
2396 
~PatternMapIterator()2397 PatternMapIterator::~PatternMapIterator() {
2398 }
2399 
2400 void
set(PatternMap & newPatternMap)2401 PatternMapIterator::set(PatternMap& newPatternMap) {
2402     this->patternMap=&newPatternMap;
2403 }
2404 
2405 PtnSkeleton*
getSkeleton() const2406 PatternMapIterator::getSkeleton() const {
2407     if ( nodePtr == nullptr ) {
2408         return nullptr;
2409     }
2410     else {
2411         return nodePtr->skeleton.getAlias();
2412     }
2413 }
2414 
2415 UBool
hasNext() const2416 PatternMapIterator::hasNext() const {
2417     int32_t headIndex = bootIndex;
2418     PtnElem *curPtr = nodePtr;
2419 
2420     if (patternMap==nullptr) {
2421         return FALSE;
2422     }
2423     while ( headIndex < MAX_PATTERN_ENTRIES ) {
2424         if ( curPtr != nullptr ) {
2425             if ( curPtr->next != nullptr ) {
2426                 return TRUE;
2427             }
2428             else {
2429                 headIndex++;
2430                 curPtr=nullptr;
2431                 continue;
2432             }
2433         }
2434         else {
2435             if ( patternMap->boot[headIndex] != nullptr ) {
2436                 return TRUE;
2437             }
2438             else {
2439                 headIndex++;
2440                 continue;
2441             }
2442         }
2443     }
2444     return FALSE;
2445 }
2446 
2447 DateTimeMatcher&
next()2448 PatternMapIterator::next() {
2449     while ( bootIndex < MAX_PATTERN_ENTRIES ) {
2450         if ( nodePtr != nullptr ) {
2451             if ( nodePtr->next != nullptr ) {
2452                 nodePtr = nodePtr->next.getAlias();
2453                 break;
2454             }
2455             else {
2456                 bootIndex++;
2457                 nodePtr=nullptr;
2458                 continue;
2459             }
2460         }
2461         else {
2462             if ( patternMap->boot[bootIndex] != nullptr ) {
2463                 nodePtr = patternMap->boot[bootIndex];
2464                 break;
2465             }
2466             else {
2467                 bootIndex++;
2468                 continue;
2469             }
2470         }
2471     }
2472     if (nodePtr!=nullptr) {
2473         matcher->copyFrom(*nodePtr->skeleton);
2474     }
2475     else {
2476         matcher->copyFrom();
2477     }
2478     return *matcher;
2479 }
2480 
2481 
SkeletonFields()2482 SkeletonFields::SkeletonFields() {
2483     // Set initial values to zero
2484     clear();
2485 }
2486 
clear()2487 void SkeletonFields::clear() {
2488     uprv_memset(chars, 0, sizeof(chars));
2489     uprv_memset(lengths, 0, sizeof(lengths));
2490 }
2491 
copyFrom(const SkeletonFields & other)2492 void SkeletonFields::copyFrom(const SkeletonFields& other) {
2493     uprv_memcpy(chars, other.chars, sizeof(chars));
2494     uprv_memcpy(lengths, other.lengths, sizeof(lengths));
2495 }
2496 
clearField(int32_t field)2497 void SkeletonFields::clearField(int32_t field) {
2498     chars[field] = 0;
2499     lengths[field] = 0;
2500 }
2501 
getFieldChar(int32_t field) const2502 UChar SkeletonFields::getFieldChar(int32_t field) const {
2503     return chars[field];
2504 }
2505 
getFieldLength(int32_t field) const2506 int32_t SkeletonFields::getFieldLength(int32_t field) const {
2507     return lengths[field];
2508 }
2509 
populate(int32_t field,const UnicodeString & value)2510 void SkeletonFields::populate(int32_t field, const UnicodeString& value) {
2511     populate(field, value.charAt(0), value.length());
2512 }
2513 
populate(int32_t field,UChar ch,int32_t length)2514 void SkeletonFields::populate(int32_t field, UChar ch, int32_t length) {
2515     chars[field] = (int8_t) ch;
2516     lengths[field] = (int8_t) length;
2517 }
2518 
isFieldEmpty(int32_t field) const2519 UBool SkeletonFields::isFieldEmpty(int32_t field) const {
2520     return lengths[field] == 0;
2521 }
2522 
appendTo(UnicodeString & string) const2523 UnicodeString& SkeletonFields::appendTo(UnicodeString& string) const {
2524     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2525         appendFieldTo(i, string);
2526     }
2527     return string;
2528 }
2529 
appendFieldTo(int32_t field,UnicodeString & string) const2530 UnicodeString& SkeletonFields::appendFieldTo(int32_t field, UnicodeString& string) const {
2531     UChar ch(chars[field]);
2532     int32_t length = (int32_t) lengths[field];
2533 
2534     for (int32_t i=0; i<length; i++) {
2535         string += ch;
2536     }
2537     return string;
2538 }
2539 
getFirstChar() const2540 UChar SkeletonFields::getFirstChar() const {
2541     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2542         if (lengths[i] != 0) {
2543             return chars[i];
2544         }
2545     }
2546     return '\0';
2547 }
2548 
2549 
PtnSkeleton()2550 PtnSkeleton::PtnSkeleton() {
2551 }
2552 
PtnSkeleton(const PtnSkeleton & other)2553 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
2554     copyFrom(other);
2555 }
2556 
copyFrom(const PtnSkeleton & other)2557 void PtnSkeleton::copyFrom(const PtnSkeleton& other) {
2558     uprv_memcpy(type, other.type, sizeof(type));
2559     original.copyFrom(other.original);
2560     baseOriginal.copyFrom(other.baseOriginal);
2561 }
2562 
clear()2563 void PtnSkeleton::clear() {
2564     uprv_memset(type, 0, sizeof(type));
2565     original.clear();
2566     baseOriginal.clear();
2567 }
2568 
2569 UBool
equals(const PtnSkeleton & other) const2570 PtnSkeleton::equals(const PtnSkeleton& other) const  {
2571     return (original == other.original)
2572         && (baseOriginal == other.baseOriginal)
2573         && (uprv_memcmp(type, other.type, sizeof(type)) == 0);
2574 }
2575 
2576 UnicodeString
getSkeleton() const2577 PtnSkeleton::getSkeleton() const {
2578     UnicodeString result;
2579     result = original.appendTo(result);
2580     int32_t pos;
2581     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2582         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2583         // was not in the provided skeleton, remove it here before returning skeleton.
2584         result.remove(pos, 1);
2585     }
2586     return result;
2587 }
2588 
2589 UnicodeString
getBaseSkeleton() const2590 PtnSkeleton::getBaseSkeleton() const {
2591     UnicodeString result;
2592     result = baseOriginal.appendTo(result);
2593     int32_t pos;
2594     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2595         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2596         // was not in the provided skeleton, remove it here before returning skeleton.
2597         result.remove(pos, 1);
2598     }
2599     return result;
2600 }
2601 
2602 UChar
getFirstChar() const2603 PtnSkeleton::getFirstChar() const {
2604     return baseOriginal.getFirstChar();
2605 }
2606 
~PtnSkeleton()2607 PtnSkeleton::~PtnSkeleton() {
2608 }
2609 
PtnElem(const UnicodeString & basePat,const UnicodeString & pat)2610 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
2611     basePattern(basePat), skeleton(nullptr), pattern(pat), next(nullptr)
2612 {
2613 }
2614 
~PtnElem()2615 PtnElem::~PtnElem() {
2616 }
2617 
DTSkeletonEnumeration(PatternMap & patternMap,dtStrEnum type,UErrorCode & status)2618 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap& patternMap, dtStrEnum type, UErrorCode& status) : fSkeletons(nullptr) {
2619     PtnElem  *curElem;
2620     PtnSkeleton *curSkeleton;
2621     UnicodeString s;
2622     int32_t bootIndex;
2623 
2624     pos=0;
2625     fSkeletons.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2626     if (U_FAILURE(status)) {
2627         return;
2628     }
2629 
2630     for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2631         curElem = patternMap.boot[bootIndex];
2632         while (curElem!=nullptr) {
2633             switch(type) {
2634                 case DT_BASESKELETON:
2635                     s=curElem->basePattern;
2636                     break;
2637                 case DT_PATTERN:
2638                     s=curElem->pattern;
2639                     break;
2640                 case DT_SKELETON:
2641                     curSkeleton=curElem->skeleton.getAlias();
2642                     s=curSkeleton->getSkeleton();
2643                     break;
2644             }
2645             if ( !isCanonicalItem(s) ) {
2646                 LocalPointer<UnicodeString> newElem(new UnicodeString(s), status);
2647                 if (U_FAILURE(status)) {
2648                     return;
2649                 }
2650                 fSkeletons->addElement(newElem.getAlias(), status);
2651                 if (U_FAILURE(status)) {
2652                     fSkeletons.adoptInstead(nullptr);
2653                     return;
2654                 }
2655                 newElem.orphan(); // fSkeletons vector now owns the UnicodeString.
2656             }
2657             curElem = curElem->next.getAlias();
2658         }
2659     }
2660     if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=nullptr) ) {
2661         status = U_BUFFER_OVERFLOW_ERROR;
2662     }
2663 }
2664 
2665 const UnicodeString*
snext(UErrorCode & status)2666 DTSkeletonEnumeration::snext(UErrorCode& status) {
2667     if (U_SUCCESS(status) && fSkeletons.isValid() && pos < fSkeletons->size()) {
2668         return (const UnicodeString*)fSkeletons->elementAt(pos++);
2669     }
2670     return nullptr;
2671 }
2672 
2673 void
reset(UErrorCode &)2674 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2675     pos=0;
2676 }
2677 
2678 int32_t
count(UErrorCode &) const2679 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2680    return (fSkeletons.isNull()) ? 0 : fSkeletons->size();
2681 }
2682 
2683 UBool
isCanonicalItem(const UnicodeString & item)2684 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2685     if ( item.length() != 1 ) {
2686         return FALSE;
2687     }
2688     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2689         if (item.charAt(0)==Canonical_Items[i]) {
2690             return TRUE;
2691         }
2692     }
2693     return FALSE;
2694 }
2695 
~DTSkeletonEnumeration()2696 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2697     UnicodeString *s;
2698     if (fSkeletons.isValid()) {
2699         for (int32_t i = 0; i < fSkeletons->size(); ++i) {
2700             if ((s = (UnicodeString *)fSkeletons->elementAt(i)) != nullptr) {
2701                 delete s;
2702             }
2703         }
2704     }
2705 }
2706 
DTRedundantEnumeration()2707 DTRedundantEnumeration::DTRedundantEnumeration() : pos(0), fPatterns(nullptr) {
2708 }
2709 
2710 void
add(const UnicodeString & pattern,UErrorCode & status)2711 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2712     if (U_FAILURE(status)) { return; }
2713     if (fPatterns.isNull())  {
2714         fPatterns.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2715         if (U_FAILURE(status)) {
2716             return;
2717        }
2718     }
2719     LocalPointer<UnicodeString> newElem(new UnicodeString(pattern), status);
2720     if (U_FAILURE(status)) {
2721         return;
2722     }
2723     fPatterns->addElement(newElem.getAlias(), status);
2724     if (U_FAILURE(status)) {
2725         fPatterns.adoptInstead(nullptr);
2726         return;
2727     }
2728     newElem.orphan(); // fPatterns now owns the string.
2729 }
2730 
2731 const UnicodeString*
snext(UErrorCode & status)2732 DTRedundantEnumeration::snext(UErrorCode& status) {
2733     if (U_SUCCESS(status) && fPatterns.isValid() && pos < fPatterns->size()) {
2734         return (const UnicodeString*)fPatterns->elementAt(pos++);
2735     }
2736     return nullptr;
2737 }
2738 
2739 void
reset(UErrorCode &)2740 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2741     pos=0;
2742 }
2743 
2744 int32_t
count(UErrorCode &) const2745 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2746     return (fPatterns.isNull()) ? 0 : fPatterns->size();
2747 }
2748 
2749 UBool
isCanonicalItem(const UnicodeString & item) const2750 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) const {
2751     if ( item.length() != 1 ) {
2752         return FALSE;
2753     }
2754     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2755         if (item.charAt(0)==Canonical_Items[i]) {
2756             return TRUE;
2757         }
2758     }
2759     return FALSE;
2760 }
2761 
~DTRedundantEnumeration()2762 DTRedundantEnumeration::~DTRedundantEnumeration() {
2763     UnicodeString *s;
2764     if (fPatterns.isValid()) {
2765         for (int32_t i = 0; i < fPatterns->size(); ++i) {
2766             if ((s = (UnicodeString *)fPatterns->elementAt(i)) != nullptr) {
2767                 delete s;
2768             }
2769         }
2770     }
2771 }
2772 
2773 U_NAMESPACE_END
2774 
2775 
2776 #endif /* #if !UCONFIG_NO_FORMATTING */
2777 
2778 //eof
2779