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__anon6ce53cd50211::AllowedHourFormatsSink508     AllowedHourFormatsSink() {}
509     virtual ~AllowedHourFormatsSink();
510 
put__anon6ce53cd50211::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__anon6ce53cd50211::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         char localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
791         // obtain a locale that always has the calendar key value that should be used
792         ures_getFunctionalEquivalent(
793             localeWithCalendarKey,
794             ULOC_LOCALE_IDENTIFIER_CAPACITY,
795             nullptr,
796             "calendar",
797             "calendar",
798             locale.getName(),
799             nullptr,
800             FALSE,
801             &err);
802         if (U_FAILURE(err)) { return; }
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             &err);
812         if (U_FAILURE(err)) { return; }
813         if (calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
814             destination.clear().append(calendarType, -1, err);
815             if (U_FAILURE(err)) { return; }
816         }
817         err = U_ZERO_ERROR;
818     }
819 }
820 
821 void
consumeShortTimePattern(const UnicodeString & shortTimePattern,UErrorCode & status)822 DateTimePatternGenerator::consumeShortTimePattern(const UnicodeString& shortTimePattern,
823         UErrorCode& status) {
824     if (U_FAILURE(status)) { return; }
825     // ICU-20383 No longer set fDefaultHourFormatChar to the hour format character from
826     // this pattern; instead it is set from localeToAllowedHourFormatsMap which now
827     // includes entries for both preferred and allowed formats.
828 
829     // HACK for hh:ss
830     hackTimes(shortTimePattern, status);
831 }
832 
833 struct DateTimePatternGenerator::AppendItemFormatsSink : public ResourceSink {
834 
835     // Destination for data, modified via setters.
836     DateTimePatternGenerator& dtpg;
837 
AppendItemFormatsSinkDateTimePatternGenerator::AppendItemFormatsSink838     AppendItemFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
839     virtual ~AppendItemFormatsSink();
840 
putDateTimePatternGenerator::AppendItemFormatsSink841     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
842             UErrorCode &errorCode) {
843         ResourceTable itemsTable = value.getTable(errorCode);
844         if (U_FAILURE(errorCode)) { return; }
845         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
846             UDateTimePatternField field = dtpg.getAppendFormatNumber(key);
847             if (field == UDATPG_FIELD_COUNT) { continue; }
848             const UnicodeString& valueStr = value.getUnicodeString(errorCode);
849             if (dtpg.getAppendItemFormat(field).isEmpty() && !valueStr.isEmpty()) {
850                 dtpg.setAppendItemFormat(field, valueStr);
851             }
852         }
853     }
854 
fillInMissingDateTimePatternGenerator::AppendItemFormatsSink855     void fillInMissing() {
856         UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, UPRV_LENGTHOF(UDATPG_ItemFormat)-1);  // Read-only alias.
857         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
858             UDateTimePatternField field = (UDateTimePatternField)i;
859             if (dtpg.getAppendItemFormat(field).isEmpty()) {
860                 dtpg.setAppendItemFormat(field, defaultItemFormat);
861             }
862         }
863     }
864 };
865 
866 struct DateTimePatternGenerator::AppendItemNamesSink : public ResourceSink {
867 
868     // Destination for data, modified via setters.
869     DateTimePatternGenerator& dtpg;
870 
AppendItemNamesSinkDateTimePatternGenerator::AppendItemNamesSink871     AppendItemNamesSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
872     virtual ~AppendItemNamesSink();
873 
putDateTimePatternGenerator::AppendItemNamesSink874     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
875             UErrorCode &errorCode) {
876         ResourceTable itemsTable = value.getTable(errorCode);
877         if (U_FAILURE(errorCode)) { return; }
878         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
879             UDateTimePGDisplayWidth width;
880             UDateTimePatternField field = dtpg.getFieldAndWidthIndices(key, &width);
881             if (field == UDATPG_FIELD_COUNT) { continue; }
882             ResourceTable detailsTable = value.getTable(errorCode);
883             if (U_FAILURE(errorCode)) { return; }
884             for (int32_t j = 0; detailsTable.getKeyAndValue(j, key, value); ++j) {
885                 if (uprv_strcmp(key, "dn") != 0) { continue; }
886                 const UnicodeString& valueStr = value.getUnicodeString(errorCode);
887                 if (dtpg.getFieldDisplayName(field,width).isEmpty() && !valueStr.isEmpty()) {
888                     dtpg.setFieldDisplayName(field,width,valueStr);
889                 }
890                 break;
891             }
892         }
893     }
894 
fillInMissingDateTimePatternGenerator::AppendItemNamesSink895     void fillInMissing() {
896         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
897             UnicodeString& valueStr = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, UDATPG_WIDE);
898             if (valueStr.isEmpty()) {
899                 valueStr = CAP_F;
900                 U_ASSERT(i < 20);
901                 if (i < 10) {
902                     // F0, F1, ..., F9
903                     valueStr += (UChar)(i+0x30);
904                 } else {
905                     // F10, F11, ...
906                     valueStr += (UChar)0x31;
907                     valueStr += (UChar)(i-10 + 0x30);
908                 }
909                 // NUL-terminate for the C API.
910                 valueStr.getTerminatedBuffer();
911             }
912             for (int32_t j = 1; j < UDATPG_WIDTH_COUNT; j++) {
913                 UnicodeString& valueStr2 = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)j);
914                 if (valueStr2.isEmpty()) {
915                     valueStr2 = dtpg.getFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)(j-1));
916                 }
917             }
918         }
919     }
920 };
921 
922 struct DateTimePatternGenerator::AvailableFormatsSink : public ResourceSink {
923 
924     // Destination for data, modified via setters.
925     DateTimePatternGenerator& dtpg;
926 
927     // Temporary variable, required for calling addPatternWithSkeleton.
928     UnicodeString conflictingPattern;
929 
AvailableFormatsSinkDateTimePatternGenerator::AvailableFormatsSink930     AvailableFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
931     virtual ~AvailableFormatsSink();
932 
putDateTimePatternGenerator::AvailableFormatsSink933     virtual void put(const char *key, ResourceValue &value, UBool isRoot,
934             UErrorCode &errorCode) {
935         ResourceTable itemsTable = value.getTable(errorCode);
936         if (U_FAILURE(errorCode)) { return; }
937         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
938             const UnicodeString formatKey(key, -1, US_INV);
939             if (!dtpg.isAvailableFormatSet(formatKey) ) {
940                 dtpg.setAvailableFormat(formatKey, errorCode);
941                 // Add pattern with its associated skeleton. Override any duplicate
942                 // derived from std patterns, but not a previous availableFormats entry:
943                 const UnicodeString& formatValue = value.getUnicodeString(errorCode);
944                 conflictingPattern.remove();
945                 dtpg.addPatternWithSkeleton(formatValue, &formatKey, !isRoot, conflictingPattern, errorCode);
946             }
947         }
948     }
949 };
950 
951 // Virtual destructors must be defined out of line.
~AppendItemFormatsSink()952 DateTimePatternGenerator::AppendItemFormatsSink::~AppendItemFormatsSink() {}
~AppendItemNamesSink()953 DateTimePatternGenerator::AppendItemNamesSink::~AppendItemNamesSink() {}
~AvailableFormatsSink()954 DateTimePatternGenerator::AvailableFormatsSink::~AvailableFormatsSink() {}
955 
956 void
addCLDRData(const Locale & locale,UErrorCode & errorCode)957 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& errorCode) {
958     if (U_FAILURE(errorCode)) { return; }
959     UnicodeString rbPattern, value, field;
960     CharString path;
961 
962     LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &errorCode));
963     if (U_FAILURE(errorCode)) { return; }
964 
965     CharString calendarTypeToUse; // to be filled in with the type to use, if all goes well
966     getCalendarTypeToUse(locale, calendarTypeToUse, errorCode);
967     if (U_FAILURE(errorCode)) { return; }
968 
969     // Local err to ignore resource not found exceptions
970     UErrorCode err = U_ZERO_ERROR;
971 
972     // Load append item formats.
973     AppendItemFormatsSink appendItemFormatsSink(*this);
974     path.clear()
975         .append(DT_DateTimeCalendarTag, errorCode)
976         .append('/', errorCode)
977         .append(calendarTypeToUse, errorCode)
978         .append('/', errorCode)
979         .append(DT_DateTimeAppendItemsTag, errorCode); // i.e., calendar/xxx/appendItems
980     if (U_FAILURE(errorCode)) { return; }
981     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), appendItemFormatsSink, err);
982     appendItemFormatsSink.fillInMissing();
983 
984     // Load CLDR item names.
985     err = U_ZERO_ERROR;
986     AppendItemNamesSink appendItemNamesSink(*this);
987     ures_getAllItemsWithFallback(rb.getAlias(), DT_DateTimeFieldsTag, appendItemNamesSink, err);
988     appendItemNamesSink.fillInMissing();
989 
990     // Load the available formats from CLDR.
991     err = U_ZERO_ERROR;
992     initHashtable(errorCode);
993     if (U_FAILURE(errorCode)) { return; }
994     AvailableFormatsSink availableFormatsSink(*this);
995     path.clear()
996         .append(DT_DateTimeCalendarTag, errorCode)
997         .append('/', errorCode)
998         .append(calendarTypeToUse, errorCode)
999         .append('/', errorCode)
1000         .append(DT_DateTimeAvailableFormatsTag, errorCode); // i.e., calendar/xxx/availableFormats
1001     if (U_FAILURE(errorCode)) { return; }
1002     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), availableFormatsSink, err);
1003 }
1004 
1005 void
initHashtable(UErrorCode & err)1006 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
1007     if (U_FAILURE(err)) { return; }
1008     if (fAvailableFormatKeyHash!=nullptr) {
1009         return;
1010     }
1011     LocalPointer<Hashtable> hash(new Hashtable(FALSE, err), err);
1012     if (U_SUCCESS(err)) {
1013         fAvailableFormatKeyHash = hash.orphan();
1014     }
1015 }
1016 
1017 void
setAppendItemFormat(UDateTimePatternField field,const UnicodeString & value)1018 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
1019     appendItemFormats[field] = value;
1020     // NUL-terminate for the C API.
1021     appendItemFormats[field].getTerminatedBuffer();
1022 }
1023 
1024 const UnicodeString&
getAppendItemFormat(UDateTimePatternField field) const1025 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
1026     return appendItemFormats[field];
1027 }
1028 
1029 void
setAppendItemName(UDateTimePatternField field,const UnicodeString & value)1030 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
1031     setFieldDisplayName(field, UDATPG_WIDTH_APPENDITEM, value);
1032 }
1033 
1034 const UnicodeString&
getAppendItemName(UDateTimePatternField field) const1035 DateTimePatternGenerator::getAppendItemName(UDateTimePatternField field) const {
1036     return fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1037 }
1038 
1039 void
setFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width,const UnicodeString & value)1040 DateTimePatternGenerator::setFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width, const UnicodeString& value) {
1041     fieldDisplayNames[field][width] = value;
1042     // NUL-terminate for the C API.
1043     fieldDisplayNames[field][width].getTerminatedBuffer();
1044 }
1045 
1046 UnicodeString
getFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width) const1047 DateTimePatternGenerator::getFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) const {
1048     return fieldDisplayNames[field][width];
1049 }
1050 
1051 UnicodeString&
getMutableFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width)1052 DateTimePatternGenerator::getMutableFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) {
1053     return fieldDisplayNames[field][width];
1054 }
1055 
1056 void
getAppendName(UDateTimePatternField field,UnicodeString & value)1057 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
1058     value = SINGLE_QUOTE;
1059     value += fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1060     value += SINGLE_QUOTE;
1061 }
1062 
1063 UnicodeString
getBestPattern(const UnicodeString & patternForm,UErrorCode & status)1064 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
1065     return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
1066 }
1067 
1068 UnicodeString
getBestPattern(const UnicodeString & patternForm,UDateTimePatternMatchOptions options,UErrorCode & status)1069 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
1070     if (U_FAILURE(status)) {
1071         return UnicodeString();
1072     }
1073     if (U_FAILURE(internalErrorCode)) {
1074         status = internalErrorCode;
1075         return UnicodeString();
1076     }
1077     const UnicodeString *bestPattern = nullptr;
1078     UnicodeString dtFormat;
1079     UnicodeString resultPattern;
1080     int32_t flags = kDTPGNoFlags;
1081 
1082     int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
1083     int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
1084 
1085     // Replace hour metacharacters 'j', 'C' and 'J', set flags as necessary
1086     UnicodeString patternFormMapped = mapSkeletonMetacharacters(patternForm, &flags, status);
1087     if (U_FAILURE(status)) {
1088         return UnicodeString();
1089     }
1090 
1091     resultPattern.remove();
1092     dtMatcher->set(patternFormMapped, fp);
1093     const PtnSkeleton* specifiedSkeleton = nullptr;
1094     bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, status, &specifiedSkeleton);
1095     if (U_FAILURE(status)) {
1096         return UnicodeString();
1097     }
1098 
1099     if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
1100         resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, flags, options);
1101 
1102         return resultPattern;
1103     }
1104     int32_t neededFields = dtMatcher->getFieldMask();
1105     UnicodeString datePattern=getBestAppending(neededFields & dateMask, flags, status, options);
1106     UnicodeString timePattern=getBestAppending(neededFields & timeMask, flags, status, options);
1107     if (U_FAILURE(status)) {
1108         return UnicodeString();
1109     }
1110     if (datePattern.length()==0) {
1111         if (timePattern.length()==0) {
1112             resultPattern.remove();
1113         }
1114         else {
1115             return timePattern;
1116         }
1117     }
1118     if (timePattern.length()==0) {
1119         return datePattern;
1120     }
1121     resultPattern.remove();
1122     status = U_ZERO_ERROR;
1123     dtFormat=getDateTimeFormat();
1124     SimpleFormatter(dtFormat, 2, 2, status).format(timePattern, datePattern, resultPattern, status);
1125     return resultPattern;
1126 }
1127 
1128 /*
1129  * Map a skeleton that may have metacharacters jJC to one without, by replacing
1130  * the metacharacters with locale-appropriate fields of h/H/k/K and of a/b/B
1131  * (depends on fDefaultHourFormatChar and fAllowedHourFormats being set, which in
1132  * turn depends on initData having been run). This method also updates the flags
1133  * as necessary. Returns the updated skeleton.
1134  */
1135 UnicodeString
mapSkeletonMetacharacters(const UnicodeString & patternForm,int32_t * flags,UErrorCode & status)1136 DateTimePatternGenerator::mapSkeletonMetacharacters(const UnicodeString& patternForm, int32_t* flags, UErrorCode& status) {
1137     UnicodeString patternFormMapped;
1138     patternFormMapped.remove();
1139     UBool inQuoted = FALSE;
1140     int32_t patPos, patLen = patternForm.length();
1141     for (patPos = 0; patPos < patLen; patPos++) {
1142         UChar patChr = patternForm.charAt(patPos);
1143         if (patChr == SINGLE_QUOTE) {
1144             inQuoted = !inQuoted;
1145         } else if (!inQuoted) {
1146             // Handle special mappings for 'j' and 'C' in which fields lengths
1147             // 1,3,5 => hour field length 1
1148             // 2,4,6 => hour field length 2
1149             // 1,2 => abbreviated dayPeriod (field length 1..3)
1150             // 3,4 => long dayPeriod (field length 4)
1151             // 5,6 => narrow dayPeriod (field length 5)
1152             if (patChr == LOW_J || patChr == CAP_C) {
1153                 int32_t extraLen = 0; // 1 less than total field length
1154                 while (patPos+1 < patLen && patternForm.charAt(patPos+1)==patChr) {
1155                     extraLen++;
1156                     patPos++;
1157                 }
1158                 int32_t hourLen = 1 + (extraLen & 1);
1159                 int32_t dayPeriodLen = (extraLen < 2)? 1: 3 + (extraLen >> 1);
1160                 UChar hourChar = LOW_H;
1161                 UChar dayPeriodChar = LOW_A;
1162                 if (patChr == LOW_J) {
1163                     hourChar = fDefaultHourFormatChar;
1164                 } else {
1165                     AllowedHourFormat bestAllowed;
1166                     if (fAllowedHourFormats[0] != ALLOWED_HOUR_FORMAT_UNKNOWN) {
1167                         bestAllowed = (AllowedHourFormat)fAllowedHourFormats[0];
1168                     } else {
1169                         status = U_INVALID_FORMAT_ERROR;
1170                         return UnicodeString();
1171                     }
1172                     if (bestAllowed == ALLOWED_HOUR_FORMAT_H || bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_Hb) {
1173                         hourChar = CAP_H;
1174                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_K || bestAllowed == ALLOWED_HOUR_FORMAT_KB || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) {
1175                         hourChar = CAP_K;
1176                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_k) {
1177                         hourChar = LOW_K;
1178                     }
1179                     // in #13183 just add b/B to skeleton, no longer need to set special flags
1180                     if (bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_hB || bestAllowed == ALLOWED_HOUR_FORMAT_KB) {
1181                         dayPeriodChar = CAP_B;
1182                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_Hb || bestAllowed == ALLOWED_HOUR_FORMAT_hb || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) {
1183                         dayPeriodChar = LOW_B;
1184                     }
1185                 }
1186                 if (hourChar==CAP_H || hourChar==LOW_K) {
1187                     dayPeriodLen = 0;
1188                 }
1189                 while (dayPeriodLen-- > 0) {
1190                     patternFormMapped.append(dayPeriodChar);
1191                 }
1192                 while (hourLen-- > 0) {
1193                     patternFormMapped.append(hourChar);
1194                 }
1195             } else if (patChr == CAP_J) {
1196                 // Get pattern for skeleton with H, then replace H or k
1197                 // with fDefaultHourFormatChar (if different)
1198                 patternFormMapped.append(CAP_H);
1199                 *flags |= kDTPGSkeletonUsesCapJ;
1200             } else {
1201                 patternFormMapped.append(patChr);
1202             }
1203         }
1204     }
1205     return patternFormMapped;
1206 }
1207 
1208 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UErrorCode & status)1209 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1210                                             const UnicodeString& skeleton,
1211                                             UErrorCode& status) {
1212     return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
1213 }
1214 
1215 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UDateTimePatternMatchOptions options,UErrorCode & status)1216 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1217                                             const UnicodeString& skeleton,
1218                                             UDateTimePatternMatchOptions options,
1219                                             UErrorCode& status) {
1220     if (U_FAILURE(status)) {
1221         return UnicodeString();
1222     }
1223     if (U_FAILURE(internalErrorCode)) {
1224         status = internalErrorCode;
1225         return UnicodeString();
1226     }
1227     dtMatcher->set(skeleton, fp);
1228     UnicodeString result = adjustFieldTypes(pattern, nullptr, kDTPGNoFlags, options);
1229     return result;
1230 }
1231 
1232 void
setDecimal(const UnicodeString & newDecimal)1233 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
1234     this->decimal = newDecimal;
1235     // NUL-terminate for the C API.
1236     this->decimal.getTerminatedBuffer();
1237 }
1238 
1239 const UnicodeString&
getDecimal() const1240 DateTimePatternGenerator::getDecimal() const {
1241     return decimal;
1242 }
1243 
1244 void
addCanonicalItems(UErrorCode & status)1245 DateTimePatternGenerator::addCanonicalItems(UErrorCode& status) {
1246     if (U_FAILURE(status)) { return; }
1247     UnicodeString  conflictingPattern;
1248 
1249     for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
1250         if (Canonical_Items[i] > 0) {
1251             addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
1252         }
1253         if (U_FAILURE(status)) { return; }
1254     }
1255 }
1256 
1257 void
setDateTimeFormat(const UnicodeString & dtFormat)1258 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
1259     dateTimeFormat = dtFormat;
1260     // NUL-terminate for the C API.
1261     dateTimeFormat.getTerminatedBuffer();
1262 }
1263 
1264 const UnicodeString&
getDateTimeFormat() const1265 DateTimePatternGenerator::getDateTimeFormat() const {
1266     return dateTimeFormat;
1267 }
1268 
1269 void
setDateTimeFromCalendar(const Locale & locale,UErrorCode & status)1270 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
1271     if (U_FAILURE(status)) { return; }
1272 
1273     const UChar *resStr;
1274     int32_t resStrLen = 0;
1275 
1276     LocalPointer<Calendar> fCalendar(Calendar::createInstance(locale, status), status);
1277     if (U_FAILURE(status)) { return; }
1278 
1279     LocalUResourceBundlePointer calData(ures_open(nullptr, locale.getBaseName(), &status));
1280     if (U_FAILURE(status)) { return; }
1281     ures_getByKey(calData.getAlias(), DT_DateTimeCalendarTag, calData.getAlias(), &status);
1282     if (U_FAILURE(status)) { return; }
1283 
1284     LocalUResourceBundlePointer dateTimePatterns;
1285     if (fCalendar->getType() != nullptr && *fCalendar->getType() != '\0'
1286             && uprv_strcmp(fCalendar->getType(), DT_DateTimeGregorianTag) != 0) {
1287         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), fCalendar->getType(),
1288                                                                 nullptr, &status));
1289         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1290                                   dateTimePatterns.getAlias(), &status);
1291     }
1292 
1293     if (dateTimePatterns.isNull() || status == U_MISSING_RESOURCE_ERROR) {
1294         status = U_ZERO_ERROR;
1295         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), DT_DateTimeGregorianTag,
1296                                                                 dateTimePatterns.orphan(), &status));
1297         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1298                                   dateTimePatterns.getAlias(), &status);
1299     }
1300     if (U_FAILURE(status)) { return; }
1301 
1302     if (ures_getSize(dateTimePatterns.getAlias()) <= DateFormat::kDateTime)
1303     {
1304         status = U_INVALID_FORMAT_ERROR;
1305         return;
1306     }
1307     resStr = ures_getStringByIndex(dateTimePatterns.getAlias(), (int32_t)DateFormat::kDateTime, &resStrLen, &status);
1308     setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
1309 }
1310 
1311 void
setDecimalSymbols(const Locale & locale,UErrorCode & status)1312 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
1313     DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
1314     if(U_SUCCESS(status)) {
1315         decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
1316         // NUL-terminate for the C API.
1317         decimal.getTerminatedBuffer();
1318     }
1319 }
1320 
1321 UDateTimePatternConflict
addPattern(const UnicodeString & pattern,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1322 DateTimePatternGenerator::addPattern(
1323     const UnicodeString& pattern,
1324     UBool override,
1325     UnicodeString &conflictingPattern,
1326     UErrorCode& status)
1327 {
1328     if (U_FAILURE(internalErrorCode)) {
1329         status = internalErrorCode;
1330         return UDATPG_NO_CONFLICT;
1331     }
1332 
1333     return addPatternWithSkeleton(pattern, nullptr, override, conflictingPattern, status);
1334 }
1335 
1336 // For DateTimePatternGenerator::addPatternWithSkeleton -
1337 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
1338 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
1339 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
1340 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
1341 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
1342 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
1343 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
1344 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
1345 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
1346 UDateTimePatternConflict
addPatternWithSkeleton(const UnicodeString & pattern,const UnicodeString * skeletonToUse,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1347 DateTimePatternGenerator::addPatternWithSkeleton(
1348     const UnicodeString& pattern,
1349     const UnicodeString* skeletonToUse,
1350     UBool override,
1351     UnicodeString& conflictingPattern,
1352     UErrorCode& status)
1353 {
1354     if (U_FAILURE(internalErrorCode)) {
1355         status = internalErrorCode;
1356         return UDATPG_NO_CONFLICT;
1357     }
1358 
1359     UnicodeString basePattern;
1360     PtnSkeleton   skeleton;
1361     UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
1362 
1363     DateTimeMatcher matcher;
1364     if ( skeletonToUse == nullptr ) {
1365         matcher.set(pattern, fp, skeleton);
1366         matcher.getBasePattern(basePattern);
1367     } else {
1368         matcher.set(*skeletonToUse, fp, skeleton); // no longer trims skeleton fields to max len 3, per #7930
1369         matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
1370     }
1371     // We only care about base conflicts - and replacing the pattern associated with a base - if:
1372     // 1. the conflicting previous base pattern did *not* have an explicit skeleton; in that case the previous
1373     // base + pattern combination was derived from either (a) a canonical item, (b) a standard format, or
1374     // (c) a pattern specified programmatically with a previous call to addPattern (which would only happen
1375     // if we are getting here from a subsequent call to addPattern).
1376     // 2. a skeleton is specified for the current pattern, but override=false; in that case we are checking
1377     // availableFormats items from root, which should not override any previous entry with the same base.
1378     UBool entryHadSpecifiedSkeleton;
1379     const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
1380     if (duplicatePattern != nullptr && (!entryHadSpecifiedSkeleton || (skeletonToUse != nullptr && !override))) {
1381         conflictingStatus = UDATPG_BASE_CONFLICT;
1382         conflictingPattern = *duplicatePattern;
1383         if (!override) {
1384             return conflictingStatus;
1385         }
1386     }
1387     // The only time we get here with override=true and skeletonToUse!=null is when adding availableFormats
1388     // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with
1389     // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for
1390     // the previously-specified conflicting item.
1391     const PtnSkeleton* entrySpecifiedSkeleton = nullptr;
1392     duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
1393     if (duplicatePattern != nullptr ) {
1394         conflictingStatus = UDATPG_CONFLICT;
1395         conflictingPattern = *duplicatePattern;
1396         if (!override || (skeletonToUse != nullptr && entrySpecifiedSkeleton != nullptr)) {
1397             return conflictingStatus;
1398         }
1399     }
1400     patternMap->add(basePattern, skeleton, pattern, skeletonToUse != nullptr, status);
1401     if(U_FAILURE(status)) {
1402         return conflictingStatus;
1403     }
1404 
1405     return UDATPG_NO_CONFLICT;
1406 }
1407 
1408 
1409 UDateTimePatternField
getAppendFormatNumber(const char * field) const1410 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
1411     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1412         if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
1413             return (UDateTimePatternField)i;
1414         }
1415     }
1416     return UDATPG_FIELD_COUNT;
1417 }
1418 
1419 UDateTimePatternField
getFieldAndWidthIndices(const char * key,UDateTimePGDisplayWidth * widthP) const1420 DateTimePatternGenerator::getFieldAndWidthIndices(const char* key, UDateTimePGDisplayWidth* widthP) const {
1421     char cldrFieldKey[UDATPG_FIELD_KEY_MAX + 1];
1422     uprv_strncpy(cldrFieldKey, key, UDATPG_FIELD_KEY_MAX);
1423     cldrFieldKey[UDATPG_FIELD_KEY_MAX]=0; // ensure termination
1424     *widthP = UDATPG_WIDE;
1425     char* hyphenPtr = uprv_strchr(cldrFieldKey, '-');
1426     if (hyphenPtr) {
1427         for (int32_t i=UDATPG_WIDTH_COUNT-1; i>0; --i) {
1428             if (uprv_strcmp(CLDR_FIELD_WIDTH[i], hyphenPtr)==0) {
1429                 *widthP=(UDateTimePGDisplayWidth)i;
1430                 break;
1431             }
1432         }
1433         *hyphenPtr = 0; // now delete width portion of key
1434     }
1435     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1436         if (uprv_strcmp(CLDR_FIELD_NAME[i],cldrFieldKey)==0) {
1437             return (UDateTimePatternField)i;
1438         }
1439     }
1440     return UDATPG_FIELD_COUNT;
1441 }
1442 
1443 const UnicodeString*
getBestRaw(DateTimeMatcher & source,int32_t includeMask,DistanceInfo * missingFields,UErrorCode & status,const PtnSkeleton ** specifiedSkeletonPtr)1444 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
1445                                      int32_t includeMask,
1446                                      DistanceInfo* missingFields,
1447                                      UErrorCode &status,
1448                                      const PtnSkeleton** specifiedSkeletonPtr) {
1449     int32_t bestDistance = 0x7fffffff;
1450     DistanceInfo tempInfo;
1451     const UnicodeString *bestPattern=nullptr;
1452     const PtnSkeleton* specifiedSkeleton=nullptr;
1453 
1454     PatternMapIterator it(status);
1455     if (U_FAILURE(status)) { return nullptr; }
1456 
1457     for (it.set(*patternMap); it.hasNext(); ) {
1458         DateTimeMatcher trial = it.next();
1459         if (trial.equals(skipMatcher)) {
1460             continue;
1461         }
1462         int32_t distance=source.getDistance(trial, includeMask, tempInfo);
1463         if (distance<bestDistance) {
1464             bestDistance=distance;
1465             bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
1466             missingFields->setTo(tempInfo);
1467             if (distance==0) {
1468                 break;
1469             }
1470         }
1471     }
1472 
1473     // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
1474     // then return it too. This generally happens when the caller needs to pass that skeleton
1475     // through to adjustFieldTypes so the latter can do a better job.
1476     if (bestPattern && specifiedSkeletonPtr) {
1477         *specifiedSkeletonPtr = specifiedSkeleton;
1478     }
1479     return bestPattern;
1480 }
1481 
1482 UnicodeString
adjustFieldTypes(const UnicodeString & pattern,const PtnSkeleton * specifiedSkeleton,int32_t flags,UDateTimePatternMatchOptions options)1483 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
1484                                            const PtnSkeleton* specifiedSkeleton,
1485                                            int32_t flags,
1486                                            UDateTimePatternMatchOptions options) {
1487     UnicodeString newPattern;
1488     fp->set(pattern);
1489     for (int32_t i=0; i < fp->itemNumber; i++) {
1490         UnicodeString field = fp->items[i];
1491         if ( fp->isQuoteLiteral(field) ) {
1492 
1493             UnicodeString quoteLiteral;
1494             fp->getQuoteLiteral(quoteLiteral, &i);
1495             newPattern += quoteLiteral;
1496         }
1497         else {
1498             if (fp->isPatternSeparator(field)) {
1499                 newPattern+=field;
1500                 continue;
1501             }
1502             int32_t canonicalIndex = fp->getCanonicalIndex(field);
1503             if (canonicalIndex < 0) {
1504                 newPattern+=field;
1505                 continue;  // don't adjust
1506             }
1507             const dtTypeElem *row = &dtTypes[canonicalIndex];
1508             int32_t typeValue = row->field;
1509 
1510             // handle day periods - with #13183, no longer need special handling here, integrated with normal types
1511 
1512             if ((flags & kDTPGFixFractionalSeconds) != 0 && typeValue == UDATPG_SECOND_FIELD) {
1513                 field += decimal;
1514                 dtMatcher->skeleton.original.appendFieldTo(UDATPG_FRACTIONAL_SECOND_FIELD, field);
1515             } else if (dtMatcher->skeleton.type[typeValue]!=0) {
1516                     // Here:
1517                     // - "reqField" is the field from the originally requested skeleton, with length
1518                     // "reqFieldLen".
1519                     // - "field" is the field from the found pattern.
1520                     //
1521                     // The adjusted field should consist of characters from the originally requested
1522                     // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1523                     // UDATPG_WEEKDAY_FIELD or UDATPG_YEAR_FIELD, in which case it should consist
1524                     // of characters from the  found pattern.
1525                     //
1526                     // The length of the adjusted field (adjFieldLen) should match that in the originally
1527                     // requested skeleton, except that in the following cases the length of the adjusted field
1528                     // should match that in the found pattern (i.e. the length of this pattern field should
1529                     // not be adjusted):
1530                     // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1531                     //    not set (ticket #7180). Note, we may want to implement a similar change for other
1532                     //    numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1533                     //    field length, but options bits can be used to override this.
1534                     // 2. There is a specified skeleton for the found pattern and one of the following is true:
1535                     //    a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1536                     //    b) The pattern field is numeric and the skeleton field is not, or vice versa.
1537 
1538                     UChar reqFieldChar = dtMatcher->skeleton.original.getFieldChar(typeValue);
1539                     int32_t reqFieldLen = dtMatcher->skeleton.original.getFieldLength(typeValue);
1540                     if (reqFieldChar == CAP_E && reqFieldLen < 3)
1541                         reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1542                     int32_t adjFieldLen = reqFieldLen;
1543                     if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1544                          (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1545                          (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1546                          adjFieldLen = field.length();
1547                     } else if (specifiedSkeleton) {
1548                         int32_t skelFieldLen = specifiedSkeleton->original.getFieldLength(typeValue);
1549                         UBool patFieldIsNumeric = (row->type > 0);
1550                         UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1551                         if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1552                             // don't adjust the field length in the found pattern
1553                             adjFieldLen = field.length();
1554                         }
1555                     }
1556                     UChar c = (typeValue!= UDATPG_HOUR_FIELD
1557                             && typeValue!= UDATPG_MONTH_FIELD
1558                             && typeValue!= UDATPG_WEEKDAY_FIELD
1559                             && (typeValue!= UDATPG_YEAR_FIELD || reqFieldChar==CAP_Y))
1560                             ? reqFieldChar
1561                             : field.charAt(0);
1562                     if (typeValue == UDATPG_HOUR_FIELD && (flags & kDTPGSkeletonUsesCapJ) != 0) {
1563                         c = fDefaultHourFormatChar;
1564                     }
1565                     field.remove();
1566                     for (int32_t j=adjFieldLen; j>0; --j) {
1567                         field += c;
1568                     }
1569             }
1570             newPattern+=field;
1571         }
1572     }
1573     return newPattern;
1574 }
1575 
1576 UnicodeString
getBestAppending(int32_t missingFields,int32_t flags,UErrorCode & status,UDateTimePatternMatchOptions options)1577 DateTimePatternGenerator::getBestAppending(int32_t missingFields, int32_t flags, UErrorCode &status, UDateTimePatternMatchOptions options) {
1578     if (U_FAILURE(status)) {
1579         return UnicodeString();
1580     }
1581     UnicodeString  resultPattern, tempPattern;
1582     const UnicodeString* tempPatternPtr;
1583     int32_t lastMissingFieldMask=0;
1584     if (missingFields!=0) {
1585         resultPattern=UnicodeString();
1586         const PtnSkeleton* specifiedSkeleton=nullptr;
1587         tempPatternPtr = getBestRaw(*dtMatcher, missingFields, distanceInfo, status, &specifiedSkeleton);
1588         if (U_FAILURE(status)) {
1589             return UnicodeString();
1590         }
1591         tempPattern = *tempPatternPtr;
1592         resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1593         if ( distanceInfo->missingFieldMask==0 ) {
1594             return resultPattern;
1595         }
1596         while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1597             if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1598                 break;  // cannot find the proper missing field
1599             }
1600             if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1601                 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1602                 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, flags | kDTPGFixFractionalSeconds, options);
1603                 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1604                 continue;
1605             }
1606             int32_t startingMask = distanceInfo->missingFieldMask;
1607             tempPatternPtr = getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, status, &specifiedSkeleton);
1608             if (U_FAILURE(status)) {
1609                 return UnicodeString();
1610             }
1611             tempPattern = *tempPatternPtr;
1612             tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1613             int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1614             int32_t topField=getTopBitNumber(foundMask);
1615 
1616             if (appendItemFormats[topField].length() != 0) {
1617                 UnicodeString appendName;
1618                 getAppendName((UDateTimePatternField)topField, appendName);
1619                 const UnicodeString *values[3] = {
1620                     &resultPattern,
1621                     &tempPattern,
1622                     &appendName
1623                 };
1624                 SimpleFormatter(appendItemFormats[topField], 2, 3, status).
1625                     formatAndReplace(values, 3, resultPattern, nullptr, 0, status);
1626             }
1627             lastMissingFieldMask = distanceInfo->missingFieldMask;
1628         }
1629     }
1630     return resultPattern;
1631 }
1632 
1633 int32_t
getTopBitNumber(int32_t foundMask) const1634 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) const {
1635     if ( foundMask==0 ) {
1636         return 0;
1637     }
1638     int32_t i=0;
1639     while (foundMask!=0) {
1640         foundMask >>=1;
1641         ++i;
1642     }
1643     if (i-1 >UDATPG_ZONE_FIELD) {
1644         return UDATPG_ZONE_FIELD;
1645     }
1646     else
1647         return i-1;
1648 }
1649 
1650 void
setAvailableFormat(const UnicodeString & key,UErrorCode & err)1651 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1652 {
1653     fAvailableFormatKeyHash->puti(key, 1, err);
1654 }
1655 
1656 UBool
isAvailableFormatSet(const UnicodeString & key) const1657 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1658     return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1659 }
1660 
1661 void
copyHashtable(Hashtable * other,UErrorCode & status)1662 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1663     if (other == nullptr || U_FAILURE(status)) {
1664         return;
1665     }
1666     if (fAvailableFormatKeyHash != nullptr) {
1667         delete fAvailableFormatKeyHash;
1668         fAvailableFormatKeyHash = nullptr;
1669     }
1670     initHashtable(status);
1671     if(U_FAILURE(status)){
1672         return;
1673     }
1674     int32_t pos = UHASH_FIRST;
1675     const UHashElement* elem = nullptr;
1676     // walk through the hash table and create a deep clone
1677     while((elem = other->nextElement(pos))!= nullptr){
1678         const UHashTok otherKeyTok = elem->key;
1679         UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1680         fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1681         if(U_FAILURE(status)){
1682             return;
1683         }
1684     }
1685 }
1686 
1687 StringEnumeration*
getSkeletons(UErrorCode & status) const1688 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1689     if (U_FAILURE(status)) {
1690         return nullptr;
1691     }
1692     if (U_FAILURE(internalErrorCode)) {
1693         status = internalErrorCode;
1694         return nullptr;
1695     }
1696     LocalPointer<StringEnumeration> skeletonEnumerator(
1697         new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status), status);
1698 
1699     return U_SUCCESS(status) ? skeletonEnumerator.orphan() : nullptr;
1700 }
1701 
1702 const UnicodeString&
getPatternForSkeleton(const UnicodeString & skeleton) const1703 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1704     PtnElem *curElem;
1705 
1706     if (skeleton.length() ==0) {
1707         return emptyString;
1708     }
1709     curElem = patternMap->getHeader(skeleton.charAt(0));
1710     while ( curElem != nullptr ) {
1711         if ( curElem->skeleton->getSkeleton()==skeleton ) {
1712             return curElem->pattern;
1713         }
1714         curElem = curElem->next.getAlias();
1715     }
1716     return emptyString;
1717 }
1718 
1719 StringEnumeration*
getBaseSkeletons(UErrorCode & status) const1720 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1721     if (U_FAILURE(status)) {
1722         return nullptr;
1723     }
1724     if (U_FAILURE(internalErrorCode)) {
1725         status = internalErrorCode;
1726         return nullptr;
1727     }
1728     LocalPointer<StringEnumeration> baseSkeletonEnumerator(
1729         new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status), status);
1730 
1731     return U_SUCCESS(status) ? baseSkeletonEnumerator.orphan() : nullptr;
1732 }
1733 
1734 StringEnumeration*
getRedundants(UErrorCode & status)1735 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1736     if (U_FAILURE(status)) { return nullptr; }
1737     if (U_FAILURE(internalErrorCode)) {
1738         status = internalErrorCode;
1739         return nullptr;
1740     }
1741     LocalPointer<StringEnumeration> output(new DTRedundantEnumeration(), status);
1742     if (U_FAILURE(status)) { return nullptr; }
1743     const UnicodeString *pattern;
1744     PatternMapIterator it(status);
1745     if (U_FAILURE(status)) { return nullptr; }
1746 
1747     for (it.set(*patternMap); it.hasNext(); ) {
1748         DateTimeMatcher current = it.next();
1749         pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1750         if ( isCanonicalItem(*pattern) ) {
1751             continue;
1752         }
1753         if ( skipMatcher == nullptr ) {
1754             skipMatcher = new DateTimeMatcher(current);
1755             if (skipMatcher == nullptr) {
1756                 status = U_MEMORY_ALLOCATION_ERROR;
1757                 return nullptr;
1758             }
1759         }
1760         else {
1761             *skipMatcher = current;
1762         }
1763         UnicodeString trial = getBestPattern(current.getPattern(), status);
1764         if (U_FAILURE(status)) { return nullptr; }
1765         if (trial == *pattern) {
1766             ((DTRedundantEnumeration *)output.getAlias())->add(*pattern, status);
1767             if (U_FAILURE(status)) { return nullptr; }
1768         }
1769         if (current.equals(skipMatcher)) {
1770             continue;
1771         }
1772     }
1773     return output.orphan();
1774 }
1775 
1776 UBool
isCanonicalItem(const UnicodeString & item) const1777 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1778     if ( item.length() != 1 ) {
1779         return FALSE;
1780     }
1781     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1782         if (item.charAt(0)==Canonical_Items[i]) {
1783             return TRUE;
1784         }
1785     }
1786     return FALSE;
1787 }
1788 
1789 
1790 DateTimePatternGenerator*
clone() const1791 DateTimePatternGenerator::clone() const {
1792     return new DateTimePatternGenerator(*this);
1793 }
1794 
PatternMap()1795 PatternMap::PatternMap() {
1796    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1797        boot[i] = nullptr;
1798    }
1799    isDupAllowed = TRUE;
1800 }
1801 
1802 void
copyFrom(const PatternMap & other,UErrorCode & status)1803 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1804     if (U_FAILURE(status)) {
1805         return;
1806     }
1807     this->isDupAllowed = other.isDupAllowed;
1808     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
1809         PtnElem *curElem, *otherElem, *prevElem=nullptr;
1810         otherElem = other.boot[bootIndex];
1811         while (otherElem != nullptr) {
1812             LocalPointer<PtnElem> newElem(new PtnElem(otherElem->basePattern, otherElem->pattern), status);
1813             if (U_FAILURE(status)) {
1814                 return; // out of memory
1815             }
1816             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(*(otherElem->skeleton)), status);
1817             if (U_FAILURE(status)) {
1818                 return; // out of memory
1819             }
1820             newElem->skeletonWasSpecified = otherElem->skeletonWasSpecified;
1821 
1822             // Release ownership from the LocalPointer of the PtnElem object.
1823             // The PtnElem will now be owned by either the boot (for the first entry in the linked-list)
1824             // or owned by the previous PtnElem object in the linked-list.
1825             curElem = newElem.orphan();
1826 
1827             if (this->boot[bootIndex] == nullptr) {
1828                 this->boot[bootIndex] = curElem;
1829             } else {
1830                 if (prevElem != nullptr) {
1831                     prevElem->next.adoptInstead(curElem);
1832                 } else {
1833                     UPRV_UNREACHABLE;
1834                 }
1835             }
1836             prevElem = curElem;
1837             otherElem = otherElem->next.getAlias();
1838         }
1839 
1840     }
1841 }
1842 
1843 PtnElem*
getHeader(UChar baseChar) const1844 PatternMap::getHeader(UChar baseChar) const {
1845     PtnElem* curElem;
1846 
1847     if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1848          curElem = boot[baseChar-CAP_A];
1849     }
1850     else {
1851         if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1852             curElem = boot[26+baseChar-LOW_A];
1853         }
1854         else {
1855             return nullptr;
1856         }
1857     }
1858     return curElem;
1859 }
1860 
~PatternMap()1861 PatternMap::~PatternMap() {
1862    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1863        if (boot[i] != nullptr ) {
1864            delete boot[i];
1865            boot[i] = nullptr;
1866        }
1867    }
1868 }  // PatternMap destructor
1869 
1870 void
add(const UnicodeString & basePattern,const PtnSkeleton & skeleton,const UnicodeString & value,UBool skeletonWasSpecified,UErrorCode & status)1871 PatternMap::add(const UnicodeString& basePattern,
1872                 const PtnSkeleton& skeleton,
1873                 const UnicodeString& value,// mapped pattern value
1874                 UBool skeletonWasSpecified,
1875                 UErrorCode &status) {
1876     UChar baseChar = basePattern.charAt(0);
1877     PtnElem *curElem, *baseElem;
1878     status = U_ZERO_ERROR;
1879 
1880     // the baseChar must be A-Z or a-z
1881     if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1882         baseElem = boot[baseChar-CAP_A];
1883     }
1884     else {
1885         if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1886             baseElem = boot[26+baseChar-LOW_A];
1887          }
1888          else {
1889              status = U_ILLEGAL_CHARACTER;
1890              return;
1891          }
1892     }
1893 
1894     if (baseElem == nullptr) {
1895         LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1896         if (U_FAILURE(status)) {
1897             return; // out of memory
1898         }
1899         newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1900         if (U_FAILURE(status)) {
1901             return; // out of memory
1902         }
1903         newElem->skeletonWasSpecified = skeletonWasSpecified;
1904         if (baseChar >= LOW_A) {
1905             boot[26 + (baseChar - LOW_A)] = newElem.orphan(); // the boot array now owns the PtnElem.
1906         }
1907         else {
1908             boot[baseChar - CAP_A] = newElem.orphan(); // the boot array now owns the PtnElem.
1909         }
1910     }
1911     if ( baseElem != nullptr ) {
1912         curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1913 
1914         if (curElem == nullptr) {
1915             // add new element to the list.
1916             curElem = baseElem;
1917             while( curElem -> next != nullptr )
1918             {
1919                 curElem = curElem->next.getAlias();
1920             }
1921 
1922             LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1923             if (U_FAILURE(status)) {
1924                 return; // out of memory
1925             }
1926             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1927             if (U_FAILURE(status)) {
1928                 return; // out of memory
1929             }
1930             newElem->skeletonWasSpecified = skeletonWasSpecified;
1931             curElem->next.adoptInstead(newElem.orphan());
1932             curElem = curElem->next.getAlias();
1933         }
1934         else {
1935             // Pattern exists in the list already.
1936             if ( !isDupAllowed ) {
1937                 return;
1938             }
1939             // Overwrite the value.
1940             curElem->pattern = value;
1941             // It was a bug that we were not doing the following previously,
1942             // though that bug hid other problems by making things partly work.
1943             curElem->skeletonWasSpecified = skeletonWasSpecified;
1944         }
1945     }
1946 }  // PatternMap::add
1947 
1948 // Find the pattern from the given basePattern string.
1949 const UnicodeString *
getPatternFromBasePattern(const UnicodeString & basePattern,UBool & skeletonWasSpecified) const1950 PatternMap::getPatternFromBasePattern(const UnicodeString& basePattern, UBool& skeletonWasSpecified) const { // key to search for
1951    PtnElem *curElem;
1952 
1953    if ((curElem=getHeader(basePattern.charAt(0)))==nullptr) {
1954        return nullptr;  // no match
1955    }
1956 
1957    do  {
1958        if ( basePattern.compare(curElem->basePattern)==0 ) {
1959           skeletonWasSpecified = curElem->skeletonWasSpecified;
1960           return &(curElem->pattern);
1961        }
1962        curElem = curElem->next.getAlias();
1963    } while (curElem != nullptr);
1964 
1965    return nullptr;
1966 }  // PatternMap::getFromBasePattern
1967 
1968 
1969 // Find the pattern from the given skeleton.
1970 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
1971 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
1972 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
1973 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
1974 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
1975 const UnicodeString *
getPatternFromSkeleton(const PtnSkeleton & skeleton,const PtnSkeleton ** specifiedSkeletonPtr) const1976 PatternMap::getPatternFromSkeleton(const PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) const { // key to search for
1977    PtnElem *curElem;
1978 
1979    if (specifiedSkeletonPtr) {
1980        *specifiedSkeletonPtr = nullptr;
1981    }
1982 
1983    // find boot entry
1984    UChar baseChar = skeleton.getFirstChar();
1985    if ((curElem=getHeader(baseChar))==nullptr) {
1986        return nullptr;  // no match
1987    }
1988 
1989    do  {
1990        UBool equal;
1991        if (specifiedSkeletonPtr != nullptr) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
1992            equal = curElem->skeleton->original == skeleton.original;
1993        } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
1994            equal = curElem->skeleton->baseOriginal == skeleton.baseOriginal;
1995        }
1996        if (equal) {
1997            if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
1998                *specifiedSkeletonPtr = curElem->skeleton.getAlias();
1999            }
2000            return &(curElem->pattern);
2001        }
2002        curElem = curElem->next.getAlias();
2003    } while (curElem != nullptr);
2004 
2005    return nullptr;
2006 }
2007 
2008 UBool
equals(const PatternMap & other) const2009 PatternMap::equals(const PatternMap& other) const {
2010     if ( this==&other ) {
2011         return TRUE;
2012     }
2013     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
2014         if (boot[bootIndex] == other.boot[bootIndex]) {
2015             continue;
2016         }
2017         if ((boot[bootIndex] == nullptr) || (other.boot[bootIndex] == nullptr)) {
2018             return FALSE;
2019         }
2020         PtnElem *otherElem = other.boot[bootIndex];
2021         PtnElem *myElem = boot[bootIndex];
2022         while ((otherElem != nullptr) || (myElem != nullptr)) {
2023             if ( myElem == otherElem ) {
2024                 break;
2025             }
2026             if ((otherElem == nullptr) || (myElem == nullptr)) {
2027                 return FALSE;
2028             }
2029             if ( (myElem->basePattern != otherElem->basePattern) ||
2030                  (myElem->pattern != otherElem->pattern) ) {
2031                 return FALSE;
2032             }
2033             if ((myElem->skeleton.getAlias() != otherElem->skeleton.getAlias()) &&
2034                 !myElem->skeleton->equals(*(otherElem->skeleton))) {
2035                 return FALSE;
2036             }
2037             myElem = myElem->next.getAlias();
2038             otherElem = otherElem->next.getAlias();
2039         }
2040     }
2041     return TRUE;
2042 }
2043 
2044 // find any key existing in the mapping table already.
2045 // return TRUE if there is an existing key, otherwise return FALSE.
2046 PtnElem*
getDuplicateElem(const UnicodeString & basePattern,const PtnSkeleton & skeleton,PtnElem * baseElem)2047 PatternMap::getDuplicateElem(
2048             const UnicodeString &basePattern,
2049             const PtnSkeleton &skeleton,
2050             PtnElem *baseElem) {
2051    PtnElem *curElem;
2052 
2053    if ( baseElem == nullptr ) {
2054          return nullptr;
2055    }
2056    else {
2057          curElem = baseElem;
2058    }
2059    do {
2060      if ( basePattern.compare(curElem->basePattern)==0 ) {
2061          UBool isEqual = TRUE;
2062          for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2063             if (curElem->skeleton->type[i] != skeleton.type[i] ) {
2064                 isEqual = FALSE;
2065                 break;
2066             }
2067         }
2068         if (isEqual) {
2069             return curElem;
2070         }
2071      }
2072      curElem = curElem->next.getAlias();
2073    } while( curElem != nullptr );
2074 
2075    // end of the list
2076    return nullptr;
2077 
2078 }  // PatternMap::getDuplicateElem
2079 
DateTimeMatcher(void)2080 DateTimeMatcher::DateTimeMatcher(void) {
2081 }
2082 
~DateTimeMatcher()2083 DateTimeMatcher::~DateTimeMatcher() {}
2084 
DateTimeMatcher(const DateTimeMatcher & other)2085 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
2086     copyFrom(other.skeleton);
2087 }
2088 
2089 
2090 void
set(const UnicodeString & pattern,FormatParser * fp)2091 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
2092     PtnSkeleton localSkeleton;
2093     return set(pattern, fp, localSkeleton);
2094 }
2095 
2096 void
set(const UnicodeString & pattern,FormatParser * fp,PtnSkeleton & skeletonResult)2097 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
2098     int32_t i;
2099     for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
2100         skeletonResult.type[i] = NONE;
2101     }
2102     skeletonResult.original.clear();
2103     skeletonResult.baseOriginal.clear();
2104     skeletonResult.addedDefaultDayPeriod = FALSE;
2105 
2106     fp->set(pattern);
2107     for (i=0; i < fp->itemNumber; i++) {
2108         const UnicodeString& value = fp->items[i];
2109         // don't skip 'a' anymore, dayPeriod handled specially below
2110 
2111         if ( fp->isQuoteLiteral(value) ) {
2112             UnicodeString quoteLiteral;
2113             fp->getQuoteLiteral(quoteLiteral, &i);
2114             continue;
2115         }
2116         int32_t canonicalIndex = fp->getCanonicalIndex(value);
2117         if (canonicalIndex < 0) {
2118             continue;
2119         }
2120         const dtTypeElem *row = &dtTypes[canonicalIndex];
2121         int32_t field = row->field;
2122         skeletonResult.original.populate(field, value);
2123         UChar repeatChar = row->patternChar;
2124         int32_t repeatCount = row->minLen;
2125         skeletonResult.baseOriginal.populate(field, repeatChar, repeatCount);
2126         int16_t subField = row->type;
2127         if (row->type > 0) {
2128             U_ASSERT(value.length() < INT16_MAX);
2129             subField += static_cast<int16_t>(value.length());
2130         }
2131         skeletonResult.type[field] = subField;
2132     }
2133     // #13183, handle special behavior for day period characters (a, b, B)
2134     if (!skeletonResult.original.isFieldEmpty(UDATPG_HOUR_FIELD)) {
2135         if (skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==LOW_H || skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==CAP_K) {
2136             // We have a skeleton with 12-hour-cycle format
2137             if (skeletonResult.original.isFieldEmpty(UDATPG_DAYPERIOD_FIELD)) {
2138                 // But we do not have a day period in the skeleton; add the default DAYPERIOD (currently "a")
2139                 for (i = 0; dtTypes[i].patternChar != 0; i++) {
2140                     if ( dtTypes[i].field == UDATPG_DAYPERIOD_FIELD ) {
2141                         // first entry for UDATPG_DAYPERIOD_FIELD
2142                         skeletonResult.original.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2143                         skeletonResult.baseOriginal.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2144                         skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = dtTypes[i].type;
2145                         skeletonResult.addedDefaultDayPeriod = TRUE;
2146                         break;
2147                     }
2148                 }
2149             }
2150         } else {
2151             // Skeleton has 24-hour-cycle hour format and has dayPeriod, delete dayPeriod (i.e. ignore it)
2152             skeletonResult.original.clearField(UDATPG_DAYPERIOD_FIELD);
2153             skeletonResult.baseOriginal.clearField(UDATPG_DAYPERIOD_FIELD);
2154             skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = NONE;
2155         }
2156     }
2157     copyFrom(skeletonResult);
2158 }
2159 
2160 void
getBasePattern(UnicodeString & result)2161 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
2162     result.remove(); // Reset the result first.
2163     skeleton.baseOriginal.appendTo(result);
2164 }
2165 
2166 UnicodeString
getPattern()2167 DateTimeMatcher::getPattern() {
2168     UnicodeString result;
2169     return skeleton.original.appendTo(result);
2170 }
2171 
2172 int32_t
getDistance(const DateTimeMatcher & other,int32_t includeMask,DistanceInfo & distanceInfo) const2173 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) const {
2174     int32_t result = 0;
2175     distanceInfo.clear();
2176     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
2177         int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
2178         int32_t otherType = other.skeleton.type[i];
2179         if (myType==otherType) {
2180             continue;
2181         }
2182         if (myType==0) {// and other is not
2183             result += EXTRA_FIELD;
2184             distanceInfo.addExtra(i);
2185         }
2186         else {
2187             if (otherType==0) {
2188                 result += MISSING_FIELD;
2189                 distanceInfo.addMissing(i);
2190             }
2191             else {
2192                 result += abs(myType - otherType);
2193             }
2194         }
2195 
2196     }
2197     return result;
2198 }
2199 
2200 void
copyFrom(const PtnSkeleton & newSkeleton)2201 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
2202     skeleton.copyFrom(newSkeleton);
2203 }
2204 
2205 void
copyFrom()2206 DateTimeMatcher::copyFrom() {
2207     // same as clear
2208     skeleton.clear();
2209 }
2210 
2211 UBool
equals(const DateTimeMatcher * other) const2212 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
2213     if (other==nullptr) { return FALSE; }
2214     return skeleton.original == other->skeleton.original;
2215 }
2216 
2217 int32_t
getFieldMask() const2218 DateTimeMatcher::getFieldMask() const {
2219     int32_t result = 0;
2220 
2221     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2222         if (skeleton.type[i]!=0) {
2223             result |= (1<<i);
2224         }
2225     }
2226     return result;
2227 }
2228 
2229 PtnSkeleton*
getSkeletonPtr()2230 DateTimeMatcher::getSkeletonPtr() {
2231     return &skeleton;
2232 }
2233 
FormatParser()2234 FormatParser::FormatParser () {
2235     status = START;
2236     itemNumber = 0;
2237 }
2238 
2239 
~FormatParser()2240 FormatParser::~FormatParser () {
2241 }
2242 
2243 
2244 // Find the next token with the starting position and length
2245 // Note: the startPos may
2246 FormatParser::TokenStatus
setTokens(const UnicodeString & pattern,int32_t startPos,int32_t * len)2247 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
2248     int32_t curLoc = startPos;
2249     if ( curLoc >= pattern.length()) {
2250         return DONE;
2251     }
2252     // check the current char is between A-Z or a-z
2253     do {
2254         UChar c=pattern.charAt(curLoc);
2255         if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
2256            curLoc++;
2257         }
2258         else {
2259                startPos = curLoc;
2260                *len=1;
2261                return ADD_TOKEN;
2262         }
2263 
2264         if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
2265             break;  // not the same token
2266         }
2267     } while(curLoc <= pattern.length());
2268     *len = curLoc-startPos;
2269     return ADD_TOKEN;
2270 }
2271 
2272 void
set(const UnicodeString & pattern)2273 FormatParser::set(const UnicodeString& pattern) {
2274     int32_t startPos = 0;
2275     TokenStatus result = START;
2276     int32_t len = 0;
2277     itemNumber = 0;
2278 
2279     do {
2280         result = setTokens( pattern, startPos, &len );
2281         if ( result == ADD_TOKEN )
2282         {
2283             items[itemNumber++] = UnicodeString(pattern, startPos, len );
2284             startPos += len;
2285         }
2286         else {
2287             break;
2288         }
2289     } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
2290 }
2291 
2292 int32_t
getCanonicalIndex(const UnicodeString & s,UBool strict)2293 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
2294     int32_t len = s.length();
2295     if (len == 0) {
2296         return -1;
2297     }
2298     UChar ch = s.charAt(0);
2299 
2300     // Verify that all are the same character.
2301     for (int32_t l = 1; l < len; l++) {
2302         if (ch != s.charAt(l)) {
2303             return -1;
2304         }
2305     }
2306     int32_t i = 0;
2307     int32_t bestRow = -1;
2308     while (dtTypes[i].patternChar != 0x0000) {
2309         if ( dtTypes[i].patternChar != ch ) {
2310             ++i;
2311             continue;
2312         }
2313         bestRow = i;
2314         if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
2315             return i;
2316         }
2317         if (dtTypes[i+1].minLen <= len) {
2318             ++i;
2319             continue;
2320         }
2321         return i;
2322     }
2323     return strict ? -1 : bestRow;
2324 }
2325 
2326 UBool
isQuoteLiteral(const UnicodeString & s)2327 FormatParser::isQuoteLiteral(const UnicodeString& s) {
2328     return (UBool)(s.charAt(0) == SINGLE_QUOTE);
2329 }
2330 
2331 // This function assumes the current itemIndex points to the quote literal.
2332 // Please call isQuoteLiteral prior to this function.
2333 void
getQuoteLiteral(UnicodeString & quote,int32_t * itemIndex)2334 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
2335     int32_t i = *itemIndex;
2336 
2337     quote.remove();
2338     if (items[i].charAt(0)==SINGLE_QUOTE) {
2339         quote += items[i];
2340         ++i;
2341     }
2342     while ( i < itemNumber ) {
2343         if ( items[i].charAt(0)==SINGLE_QUOTE ) {
2344             if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
2345                 // two single quotes e.g. 'o''clock'
2346                 quote += items[i++];
2347                 quote += items[i++];
2348                 continue;
2349             }
2350             else {
2351                 quote += items[i];
2352                 break;
2353             }
2354         }
2355         else {
2356             quote += items[i];
2357         }
2358         ++i;
2359     }
2360     *itemIndex=i;
2361 }
2362 
2363 UBool
isPatternSeparator(const UnicodeString & field) const2364 FormatParser::isPatternSeparator(const UnicodeString& field) const {
2365     for (int32_t i=0; i<field.length(); ++i ) {
2366         UChar c= field.charAt(i);
2367         if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
2368              (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
2369             continue;
2370         }
2371         else {
2372             return FALSE;
2373         }
2374     }
2375     return TRUE;
2376 }
2377 
~DistanceInfo()2378 DistanceInfo::~DistanceInfo() {}
2379 
2380 void
setTo(const DistanceInfo & other)2381 DistanceInfo::setTo(const DistanceInfo& other) {
2382     missingFieldMask = other.missingFieldMask;
2383     extraFieldMask= other.extraFieldMask;
2384 }
2385 
PatternMapIterator(UErrorCode & status)2386 PatternMapIterator::PatternMapIterator(UErrorCode& status) :
2387     bootIndex(0), nodePtr(nullptr), matcher(nullptr), patternMap(nullptr)
2388 {
2389     if (U_FAILURE(status)) { return; }
2390     matcher.adoptInsteadAndCheckErrorCode(new DateTimeMatcher(), status);
2391 }
2392 
~PatternMapIterator()2393 PatternMapIterator::~PatternMapIterator() {
2394 }
2395 
2396 void
set(PatternMap & newPatternMap)2397 PatternMapIterator::set(PatternMap& newPatternMap) {
2398     this->patternMap=&newPatternMap;
2399 }
2400 
2401 PtnSkeleton*
getSkeleton() const2402 PatternMapIterator::getSkeleton() const {
2403     if ( nodePtr == nullptr ) {
2404         return nullptr;
2405     }
2406     else {
2407         return nodePtr->skeleton.getAlias();
2408     }
2409 }
2410 
2411 UBool
hasNext() const2412 PatternMapIterator::hasNext() const {
2413     int32_t headIndex = bootIndex;
2414     PtnElem *curPtr = nodePtr;
2415 
2416     if (patternMap==nullptr) {
2417         return FALSE;
2418     }
2419     while ( headIndex < MAX_PATTERN_ENTRIES ) {
2420         if ( curPtr != nullptr ) {
2421             if ( curPtr->next != nullptr ) {
2422                 return TRUE;
2423             }
2424             else {
2425                 headIndex++;
2426                 curPtr=nullptr;
2427                 continue;
2428             }
2429         }
2430         else {
2431             if ( patternMap->boot[headIndex] != nullptr ) {
2432                 return TRUE;
2433             }
2434             else {
2435                 headIndex++;
2436                 continue;
2437             }
2438         }
2439     }
2440     return FALSE;
2441 }
2442 
2443 DateTimeMatcher&
next()2444 PatternMapIterator::next() {
2445     while ( bootIndex < MAX_PATTERN_ENTRIES ) {
2446         if ( nodePtr != nullptr ) {
2447             if ( nodePtr->next != nullptr ) {
2448                 nodePtr = nodePtr->next.getAlias();
2449                 break;
2450             }
2451             else {
2452                 bootIndex++;
2453                 nodePtr=nullptr;
2454                 continue;
2455             }
2456         }
2457         else {
2458             if ( patternMap->boot[bootIndex] != nullptr ) {
2459                 nodePtr = patternMap->boot[bootIndex];
2460                 break;
2461             }
2462             else {
2463                 bootIndex++;
2464                 continue;
2465             }
2466         }
2467     }
2468     if (nodePtr!=nullptr) {
2469         matcher->copyFrom(*nodePtr->skeleton);
2470     }
2471     else {
2472         matcher->copyFrom();
2473     }
2474     return *matcher;
2475 }
2476 
2477 
SkeletonFields()2478 SkeletonFields::SkeletonFields() {
2479     // Set initial values to zero
2480     clear();
2481 }
2482 
clear()2483 void SkeletonFields::clear() {
2484     uprv_memset(chars, 0, sizeof(chars));
2485     uprv_memset(lengths, 0, sizeof(lengths));
2486 }
2487 
copyFrom(const SkeletonFields & other)2488 void SkeletonFields::copyFrom(const SkeletonFields& other) {
2489     uprv_memcpy(chars, other.chars, sizeof(chars));
2490     uprv_memcpy(lengths, other.lengths, sizeof(lengths));
2491 }
2492 
clearField(int32_t field)2493 void SkeletonFields::clearField(int32_t field) {
2494     chars[field] = 0;
2495     lengths[field] = 0;
2496 }
2497 
getFieldChar(int32_t field) const2498 UChar SkeletonFields::getFieldChar(int32_t field) const {
2499     return chars[field];
2500 }
2501 
getFieldLength(int32_t field) const2502 int32_t SkeletonFields::getFieldLength(int32_t field) const {
2503     return lengths[field];
2504 }
2505 
populate(int32_t field,const UnicodeString & value)2506 void SkeletonFields::populate(int32_t field, const UnicodeString& value) {
2507     populate(field, value.charAt(0), value.length());
2508 }
2509 
populate(int32_t field,UChar ch,int32_t length)2510 void SkeletonFields::populate(int32_t field, UChar ch, int32_t length) {
2511     chars[field] = (int8_t) ch;
2512     lengths[field] = (int8_t) length;
2513 }
2514 
isFieldEmpty(int32_t field) const2515 UBool SkeletonFields::isFieldEmpty(int32_t field) const {
2516     return lengths[field] == 0;
2517 }
2518 
appendTo(UnicodeString & string) const2519 UnicodeString& SkeletonFields::appendTo(UnicodeString& string) const {
2520     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2521         appendFieldTo(i, string);
2522     }
2523     return string;
2524 }
2525 
appendFieldTo(int32_t field,UnicodeString & string) const2526 UnicodeString& SkeletonFields::appendFieldTo(int32_t field, UnicodeString& string) const {
2527     UChar ch(chars[field]);
2528     int32_t length = (int32_t) lengths[field];
2529 
2530     for (int32_t i=0; i<length; i++) {
2531         string += ch;
2532     }
2533     return string;
2534 }
2535 
getFirstChar() const2536 UChar SkeletonFields::getFirstChar() const {
2537     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2538         if (lengths[i] != 0) {
2539             return chars[i];
2540         }
2541     }
2542     return '\0';
2543 }
2544 
2545 
PtnSkeleton()2546 PtnSkeleton::PtnSkeleton() {
2547 }
2548 
PtnSkeleton(const PtnSkeleton & other)2549 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
2550     copyFrom(other);
2551 }
2552 
copyFrom(const PtnSkeleton & other)2553 void PtnSkeleton::copyFrom(const PtnSkeleton& other) {
2554     uprv_memcpy(type, other.type, sizeof(type));
2555     original.copyFrom(other.original);
2556     baseOriginal.copyFrom(other.baseOriginal);
2557 }
2558 
clear()2559 void PtnSkeleton::clear() {
2560     uprv_memset(type, 0, sizeof(type));
2561     original.clear();
2562     baseOriginal.clear();
2563 }
2564 
2565 UBool
equals(const PtnSkeleton & other) const2566 PtnSkeleton::equals(const PtnSkeleton& other) const  {
2567     return (original == other.original)
2568         && (baseOriginal == other.baseOriginal)
2569         && (uprv_memcmp(type, other.type, sizeof(type)) == 0);
2570 }
2571 
2572 UnicodeString
getSkeleton() const2573 PtnSkeleton::getSkeleton() const {
2574     UnicodeString result;
2575     result = original.appendTo(result);
2576     int32_t pos;
2577     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2578         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2579         // was not in the provided skeleton, remove it here before returning skeleton.
2580         result.remove(pos, 1);
2581     }
2582     return result;
2583 }
2584 
2585 UnicodeString
getBaseSkeleton() const2586 PtnSkeleton::getBaseSkeleton() const {
2587     UnicodeString result;
2588     result = baseOriginal.appendTo(result);
2589     int32_t pos;
2590     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2591         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2592         // was not in the provided skeleton, remove it here before returning skeleton.
2593         result.remove(pos, 1);
2594     }
2595     return result;
2596 }
2597 
2598 UChar
getFirstChar() const2599 PtnSkeleton::getFirstChar() const {
2600     return baseOriginal.getFirstChar();
2601 }
2602 
~PtnSkeleton()2603 PtnSkeleton::~PtnSkeleton() {
2604 }
2605 
PtnElem(const UnicodeString & basePat,const UnicodeString & pat)2606 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
2607     basePattern(basePat), skeleton(nullptr), pattern(pat), next(nullptr)
2608 {
2609 }
2610 
~PtnElem()2611 PtnElem::~PtnElem() {
2612 }
2613 
DTSkeletonEnumeration(PatternMap & patternMap,dtStrEnum type,UErrorCode & status)2614 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap& patternMap, dtStrEnum type, UErrorCode& status) : fSkeletons(nullptr) {
2615     PtnElem  *curElem;
2616     PtnSkeleton *curSkeleton;
2617     UnicodeString s;
2618     int32_t bootIndex;
2619 
2620     pos=0;
2621     fSkeletons.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2622     if (U_FAILURE(status)) {
2623         return;
2624     }
2625 
2626     for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2627         curElem = patternMap.boot[bootIndex];
2628         while (curElem!=nullptr) {
2629             switch(type) {
2630                 case DT_BASESKELETON:
2631                     s=curElem->basePattern;
2632                     break;
2633                 case DT_PATTERN:
2634                     s=curElem->pattern;
2635                     break;
2636                 case DT_SKELETON:
2637                     curSkeleton=curElem->skeleton.getAlias();
2638                     s=curSkeleton->getSkeleton();
2639                     break;
2640             }
2641             if ( !isCanonicalItem(s) ) {
2642                 LocalPointer<UnicodeString> newElem(new UnicodeString(s), status);
2643                 if (U_FAILURE(status)) {
2644                     return;
2645                 }
2646                 fSkeletons->addElement(newElem.getAlias(), status);
2647                 if (U_FAILURE(status)) {
2648                     fSkeletons.adoptInstead(nullptr);
2649                     return;
2650                 }
2651                 newElem.orphan(); // fSkeletons vector now owns the UnicodeString.
2652             }
2653             curElem = curElem->next.getAlias();
2654         }
2655     }
2656     if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=nullptr) ) {
2657         status = U_BUFFER_OVERFLOW_ERROR;
2658     }
2659 }
2660 
2661 const UnicodeString*
snext(UErrorCode & status)2662 DTSkeletonEnumeration::snext(UErrorCode& status) {
2663     if (U_SUCCESS(status) && fSkeletons.isValid() && pos < fSkeletons->size()) {
2664         return (const UnicodeString*)fSkeletons->elementAt(pos++);
2665     }
2666     return nullptr;
2667 }
2668 
2669 void
reset(UErrorCode &)2670 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2671     pos=0;
2672 }
2673 
2674 int32_t
count(UErrorCode &) const2675 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2676    return (fSkeletons.isNull()) ? 0 : fSkeletons->size();
2677 }
2678 
2679 UBool
isCanonicalItem(const UnicodeString & item)2680 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2681     if ( item.length() != 1 ) {
2682         return FALSE;
2683     }
2684     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2685         if (item.charAt(0)==Canonical_Items[i]) {
2686             return TRUE;
2687         }
2688     }
2689     return FALSE;
2690 }
2691 
~DTSkeletonEnumeration()2692 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2693     UnicodeString *s;
2694     if (fSkeletons.isValid()) {
2695         for (int32_t i = 0; i < fSkeletons->size(); ++i) {
2696             if ((s = (UnicodeString *)fSkeletons->elementAt(i)) != nullptr) {
2697                 delete s;
2698             }
2699         }
2700     }
2701 }
2702 
DTRedundantEnumeration()2703 DTRedundantEnumeration::DTRedundantEnumeration() : pos(0), fPatterns(nullptr) {
2704 }
2705 
2706 void
add(const UnicodeString & pattern,UErrorCode & status)2707 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2708     if (U_FAILURE(status)) { return; }
2709     if (fPatterns.isNull())  {
2710         fPatterns.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2711         if (U_FAILURE(status)) {
2712             return;
2713        }
2714     }
2715     LocalPointer<UnicodeString> newElem(new UnicodeString(pattern), status);
2716     if (U_FAILURE(status)) {
2717         return;
2718     }
2719     fPatterns->addElement(newElem.getAlias(), status);
2720     if (U_FAILURE(status)) {
2721         fPatterns.adoptInstead(nullptr);
2722         return;
2723     }
2724     newElem.orphan(); // fPatterns now owns the string.
2725 }
2726 
2727 const UnicodeString*
snext(UErrorCode & status)2728 DTRedundantEnumeration::snext(UErrorCode& status) {
2729     if (U_SUCCESS(status) && fPatterns.isValid() && pos < fPatterns->size()) {
2730         return (const UnicodeString*)fPatterns->elementAt(pos++);
2731     }
2732     return nullptr;
2733 }
2734 
2735 void
reset(UErrorCode &)2736 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2737     pos=0;
2738 }
2739 
2740 int32_t
count(UErrorCode &) const2741 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2742     return (fPatterns.isNull()) ? 0 : fPatterns->size();
2743 }
2744 
2745 UBool
isCanonicalItem(const UnicodeString & item) const2746 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) const {
2747     if ( item.length() != 1 ) {
2748         return FALSE;
2749     }
2750     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2751         if (item.charAt(0)==Canonical_Items[i]) {
2752             return TRUE;
2753         }
2754     }
2755     return FALSE;
2756 }
2757 
~DTRedundantEnumeration()2758 DTRedundantEnumeration::~DTRedundantEnumeration() {
2759     UnicodeString *s;
2760     if (fPatterns.isValid()) {
2761         for (int32_t i = 0; i < fPatterns->size(); ++i) {
2762             if ((s = (UnicodeString *)fPatterns->elementAt(i)) != nullptr) {
2763                 delete s;
2764             }
2765         }
2766     }
2767 }
2768 
2769 U_NAMESPACE_END
2770 
2771 
2772 #endif /* #if !UCONFIG_NO_FORMATTING */
2773 
2774 //eof
2775