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 "unicode/region.h"
32 #include "cpputils.h"
33 #include "mutex.h"
34 #include "umutex.h"
35 #include "cmemory.h"
36 #include "cstring.h"
37 #include "locbased.h"
38 #include "hash.h"
39 #include "uhash.h"
40 #include "uresimp.h"
41 #include "dtptngen_impl.h"
42 #include "ucln_in.h"
43 #include "charstr.h"
44 #include "uassert.h"
45 
46 #if U_CHARSET_FAMILY==U_EBCDIC_FAMILY
47 /**
48  * If we are on EBCDIC, use an iterator which will
49  * traverse the bundles in ASCII order.
50  */
51 #define U_USE_ASCII_BUNDLE_ITERATOR
52 #define U_SORT_ASCII_BUNDLE_ITERATOR
53 #endif
54 
55 #if defined(U_USE_ASCII_BUNDLE_ITERATOR)
56 
57 #include "unicode/ustring.h"
58 #include "uarrsort.h"
59 
60 struct UResAEntry {
61     UChar *key;
62     UResourceBundle *item;
63 };
64 
65 struct UResourceBundleAIterator {
66     UResourceBundle  *bund;
67     UResAEntry *entries;
68     int32_t num;
69     int32_t cursor;
70 };
71 
72 /* Must be C linkage to pass function pointer to the sort function */
73 
74 U_CDECL_BEGIN
75 
76 static int32_t U_CALLCONV
ures_a_codepointSort(const void * context,const void * left,const void * right)77 ures_a_codepointSort(const void *context, const void *left, const void *right) {
78     //CompareContext *cmp=(CompareContext *)context;
79     return u_strcmp(((const UResAEntry *)left)->key,
80                     ((const UResAEntry *)right)->key);
81 }
82 
83 U_CDECL_END
84 
ures_a_open(UResourceBundleAIterator * aiter,UResourceBundle * bund,UErrorCode * status)85 static void ures_a_open(UResourceBundleAIterator *aiter, UResourceBundle *bund, UErrorCode *status) {
86     if(U_FAILURE(*status)) {
87         return;
88     }
89     aiter->bund = bund;
90     aiter->num = ures_getSize(aiter->bund);
91     aiter->cursor = 0;
92 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
93     aiter->entries = nullptr;
94 #else
95     aiter->entries = (UResAEntry*)uprv_malloc(sizeof(UResAEntry)*aiter->num);
96     for(int i=0;i<aiter->num;i++) {
97         aiter->entries[i].item = ures_getByIndex(aiter->bund, i, nullptr, status);
98         const char *akey = ures_getKey(aiter->entries[i].item);
99         int32_t len = uprv_strlen(akey)+1;
100         aiter->entries[i].key = (UChar*)uprv_malloc(len*sizeof(UChar));
101         u_charsToUChars(akey, aiter->entries[i].key, len);
102     }
103     uprv_sortArray(aiter->entries, aiter->num, sizeof(UResAEntry), ures_a_codepointSort, nullptr, TRUE, status);
104 #endif
105 }
106 
ures_a_close(UResourceBundleAIterator * aiter)107 static void ures_a_close(UResourceBundleAIterator *aiter) {
108 #if defined(U_SORT_ASCII_BUNDLE_ITERATOR)
109     for(int i=0;i<aiter->num;i++) {
110         uprv_free(aiter->entries[i].key);
111         ures_close(aiter->entries[i].item);
112     }
113 #endif
114 }
115 
ures_a_getNextString(UResourceBundleAIterator * aiter,int32_t * len,const char ** key,UErrorCode * err)116 static const UChar *ures_a_getNextString(UResourceBundleAIterator *aiter, int32_t *len, const char **key, UErrorCode *err) {
117 #if !defined(U_SORT_ASCII_BUNDLE_ITERATOR)
118     return ures_getNextString(aiter->bund, len, key, err);
119 #else
120     if(U_FAILURE(*err)) return nullptr;
121     UResourceBundle *item = aiter->entries[aiter->cursor].item;
122     const UChar* ret = ures_getString(item, len, err);
123     *key = ures_getKey(item);
124     aiter->cursor++;
125     return ret;
126 #endif
127 }
128 
129 
130 #endif
131 
132 
133 U_NAMESPACE_BEGIN
134 
135 // *****************************************************************************
136 // class DateTimePatternGenerator
137 // *****************************************************************************
138 static const UChar Canonical_Items[] = {
139     // GyQMwWEDFdaHmsSv
140     CAP_G, LOW_Y, CAP_Q, CAP_M, LOW_W, CAP_W, CAP_E,
141     CAP_D, CAP_F, LOW_D, LOW_A, // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J
142     CAP_H, LOW_M, LOW_S, CAP_S, LOW_V, 0
143 };
144 
145 static const dtTypeElem dtTypes[] = {
146     // patternChar, field, type, minLen, weight
147     {CAP_G, UDATPG_ERA_FIELD, DT_SHORT, 1, 3,},
148     {CAP_G, UDATPG_ERA_FIELD, DT_LONG,  4, 0},
149     {CAP_G, UDATPG_ERA_FIELD, DT_NARROW, 5, 0},
150 
151     {LOW_Y, UDATPG_YEAR_FIELD, DT_NUMERIC, 1, 20},
152     {CAP_Y, UDATPG_YEAR_FIELD, DT_NUMERIC + DT_DELTA, 1, 20},
153     {LOW_U, UDATPG_YEAR_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 20},
154     {LOW_R, UDATPG_YEAR_FIELD, DT_NUMERIC + 3*DT_DELTA, 1, 20},
155     {CAP_U, UDATPG_YEAR_FIELD, DT_SHORT, 1, 3},
156     {CAP_U, UDATPG_YEAR_FIELD, DT_LONG, 4, 0},
157     {CAP_U, UDATPG_YEAR_FIELD, DT_NARROW, 5, 0},
158 
159     {CAP_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC, 1, 2},
160     {CAP_Q, UDATPG_QUARTER_FIELD, DT_SHORT, 3, 0},
161     {CAP_Q, UDATPG_QUARTER_FIELD, DT_LONG, 4, 0},
162     {CAP_Q, UDATPG_QUARTER_FIELD, DT_NARROW, 5, 0},
163     {LOW_Q, UDATPG_QUARTER_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
164     {LOW_Q, UDATPG_QUARTER_FIELD, DT_SHORT - DT_DELTA, 3, 0},
165     {LOW_Q, UDATPG_QUARTER_FIELD, DT_LONG - DT_DELTA, 4, 0},
166     {LOW_Q, UDATPG_QUARTER_FIELD, DT_NARROW - DT_DELTA, 5, 0},
167 
168     {CAP_M, UDATPG_MONTH_FIELD, DT_NUMERIC, 1, 2},
169     {CAP_M, UDATPG_MONTH_FIELD, DT_SHORT, 3, 0},
170     {CAP_M, UDATPG_MONTH_FIELD, DT_LONG, 4, 0},
171     {CAP_M, UDATPG_MONTH_FIELD, DT_NARROW, 5, 0},
172     {CAP_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 2},
173     {CAP_L, UDATPG_MONTH_FIELD, DT_SHORT - DT_DELTA, 3, 0},
174     {CAP_L, UDATPG_MONTH_FIELD, DT_LONG - DT_DELTA, 4, 0},
175     {CAP_L, UDATPG_MONTH_FIELD, DT_NARROW - DT_DELTA, 5, 0},
176     {LOW_L, UDATPG_MONTH_FIELD, DT_NUMERIC + DT_DELTA, 1, 1},
177 
178     {LOW_W, UDATPG_WEEK_OF_YEAR_FIELD, DT_NUMERIC, 1, 2},
179 
180     {CAP_W, UDATPG_WEEK_OF_MONTH_FIELD, DT_NUMERIC, 1, 0},
181 
182     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORT, 1, 3},
183     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_LONG, 4, 0},
184     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_NARROW, 5, 0},
185     {CAP_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER, 6, 0},
186     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + 2*DT_DELTA, 1, 2},
187     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORT - 2*DT_DELTA, 3, 0},
188     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
189     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_NARROW - 2*DT_DELTA, 5, 0},
190     {LOW_C, UDATPG_WEEKDAY_FIELD, DT_SHORTER - 2*DT_DELTA, 6, 0},
191     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // LOW_E is currently not used in CLDR data, should not be canonical
192     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORT - DT_DELTA, 3, 0},
193     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_LONG - DT_DELTA, 4, 0},
194     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_NARROW - DT_DELTA, 5, 0},
195     {LOW_E, UDATPG_WEEKDAY_FIELD, DT_SHORTER - DT_DELTA, 6, 0},
196 
197     {LOW_D, UDATPG_DAY_FIELD, DT_NUMERIC, 1, 2},
198     {LOW_G, UDATPG_DAY_FIELD, DT_NUMERIC + DT_DELTA, 1, 20}, // really internal use, so we don't care
199 
200     {CAP_D, UDATPG_DAY_OF_YEAR_FIELD, DT_NUMERIC, 1, 3},
201 
202     {CAP_F, UDATPG_DAY_OF_WEEK_IN_MONTH_FIELD, DT_NUMERIC, 1, 0},
203 
204     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_SHORT, 1, 3},
205     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_LONG, 4, 0},
206     {LOW_A, UDATPG_DAYPERIOD_FIELD, DT_NARROW, 5, 0},
207     {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - DT_DELTA, 1, 3},
208     {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - DT_DELTA, 4, 0},
209     {LOW_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - DT_DELTA, 5, 0},
210     // b needs to be closer to a than to B, so we make this 3*DT_DELTA
211     {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_SHORT - 3*DT_DELTA, 1, 3},
212     {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_LONG - 3*DT_DELTA, 4, 0},
213     {CAP_B, UDATPG_DAYPERIOD_FIELD, DT_NARROW - 3*DT_DELTA, 5, 0},
214 
215     {CAP_H, UDATPG_HOUR_FIELD, DT_NUMERIC + 10*DT_DELTA, 1, 2}, // 24 hour
216     {LOW_K, UDATPG_HOUR_FIELD, DT_NUMERIC + 11*DT_DELTA, 1, 2}, // 24 hour
217     {LOW_H, UDATPG_HOUR_FIELD, DT_NUMERIC, 1, 2}, // 12 hour
218     {CAP_K, UDATPG_HOUR_FIELD, DT_NUMERIC + DT_DELTA, 1, 2}, // 12 hour
219     // The C code has had versions of the following 3, keep & update. Should not need these, but...
220     // Without these, certain tests using e.g. staticGetSkeleton fail because j/J in patterns
221     // get skipped instead of mapped to the right hour chars, for example in
222     //   DateFormatTest::TestPatternFromSkeleton
223     //   IntlTestDateTimePatternGeneratorAPI:: testStaticGetSkeleton
224     //   DateIntervalFormatTest::testTicket11985
225     // Need to investigate better handling of jJC replacement e.g. in staticGetSkeleton.
226     {CAP_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 5*DT_DELTA, 1, 2}, // 12/24 hour no AM/PM
227     {LOW_J, UDATPG_HOUR_FIELD, DT_NUMERIC + 6*DT_DELTA, 1, 6}, // 12/24 hour
228     {CAP_C, UDATPG_HOUR_FIELD, DT_NUMERIC + 7*DT_DELTA, 1, 6}, // 12/24 hour with preferred dayPeriods for 12
229 
230     {LOW_M, UDATPG_MINUTE_FIELD, DT_NUMERIC, 1, 2},
231 
232     {LOW_S, UDATPG_SECOND_FIELD, DT_NUMERIC, 1, 2},
233     {CAP_A, UDATPG_SECOND_FIELD, DT_NUMERIC + DT_DELTA, 1, 1000},
234 
235     {CAP_S, UDATPG_FRACTIONAL_SECOND_FIELD, DT_NUMERIC, 1, 1000},
236 
237     {LOW_V, UDATPG_ZONE_FIELD, DT_SHORT - 2*DT_DELTA, 1, 0},
238     {LOW_V, UDATPG_ZONE_FIELD, DT_LONG - 2*DT_DELTA, 4, 0},
239     {LOW_Z, UDATPG_ZONE_FIELD, DT_SHORT, 1, 3},
240     {LOW_Z, UDATPG_ZONE_FIELD, DT_LONG, 4, 0},
241     {CAP_Z, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 3},
242     {CAP_Z, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
243     {CAP_Z, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 5, 0},
244     {CAP_O, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0},
245     {CAP_O, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
246     {CAP_V, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 1, 0},
247     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 2, 0},
248     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-1 - DT_DELTA, 3, 0},
249     {CAP_V, UDATPG_ZONE_FIELD, DT_LONG-2 - DT_DELTA, 4, 0},
250     {CAP_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0},
251     {CAP_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0},
252     {CAP_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
253     {LOW_X, UDATPG_ZONE_FIELD, DT_NARROW - DT_DELTA, 1, 0},
254     {LOW_X, UDATPG_ZONE_FIELD, DT_SHORT - DT_DELTA, 2, 0},
255     {LOW_X, UDATPG_ZONE_FIELD, DT_LONG - DT_DELTA, 4, 0},
256 
257     {0, UDATPG_FIELD_COUNT, 0, 0, 0} , // last row of dtTypes[]
258  };
259 
260 static const char* const CLDR_FIELD_APPEND[] = {
261     "Era", "Year", "Quarter", "Month", "Week", "*", "Day-Of-Week",
262     "*", "*", "Day", "*", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J
263     "Hour", "Minute", "Second", "*", "Timezone"
264 };
265 
266 static const char* const CLDR_FIELD_NAME[UDATPG_FIELD_COUNT] = {
267     "era", "year", "quarter", "month", "week", "weekOfMonth", "weekday",
268     "dayOfYear", "weekdayOfMonth", "day", "dayperiod", // The UDATPG_x_FIELD constants and these fields have a different order than in ICU4J
269     "hour", "minute", "second", "*", "zone"
270 };
271 
272 static const char* const CLDR_FIELD_WIDTH[] = { // [UDATPG_WIDTH_COUNT]
273     "", "-short", "-narrow"
274 };
275 
276 // TODO(ticket:13619): remove when definition uncommented in dtptngen.h.
277 static const int32_t UDATPG_WIDTH_COUNT = UDATPG_NARROW + 1;
278 static constexpr UDateTimePGDisplayWidth UDATPG_WIDTH_APPENDITEM = UDATPG_WIDE;
279 static constexpr int32_t UDATPG_FIELD_KEY_MAX = 24; // max length of CLDR field tag (type + width)
280 
281 // For appendItems
282 static const UChar UDATPG_ItemFormat[]= {0x7B, 0x30, 0x7D, 0x20, 0x251C, 0x7B, 0x32, 0x7D, 0x3A,
283     0x20, 0x7B, 0x31, 0x7D, 0x2524, 0};  // {0} \u251C{2}: {1}\u2524
284 
285 //static const UChar repeatedPatterns[6]={CAP_G, CAP_E, LOW_Z, LOW_V, CAP_Q, 0}; // "GEzvQ"
286 
287 static const char DT_DateTimePatternsTag[]="DateTimePatterns";
288 static const char DT_DateTimeCalendarTag[]="calendar";
289 static const char DT_DateTimeGregorianTag[]="gregorian";
290 static const char DT_DateTimeAppendItemsTag[]="appendItems";
291 static const char DT_DateTimeFieldsTag[]="fields";
292 static const char DT_DateTimeAvailableFormatsTag[]="availableFormats";
293 //static const UnicodeString repeatedPattern=UnicodeString(repeatedPatterns);
294 
295 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateTimePatternGenerator)
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)296 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTSkeletonEnumeration)
297 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DTRedundantEnumeration)
298 
299 DateTimePatternGenerator*  U_EXPORT2
300 DateTimePatternGenerator::createInstance(UErrorCode& status) {
301     return createInstance(Locale::getDefault(), status);
302 }
303 
304 DateTimePatternGenerator* U_EXPORT2
createInstance(const Locale & locale,UErrorCode & status)305 DateTimePatternGenerator::createInstance(const Locale& locale, UErrorCode& status) {
306     if (U_FAILURE(status)) {
307         return nullptr;
308     }
309     LocalPointer<DateTimePatternGenerator> result(
310             new DateTimePatternGenerator(locale, status), status);
311     return U_SUCCESS(status) ? result.orphan() : nullptr;
312 }
313 
314 DateTimePatternGenerator*  U_EXPORT2
createEmptyInstance(UErrorCode & status)315 DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) {
316     if (U_FAILURE(status)) {
317         return nullptr;
318     }
319     LocalPointer<DateTimePatternGenerator> result(
320             new DateTimePatternGenerator(status), status);
321     return U_SUCCESS(status) ? result.orphan() : nullptr;
322 }
323 
DateTimePatternGenerator(UErrorCode & status)324 DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) :
325     skipMatcher(nullptr),
326     fAvailableFormatKeyHash(nullptr),
327     internalErrorCode(U_ZERO_ERROR)
328 {
329     fp = new FormatParser();
330     dtMatcher = new DateTimeMatcher();
331     distanceInfo = new DistanceInfo();
332     patternMap = new PatternMap();
333     if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) {
334         internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR;
335     }
336 }
337 
DateTimePatternGenerator(const Locale & locale,UErrorCode & status)338 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status) :
339     skipMatcher(nullptr),
340     fAvailableFormatKeyHash(nullptr),
341     internalErrorCode(U_ZERO_ERROR)
342 {
343     fp = new FormatParser();
344     dtMatcher = new DateTimeMatcher();
345     distanceInfo = new DistanceInfo();
346     patternMap = new PatternMap();
347     if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) {
348         internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR;
349     }
350     else {
351         initData(locale, status);
352     }
353 }
354 
DateTimePatternGenerator(const DateTimePatternGenerator & other)355 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) :
356     UObject(),
357     skipMatcher(nullptr),
358     fAvailableFormatKeyHash(nullptr),
359     internalErrorCode(U_ZERO_ERROR)
360 {
361     fp = new FormatParser();
362     dtMatcher = new DateTimeMatcher();
363     distanceInfo = new DistanceInfo();
364     patternMap = new PatternMap();
365     if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) {
366         internalErrorCode = U_MEMORY_ALLOCATION_ERROR;
367     }
368     *this=other;
369 }
370 
371 DateTimePatternGenerator&
operator =(const DateTimePatternGenerator & other)372 DateTimePatternGenerator::operator=(const DateTimePatternGenerator& other) {
373     // reflexive case
374     if (&other == this) {
375         return *this;
376     }
377     internalErrorCode = other.internalErrorCode;
378     pLocale = other.pLocale;
379     fDefaultHourFormatChar = other.fDefaultHourFormatChar;
380     *fp = *(other.fp);
381     dtMatcher->copyFrom(other.dtMatcher->skeleton);
382     *distanceInfo = *(other.distanceInfo);
383     dateTimeFormat = other.dateTimeFormat;
384     decimal = other.decimal;
385     // NUL-terminate for the C API.
386     dateTimeFormat.getTerminatedBuffer();
387     decimal.getTerminatedBuffer();
388     delete skipMatcher;
389     if ( other.skipMatcher == nullptr ) {
390         skipMatcher = nullptr;
391     }
392     else {
393         skipMatcher = new DateTimeMatcher(*other.skipMatcher);
394         if (skipMatcher == nullptr)
395         {
396             internalErrorCode = U_MEMORY_ALLOCATION_ERROR;
397             return *this;
398         }
399     }
400     for (int32_t i=0; i< UDATPG_FIELD_COUNT; ++i ) {
401         appendItemFormats[i] = other.appendItemFormats[i];
402         appendItemFormats[i].getTerminatedBuffer(); // NUL-terminate for the C API.
403         for (int32_t j=0; j< UDATPG_WIDTH_COUNT; ++j ) {
404             fieldDisplayNames[i][j] = other.fieldDisplayNames[i][j];
405             fieldDisplayNames[i][j].getTerminatedBuffer(); // NUL-terminate for the C API.
406         }
407     }
408     patternMap->copyFrom(*other.patternMap, internalErrorCode);
409     copyHashtable(other.fAvailableFormatKeyHash, internalErrorCode);
410     return *this;
411 }
412 
413 
414 UBool
operator ==(const DateTimePatternGenerator & other) const415 DateTimePatternGenerator::operator==(const DateTimePatternGenerator& other) const {
416     if (this == &other) {
417         return TRUE;
418     }
419     if ((pLocale==other.pLocale) && (patternMap->equals(*other.patternMap)) &&
420         (dateTimeFormat==other.dateTimeFormat) && (decimal==other.decimal)) {
421         for ( int32_t i=0 ; i<UDATPG_FIELD_COUNT; ++i ) {
422             if (appendItemFormats[i] != other.appendItemFormats[i]) {
423                 return FALSE;
424             }
425             for (int32_t j=0; j< UDATPG_WIDTH_COUNT; ++j ) {
426                 if (fieldDisplayNames[i][j] != other.fieldDisplayNames[i][j]) {
427                     return FALSE;
428                 }
429             }
430         }
431         return TRUE;
432     }
433     else {
434         return FALSE;
435     }
436 }
437 
438 UBool
operator !=(const DateTimePatternGenerator & other) const439 DateTimePatternGenerator::operator!=(const DateTimePatternGenerator& other) const {
440     return  !operator==(other);
441 }
442 
~DateTimePatternGenerator()443 DateTimePatternGenerator::~DateTimePatternGenerator() {
444     if (fAvailableFormatKeyHash!=nullptr) {
445         delete fAvailableFormatKeyHash;
446     }
447 
448     if (fp != nullptr) delete fp;
449     if (dtMatcher != nullptr) delete dtMatcher;
450     if (distanceInfo != nullptr) delete distanceInfo;
451     if (patternMap != nullptr) delete patternMap;
452     if (skipMatcher != nullptr) delete skipMatcher;
453 }
454 
455 namespace {
456 
457 UInitOnce initOnce = U_INITONCE_INITIALIZER;
458 UHashtable *localeToAllowedHourFormatsMap = nullptr;
459 
460 // Value deleter for hashmap.
deleteAllowedHourFormats(void * ptr)461 U_CFUNC void U_CALLCONV deleteAllowedHourFormats(void *ptr) {
462     uprv_free(ptr);
463 }
464 
465 // Close hashmap at cleanup.
allowedHourFormatsCleanup()466 U_CFUNC UBool U_CALLCONV allowedHourFormatsCleanup() {
467     uhash_close(localeToAllowedHourFormatsMap);
468     return TRUE;
469 }
470 
471 enum AllowedHourFormat{
472     ALLOWED_HOUR_FORMAT_UNKNOWN = -1,
473     ALLOWED_HOUR_FORMAT_h,
474     ALLOWED_HOUR_FORMAT_H,
475     ALLOWED_HOUR_FORMAT_K,  // Added ICU-20383, used by JP
476     ALLOWED_HOUR_FORMAT_k,  // Added ICU-20383, not currently used
477     ALLOWED_HOUR_FORMAT_hb,
478     ALLOWED_HOUR_FORMAT_hB,
479     ALLOWED_HOUR_FORMAT_Kb, // Added ICU-20383, not currently used
480     ALLOWED_HOUR_FORMAT_KB, // Added ICU-20383, not currently used
481     // ICU-20383 The following are unlikely and not currently used
482     ALLOWED_HOUR_FORMAT_Hb,
483     ALLOWED_HOUR_FORMAT_HB
484 };
485 
486 }  // namespace
487 
488 void
initData(const Locale & locale,UErrorCode & status)489 DateTimePatternGenerator::initData(const Locale& locale, UErrorCode &status) {
490     //const char *baseLangName = locale.getBaseName(); // unused
491 
492     skipMatcher = nullptr;
493     fAvailableFormatKeyHash=nullptr;
494     addCanonicalItems(status);
495     addICUPatterns(locale, status);
496     addCLDRData(locale, status);
497     setDateTimeFromCalendar(locale, status);
498     setDecimalSymbols(locale, status);
499     umtx_initOnce(initOnce, loadAllowedHourFormatsData, status);
500     getAllowedHourFormats(locale, status);
501     // If any of the above methods failed then the object is in an invalid state.
502     internalErrorCode = status;
503 } // DateTimePatternGenerator::initData
504 
505 namespace {
506 
507 struct AllowedHourFormatsSink : public ResourceSink {
508     // Initialize sub-sinks.
AllowedHourFormatsSink__anone734f86d0211::AllowedHourFormatsSink509     AllowedHourFormatsSink() {}
510     virtual ~AllowedHourFormatsSink();
511 
put__anone734f86d0211::AllowedHourFormatsSink512     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
513                      UErrorCode &errorCode) {
514         ResourceTable timeData = value.getTable(errorCode);
515         if (U_FAILURE(errorCode)) { return; }
516         for (int32_t i = 0; timeData.getKeyAndValue(i, key, value); ++i) {
517             const char *regionOrLocale = key;
518             ResourceTable formatList = value.getTable(errorCode);
519             if (U_FAILURE(errorCode)) { return; }
520             // below we construct a list[] that has an entry for the "preferred" value at [0],
521             // followed by 1 or more entries for the "allowed" values, terminated with an
522             // entry for ALLOWED_HOUR_FORMAT_UNKNOWN (not included in length below)
523             LocalMemory<int32_t> list;
524             int32_t length = 0;
525             int32_t preferredFormat = ALLOWED_HOUR_FORMAT_UNKNOWN;
526             for (int32_t j = 0; formatList.getKeyAndValue(j, key, value); ++j) {
527                 if (uprv_strcmp(key, "allowed") == 0) {
528                     if (value.getType() == URES_STRING) {
529                         length = 2; // 1 preferred to add later, 1 allowed to add now
530                         if (list.allocateInsteadAndReset(length + 1) == nullptr) {
531                             errorCode = U_MEMORY_ALLOCATION_ERROR;
532                             return;
533                         }
534                         list[1] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
535                     }
536                     else {
537                         ResourceArray allowedFormats = value.getArray(errorCode);
538                         length = allowedFormats.getSize() + 1; // 1 preferred, getSize allowed
539                         if (list.allocateInsteadAndReset(length + 1) == nullptr) {
540                             errorCode = U_MEMORY_ALLOCATION_ERROR;
541                             return;
542                         }
543                         for (int32_t k = 1; k < length; ++k) {
544                             allowedFormats.getValue(k-1, value);
545                             list[k] = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
546                         }
547                     }
548                 } else if (uprv_strcmp(key, "preferred") == 0) {
549                     preferredFormat = getHourFormatFromUnicodeString(value.getUnicodeString(errorCode));
550                 }
551             }
552             if (length > 1) {
553                 list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: list[1];
554             } else {
555                 // fallback handling for missing data
556                 length = 2; // 1 preferred, 1 allowed
557                 if (list.allocateInsteadAndReset(length + 1) == nullptr) {
558                     errorCode = U_MEMORY_ALLOCATION_ERROR;
559                     return;
560                 }
561                 list[0] = (preferredFormat!=ALLOWED_HOUR_FORMAT_UNKNOWN)? preferredFormat: ALLOWED_HOUR_FORMAT_H;
562                 list[1] = list[0];
563             }
564             list[length] = ALLOWED_HOUR_FORMAT_UNKNOWN;
565             // At this point list[] will have at least two non-ALLOWED_HOUR_FORMAT_UNKNOWN entries,
566             // followed by ALLOWED_HOUR_FORMAT_UNKNOWN.
567             uhash_put(localeToAllowedHourFormatsMap, const_cast<char *>(regionOrLocale), list.orphan(), &errorCode);
568             if (U_FAILURE(errorCode)) { return; }
569         }
570     }
571 
getHourFormatFromUnicodeString__anone734f86d0211::AllowedHourFormatsSink572     AllowedHourFormat getHourFormatFromUnicodeString(const UnicodeString &s) {
573         if (s.length() == 1) {
574             if (s[0] == LOW_H) { return ALLOWED_HOUR_FORMAT_h; }
575             if (s[0] == CAP_H) { return ALLOWED_HOUR_FORMAT_H; }
576             if (s[0] == CAP_K) { return ALLOWED_HOUR_FORMAT_K; }
577             if (s[0] == LOW_K) { return ALLOWED_HOUR_FORMAT_k; }
578         } else if (s.length() == 2) {
579             if (s[0] == LOW_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_hb; }
580             if (s[0] == LOW_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_hB; }
581             if (s[0] == CAP_K && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Kb; }
582             if (s[0] == CAP_K && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_KB; }
583             if (s[0] == CAP_H && s[1] == LOW_B) { return ALLOWED_HOUR_FORMAT_Hb; }
584             if (s[0] == CAP_H && s[1] == CAP_B) { return ALLOWED_HOUR_FORMAT_HB; }
585         }
586 
587         return ALLOWED_HOUR_FORMAT_UNKNOWN;
588     }
589 };
590 
591 }  // namespace
592 
~AllowedHourFormatsSink()593 AllowedHourFormatsSink::~AllowedHourFormatsSink() {}
594 
loadAllowedHourFormatsData(UErrorCode & status)595 U_CFUNC void U_CALLCONV DateTimePatternGenerator::loadAllowedHourFormatsData(UErrorCode &status) {
596     if (U_FAILURE(status)) { return; }
597     localeToAllowedHourFormatsMap = uhash_open(
598         uhash_hashChars, uhash_compareChars, nullptr, &status);
599     if (U_FAILURE(status)) { return; }
600 
601     uhash_setValueDeleter(localeToAllowedHourFormatsMap, deleteAllowedHourFormats);
602     ucln_i18n_registerCleanup(UCLN_I18N_ALLOWED_HOUR_FORMATS, allowedHourFormatsCleanup);
603 
604     LocalUResourceBundlePointer rb(ures_openDirect(nullptr, "supplementalData", &status));
605     if (U_FAILURE(status)) { return; }
606 
607     AllowedHourFormatsSink sink;
608     // TODO: Currently in the enumeration each table allocates a new array.
609     // Try to reduce the number of memory allocations. Consider storing a
610     // UVector32 with the concatenation of all of the sub-arrays, put the start index
611     // into the hashmap, store 6 single-value sub-arrays right at the beginning of the
612     // vector (at index enum*2) for easy data sharing, copy sub-arrays into runtime
613     // object. Remember to clean up the vector, too.
614     ures_getAllItemsWithFallback(rb.getAlias(), "timeData", sink, status);
615 }
616 
getAllowedHourFormatsLangCountry(const char * language,const char * country,UErrorCode & status)617 static int32_t* getAllowedHourFormatsLangCountry(const char* language, const char* country, UErrorCode& status) {
618     CharString langCountry;
619     langCountry.append(language, status);
620     langCountry.append('_', status);
621     langCountry.append(country, status);
622 
623     int32_t* allowedFormats;
624     allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, langCountry.data());
625     if (allowedFormats == nullptr) {
626         allowedFormats = (int32_t *)uhash_get(localeToAllowedHourFormatsMap, const_cast<char *>(country));
627     }
628 
629     return allowedFormats;
630 }
631 
getAllowedHourFormats(const Locale & locale,UErrorCode & status)632 void DateTimePatternGenerator::getAllowedHourFormats(const Locale &locale, UErrorCode &status) {
633     if (U_FAILURE(status)) { return; }
634 
635     const char *language = locale.getLanguage();
636     const char *country = locale.getCountry();
637     Locale maxLocale;  // must be here for correct lifetime
638     if (*language == '\0' || *country == '\0') {
639         maxLocale = locale;
640         UErrorCode localStatus = U_ZERO_ERROR;
641         maxLocale.addLikelySubtags(localStatus);
642         if (U_SUCCESS(localStatus)) {
643             language = maxLocale.getLanguage();
644             country = maxLocale.getCountry();
645         }
646     }
647     if (*language == '\0') {
648         // Unexpected, but fail gracefully
649         language = "und";
650     }
651     if (*country == '\0') {
652         country = "001";
653     }
654 
655     int32_t* allowedFormats = getAllowedHourFormatsLangCountry(language, country, status);
656 
657     // Check if the region has an alias
658     if (allowedFormats == nullptr) {
659         UErrorCode localStatus = U_ZERO_ERROR;
660         const Region* region = Region::getInstance(country, localStatus);
661         if (U_SUCCESS(localStatus)) {
662             country = region->getRegionCode(); // the real region code
663             allowedFormats = getAllowedHourFormatsLangCountry(language, country, status);
664         }
665     }
666 
667     if (allowedFormats != nullptr) {  // Lookup is successful
668         // Here allowedFormats points to a list consisting of key for preferredFormat,
669         // followed by one or more keys for allowedFormats, then followed by ALLOWED_HOUR_FORMAT_UNKNOWN.
670         switch (allowedFormats[0]) {
671             case ALLOWED_HOUR_FORMAT_h: fDefaultHourFormatChar = LOW_H; break;
672             case ALLOWED_HOUR_FORMAT_H: fDefaultHourFormatChar = CAP_H; break;
673             case ALLOWED_HOUR_FORMAT_K: fDefaultHourFormatChar = CAP_K; break;
674             case ALLOWED_HOUR_FORMAT_k: fDefaultHourFormatChar = LOW_K; break;
675             default: fDefaultHourFormatChar = CAP_H; break;
676         }
677         for (int32_t i = 0; i < UPRV_LENGTHOF(fAllowedHourFormats); ++i) {
678             fAllowedHourFormats[i] = allowedFormats[i + 1];
679             if (fAllowedHourFormats[i] == ALLOWED_HOUR_FORMAT_UNKNOWN) {
680                 break;
681             }
682         }
683     } else {  // Lookup failed, twice
684         fDefaultHourFormatChar = CAP_H;
685         fAllowedHourFormats[0] = ALLOWED_HOUR_FORMAT_H;
686         fAllowedHourFormats[1] = ALLOWED_HOUR_FORMAT_UNKNOWN;
687     }
688 }
689 
690 UnicodeString
getSkeleton(const UnicodeString & pattern,UErrorCode &)691 DateTimePatternGenerator::getSkeleton(const UnicodeString& pattern, UErrorCode&
692 /*status*/) {
693     FormatParser fp2;
694     DateTimeMatcher matcher;
695     PtnSkeleton localSkeleton;
696     matcher.set(pattern, &fp2, localSkeleton);
697     return localSkeleton.getSkeleton();
698 }
699 
700 UnicodeString
staticGetSkeleton(const UnicodeString & pattern,UErrorCode &)701 DateTimePatternGenerator::staticGetSkeleton(
702         const UnicodeString& pattern, UErrorCode& /*status*/) {
703     FormatParser fp;
704     DateTimeMatcher matcher;
705     PtnSkeleton localSkeleton;
706     matcher.set(pattern, &fp, localSkeleton);
707     return localSkeleton.getSkeleton();
708 }
709 
710 UnicodeString
getBaseSkeleton(const UnicodeString & pattern,UErrorCode &)711 DateTimePatternGenerator::getBaseSkeleton(const UnicodeString& pattern, UErrorCode& /*status*/) {
712     FormatParser fp2;
713     DateTimeMatcher matcher;
714     PtnSkeleton localSkeleton;
715     matcher.set(pattern, &fp2, localSkeleton);
716     return localSkeleton.getBaseSkeleton();
717 }
718 
719 UnicodeString
staticGetBaseSkeleton(const UnicodeString & pattern,UErrorCode &)720 DateTimePatternGenerator::staticGetBaseSkeleton(
721         const UnicodeString& pattern, UErrorCode& /*status*/) {
722     FormatParser fp;
723     DateTimeMatcher matcher;
724     PtnSkeleton localSkeleton;
725     matcher.set(pattern, &fp, localSkeleton);
726     return localSkeleton.getBaseSkeleton();
727 }
728 
729 void
addICUPatterns(const Locale & locale,UErrorCode & status)730 DateTimePatternGenerator::addICUPatterns(const Locale& locale, UErrorCode& status) {
731     if (U_FAILURE(status)) { return; }
732     UnicodeString dfPattern;
733     UnicodeString conflictingString;
734     DateFormat* df;
735 
736     // Load with ICU patterns
737     for (int32_t i=DateFormat::kFull; i<=DateFormat::kShort; i++) {
738         DateFormat::EStyle style = (DateFormat::EStyle)i;
739         df = DateFormat::createDateInstance(style, locale);
740         SimpleDateFormat* sdf;
741         if (df != nullptr && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != nullptr) {
742             sdf->toPattern(dfPattern);
743             addPattern(dfPattern, FALSE, conflictingString, status);
744         }
745         // TODO Maybe we should return an error when the date format isn't simple.
746         delete df;
747         if (U_FAILURE(status)) { return; }
748 
749         df = DateFormat::createTimeInstance(style, locale);
750         if (df != nullptr && (sdf = dynamic_cast<SimpleDateFormat*>(df)) != nullptr) {
751             sdf->toPattern(dfPattern);
752             addPattern(dfPattern, FALSE, conflictingString, status);
753 
754             // TODO: C++ and Java are inconsistent (see #12568).
755             // C++ uses MEDIUM, but Java uses SHORT.
756             if ( i==DateFormat::kShort && !dfPattern.isEmpty() ) {
757                 consumeShortTimePattern(dfPattern, status);
758             }
759         }
760         // TODO Maybe we should return an error when the date format isn't simple.
761         delete df;
762         if (U_FAILURE(status)) { return; }
763     }
764 }
765 
766 void
hackTimes(const UnicodeString & hackPattern,UErrorCode & status)767 DateTimePatternGenerator::hackTimes(const UnicodeString& hackPattern, UErrorCode& status)  {
768     UnicodeString conflictingString;
769 
770     fp->set(hackPattern);
771     UnicodeString mmss;
772     UBool gotMm=FALSE;
773     for (int32_t i=0; i<fp->itemNumber; ++i) {
774         UnicodeString field = fp->items[i];
775         if ( fp->isQuoteLiteral(field) ) {
776             if ( gotMm ) {
777                UnicodeString quoteLiteral;
778                fp->getQuoteLiteral(quoteLiteral, &i);
779                mmss += quoteLiteral;
780             }
781         }
782         else {
783             if (fp->isPatternSeparator(field) && gotMm) {
784                 mmss+=field;
785             }
786             else {
787                 UChar ch=field.charAt(0);
788                 if (ch==LOW_M) {
789                     gotMm=TRUE;
790                     mmss+=field;
791                 }
792                 else {
793                     if (ch==LOW_S) {
794                         if (!gotMm) {
795                             break;
796                         }
797                         mmss+= field;
798                         addPattern(mmss, FALSE, conflictingString, status);
799                         break;
800                     }
801                     else {
802                         if (gotMm || ch==LOW_Z || ch==CAP_Z || ch==LOW_V || ch==CAP_V) {
803                             break;
804                         }
805                     }
806                 }
807             }
808         }
809     }
810 }
811 
812 #define ULOC_LOCALE_IDENTIFIER_CAPACITY (ULOC_FULLNAME_CAPACITY + 1 + ULOC_KEYWORD_AND_VALUES_CAPACITY)
813 
814 void
getCalendarTypeToUse(const Locale & locale,CharString & destination,UErrorCode & err)815 DateTimePatternGenerator::getCalendarTypeToUse(const Locale& locale, CharString& destination, UErrorCode& err) {
816     destination.clear().append(DT_DateTimeGregorianTag, -1, err); // initial default
817     if ( U_SUCCESS(err) ) {
818         UErrorCode localStatus = U_ZERO_ERROR;
819         char localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY];
820         // obtain a locale that always has the calendar key value that should be used
821         ures_getFunctionalEquivalent(
822             localeWithCalendarKey,
823             ULOC_LOCALE_IDENTIFIER_CAPACITY,
824             nullptr,
825             "calendar",
826             "calendar",
827             locale.getName(),
828             nullptr,
829             FALSE,
830             &localStatus);
831         localeWithCalendarKey[ULOC_LOCALE_IDENTIFIER_CAPACITY-1] = 0; // ensure null termination
832         // now get the calendar key value from that locale
833         char calendarType[ULOC_KEYWORDS_CAPACITY];
834         int32_t calendarTypeLen = uloc_getKeywordValue(
835             localeWithCalendarKey,
836             "calendar",
837             calendarType,
838             ULOC_KEYWORDS_CAPACITY,
839             &localStatus);
840         // If the input locale was invalid, don't fail with missing resource error, instead
841         // continue with default of Gregorian.
842         if (U_FAILURE(localStatus) && localStatus != U_MISSING_RESOURCE_ERROR) {
843             err = localStatus;
844             return;
845         }
846         if (calendarTypeLen < ULOC_KEYWORDS_CAPACITY) {
847             destination.clear().append(calendarType, -1, err);
848             if (U_FAILURE(err)) { return; }
849         }
850     }
851 }
852 
853 void
consumeShortTimePattern(const UnicodeString & shortTimePattern,UErrorCode & status)854 DateTimePatternGenerator::consumeShortTimePattern(const UnicodeString& shortTimePattern,
855         UErrorCode& status) {
856     if (U_FAILURE(status)) { return; }
857     // ICU-20383 No longer set fDefaultHourFormatChar to the hour format character from
858     // this pattern; instead it is set from localeToAllowedHourFormatsMap which now
859     // includes entries for both preferred and allowed formats.
860 
861     // HACK for hh:ss
862     hackTimes(shortTimePattern, status);
863 }
864 
865 struct DateTimePatternGenerator::AppendItemFormatsSink : public ResourceSink {
866 
867     // Destination for data, modified via setters.
868     DateTimePatternGenerator& dtpg;
869 
AppendItemFormatsSinkDateTimePatternGenerator::AppendItemFormatsSink870     AppendItemFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
871     virtual ~AppendItemFormatsSink();
872 
putDateTimePatternGenerator::AppendItemFormatsSink873     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
874             UErrorCode &errorCode) {
875         ResourceTable itemsTable = value.getTable(errorCode);
876         if (U_FAILURE(errorCode)) { return; }
877         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
878             UDateTimePatternField field = dtpg.getAppendFormatNumber(key);
879             if (field == UDATPG_FIELD_COUNT) { continue; }
880             const UnicodeString& valueStr = value.getUnicodeString(errorCode);
881             if (dtpg.getAppendItemFormat(field).isEmpty() && !valueStr.isEmpty()) {
882                 dtpg.setAppendItemFormat(field, valueStr);
883             }
884         }
885     }
886 
fillInMissingDateTimePatternGenerator::AppendItemFormatsSink887     void fillInMissing() {
888         UnicodeString defaultItemFormat(TRUE, UDATPG_ItemFormat, UPRV_LENGTHOF(UDATPG_ItemFormat)-1);  // Read-only alias.
889         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
890             UDateTimePatternField field = (UDateTimePatternField)i;
891             if (dtpg.getAppendItemFormat(field).isEmpty()) {
892                 dtpg.setAppendItemFormat(field, defaultItemFormat);
893             }
894         }
895     }
896 };
897 
898 struct DateTimePatternGenerator::AppendItemNamesSink : public ResourceSink {
899 
900     // Destination for data, modified via setters.
901     DateTimePatternGenerator& dtpg;
902 
AppendItemNamesSinkDateTimePatternGenerator::AppendItemNamesSink903     AppendItemNamesSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
904     virtual ~AppendItemNamesSink();
905 
putDateTimePatternGenerator::AppendItemNamesSink906     virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
907             UErrorCode &errorCode) {
908         ResourceTable itemsTable = value.getTable(errorCode);
909         if (U_FAILURE(errorCode)) { return; }
910         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
911             UDateTimePGDisplayWidth width;
912             UDateTimePatternField field = dtpg.getFieldAndWidthIndices(key, &width);
913             if (field == UDATPG_FIELD_COUNT) { continue; }
914             ResourceTable detailsTable = value.getTable(errorCode);
915             if (U_FAILURE(errorCode)) { return; }
916             for (int32_t j = 0; detailsTable.getKeyAndValue(j, key, value); ++j) {
917                 if (uprv_strcmp(key, "dn") != 0) { continue; }
918                 const UnicodeString& valueStr = value.getUnicodeString(errorCode);
919                 if (dtpg.getFieldDisplayName(field,width).isEmpty() && !valueStr.isEmpty()) {
920                     dtpg.setFieldDisplayName(field,width,valueStr);
921                 }
922                 break;
923             }
924         }
925     }
926 
fillInMissingDateTimePatternGenerator::AppendItemNamesSink927     void fillInMissing() {
928         for (int32_t i = 0; i < UDATPG_FIELD_COUNT; i++) {
929             UnicodeString& valueStr = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, UDATPG_WIDE);
930             if (valueStr.isEmpty()) {
931                 valueStr = CAP_F;
932                 U_ASSERT(i < 20);
933                 if (i < 10) {
934                     // F0, F1, ..., F9
935                     valueStr += (UChar)(i+0x30);
936                 } else {
937                     // F10, F11, ...
938                     valueStr += (UChar)0x31;
939                     valueStr += (UChar)(i-10 + 0x30);
940                 }
941                 // NUL-terminate for the C API.
942                 valueStr.getTerminatedBuffer();
943             }
944             for (int32_t j = 1; j < UDATPG_WIDTH_COUNT; j++) {
945                 UnicodeString& valueStr2 = dtpg.getMutableFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)j);
946                 if (valueStr2.isEmpty()) {
947                     valueStr2 = dtpg.getFieldDisplayName((UDateTimePatternField)i, (UDateTimePGDisplayWidth)(j-1));
948                 }
949             }
950         }
951     }
952 };
953 
954 struct DateTimePatternGenerator::AvailableFormatsSink : public ResourceSink {
955 
956     // Destination for data, modified via setters.
957     DateTimePatternGenerator& dtpg;
958 
959     // Temporary variable, required for calling addPatternWithSkeleton.
960     UnicodeString conflictingPattern;
961 
AvailableFormatsSinkDateTimePatternGenerator::AvailableFormatsSink962     AvailableFormatsSink(DateTimePatternGenerator& _dtpg) : dtpg(_dtpg) {}
963     virtual ~AvailableFormatsSink();
964 
putDateTimePatternGenerator::AvailableFormatsSink965     virtual void put(const char *key, ResourceValue &value, UBool isRoot,
966             UErrorCode &errorCode) {
967         ResourceTable itemsTable = value.getTable(errorCode);
968         if (U_FAILURE(errorCode)) { return; }
969         for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
970             const UnicodeString formatKey(key, -1, US_INV);
971             if (!dtpg.isAvailableFormatSet(formatKey) ) {
972                 dtpg.setAvailableFormat(formatKey, errorCode);
973                 // Add pattern with its associated skeleton. Override any duplicate
974                 // derived from std patterns, but not a previous availableFormats entry:
975                 const UnicodeString& formatValue = value.getUnicodeString(errorCode);
976                 conflictingPattern.remove();
977                 dtpg.addPatternWithSkeleton(formatValue, &formatKey, !isRoot, conflictingPattern, errorCode);
978             }
979         }
980     }
981 };
982 
983 // Virtual destructors must be defined out of line.
~AppendItemFormatsSink()984 DateTimePatternGenerator::AppendItemFormatsSink::~AppendItemFormatsSink() {}
~AppendItemNamesSink()985 DateTimePatternGenerator::AppendItemNamesSink::~AppendItemNamesSink() {}
~AvailableFormatsSink()986 DateTimePatternGenerator::AvailableFormatsSink::~AvailableFormatsSink() {}
987 
988 void
addCLDRData(const Locale & locale,UErrorCode & errorCode)989 DateTimePatternGenerator::addCLDRData(const Locale& locale, UErrorCode& errorCode) {
990     if (U_FAILURE(errorCode)) { return; }
991     UnicodeString rbPattern, value, field;
992     CharString path;
993 
994     LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &errorCode));
995     if (U_FAILURE(errorCode)) { return; }
996 
997     CharString calendarTypeToUse; // to be filled in with the type to use, if all goes well
998     getCalendarTypeToUse(locale, calendarTypeToUse, errorCode);
999     if (U_FAILURE(errorCode)) { return; }
1000 
1001     // Local err to ignore resource not found exceptions
1002     UErrorCode err = U_ZERO_ERROR;
1003 
1004     // Load append item formats.
1005     AppendItemFormatsSink appendItemFormatsSink(*this);
1006     path.clear()
1007         .append(DT_DateTimeCalendarTag, errorCode)
1008         .append('/', errorCode)
1009         .append(calendarTypeToUse, errorCode)
1010         .append('/', errorCode)
1011         .append(DT_DateTimeAppendItemsTag, errorCode); // i.e., calendar/xxx/appendItems
1012     if (U_FAILURE(errorCode)) { return; }
1013     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), appendItemFormatsSink, err);
1014     appendItemFormatsSink.fillInMissing();
1015 
1016     // Load CLDR item names.
1017     err = U_ZERO_ERROR;
1018     AppendItemNamesSink appendItemNamesSink(*this);
1019     ures_getAllItemsWithFallback(rb.getAlias(), DT_DateTimeFieldsTag, appendItemNamesSink, err);
1020     appendItemNamesSink.fillInMissing();
1021 
1022     // Load the available formats from CLDR.
1023     err = U_ZERO_ERROR;
1024     initHashtable(errorCode);
1025     if (U_FAILURE(errorCode)) { return; }
1026     AvailableFormatsSink availableFormatsSink(*this);
1027     path.clear()
1028         .append(DT_DateTimeCalendarTag, errorCode)
1029         .append('/', errorCode)
1030         .append(calendarTypeToUse, errorCode)
1031         .append('/', errorCode)
1032         .append(DT_DateTimeAvailableFormatsTag, errorCode); // i.e., calendar/xxx/availableFormats
1033     if (U_FAILURE(errorCode)) { return; }
1034     ures_getAllItemsWithFallback(rb.getAlias(), path.data(), availableFormatsSink, err);
1035 }
1036 
1037 void
initHashtable(UErrorCode & err)1038 DateTimePatternGenerator::initHashtable(UErrorCode& err) {
1039     if (U_FAILURE(err)) { return; }
1040     if (fAvailableFormatKeyHash!=nullptr) {
1041         return;
1042     }
1043     LocalPointer<Hashtable> hash(new Hashtable(FALSE, err), err);
1044     if (U_SUCCESS(err)) {
1045         fAvailableFormatKeyHash = hash.orphan();
1046     }
1047 }
1048 
1049 void
setAppendItemFormat(UDateTimePatternField field,const UnicodeString & value)1050 DateTimePatternGenerator::setAppendItemFormat(UDateTimePatternField field, const UnicodeString& value) {
1051     appendItemFormats[field] = value;
1052     // NUL-terminate for the C API.
1053     appendItemFormats[field].getTerminatedBuffer();
1054 }
1055 
1056 const UnicodeString&
getAppendItemFormat(UDateTimePatternField field) const1057 DateTimePatternGenerator::getAppendItemFormat(UDateTimePatternField field) const {
1058     return appendItemFormats[field];
1059 }
1060 
1061 void
setAppendItemName(UDateTimePatternField field,const UnicodeString & value)1062 DateTimePatternGenerator::setAppendItemName(UDateTimePatternField field, const UnicodeString& value) {
1063     setFieldDisplayName(field, UDATPG_WIDTH_APPENDITEM, value);
1064 }
1065 
1066 const UnicodeString&
getAppendItemName(UDateTimePatternField field) const1067 DateTimePatternGenerator::getAppendItemName(UDateTimePatternField field) const {
1068     return fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1069 }
1070 
1071 void
setFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width,const UnicodeString & value)1072 DateTimePatternGenerator::setFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width, const UnicodeString& value) {
1073     fieldDisplayNames[field][width] = value;
1074     // NUL-terminate for the C API.
1075     fieldDisplayNames[field][width].getTerminatedBuffer();
1076 }
1077 
1078 UnicodeString
getFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width) const1079 DateTimePatternGenerator::getFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) const {
1080     return fieldDisplayNames[field][width];
1081 }
1082 
1083 UnicodeString&
getMutableFieldDisplayName(UDateTimePatternField field,UDateTimePGDisplayWidth width)1084 DateTimePatternGenerator::getMutableFieldDisplayName(UDateTimePatternField field, UDateTimePGDisplayWidth width) {
1085     return fieldDisplayNames[field][width];
1086 }
1087 
1088 void
getAppendName(UDateTimePatternField field,UnicodeString & value)1089 DateTimePatternGenerator::getAppendName(UDateTimePatternField field, UnicodeString& value) {
1090     value = SINGLE_QUOTE;
1091     value += fieldDisplayNames[field][UDATPG_WIDTH_APPENDITEM];
1092     value += SINGLE_QUOTE;
1093 }
1094 
1095 UnicodeString
getBestPattern(const UnicodeString & patternForm,UErrorCode & status)1096 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UErrorCode& status) {
1097     return getBestPattern(patternForm, UDATPG_MATCH_NO_OPTIONS, status);
1098 }
1099 
1100 UnicodeString
getBestPattern(const UnicodeString & patternForm,UDateTimePatternMatchOptions options,UErrorCode & status)1101 DateTimePatternGenerator::getBestPattern(const UnicodeString& patternForm, UDateTimePatternMatchOptions options, UErrorCode& status) {
1102     if (U_FAILURE(status)) {
1103         return UnicodeString();
1104     }
1105     if (U_FAILURE(internalErrorCode)) {
1106         status = internalErrorCode;
1107         return UnicodeString();
1108     }
1109     const UnicodeString *bestPattern = nullptr;
1110     UnicodeString dtFormat;
1111     UnicodeString resultPattern;
1112     int32_t flags = kDTPGNoFlags;
1113 
1114     int32_t dateMask=(1<<UDATPG_DAYPERIOD_FIELD) - 1;
1115     int32_t timeMask=(1<<UDATPG_FIELD_COUNT) - 1 - dateMask;
1116 
1117     // Replace hour metacharacters 'j', 'C' and 'J', set flags as necessary
1118     UnicodeString patternFormMapped = mapSkeletonMetacharacters(patternForm, &flags, status);
1119     if (U_FAILURE(status)) {
1120         return UnicodeString();
1121     }
1122 
1123     resultPattern.remove();
1124     dtMatcher->set(patternFormMapped, fp);
1125     const PtnSkeleton* specifiedSkeleton = nullptr;
1126     bestPattern=getBestRaw(*dtMatcher, -1, distanceInfo, status, &specifiedSkeleton);
1127     if (U_FAILURE(status)) {
1128         return UnicodeString();
1129     }
1130 
1131     if ( distanceInfo->missingFieldMask==0 && distanceInfo->extraFieldMask==0 ) {
1132         resultPattern = adjustFieldTypes(*bestPattern, specifiedSkeleton, flags, options);
1133 
1134         return resultPattern;
1135     }
1136     int32_t neededFields = dtMatcher->getFieldMask();
1137     UnicodeString datePattern=getBestAppending(neededFields & dateMask, flags, status, options);
1138     UnicodeString timePattern=getBestAppending(neededFields & timeMask, flags, status, options);
1139     if (U_FAILURE(status)) {
1140         return UnicodeString();
1141     }
1142     if (datePattern.length()==0) {
1143         if (timePattern.length()==0) {
1144             resultPattern.remove();
1145         }
1146         else {
1147             return timePattern;
1148         }
1149     }
1150     if (timePattern.length()==0) {
1151         return datePattern;
1152     }
1153     resultPattern.remove();
1154     status = U_ZERO_ERROR;
1155     dtFormat=getDateTimeFormat();
1156     SimpleFormatter(dtFormat, 2, 2, status).format(timePattern, datePattern, resultPattern, status);
1157     return resultPattern;
1158 }
1159 
1160 /*
1161  * Map a skeleton that may have metacharacters jJC to one without, by replacing
1162  * the metacharacters with locale-appropriate fields of h/H/k/K and of a/b/B
1163  * (depends on fDefaultHourFormatChar and fAllowedHourFormats being set, which in
1164  * turn depends on initData having been run). This method also updates the flags
1165  * as necessary. Returns the updated skeleton.
1166  */
1167 UnicodeString
mapSkeletonMetacharacters(const UnicodeString & patternForm,int32_t * flags,UErrorCode & status)1168 DateTimePatternGenerator::mapSkeletonMetacharacters(const UnicodeString& patternForm, int32_t* flags, UErrorCode& status) {
1169     UnicodeString patternFormMapped;
1170     patternFormMapped.remove();
1171     UBool inQuoted = FALSE;
1172     int32_t patPos, patLen = patternForm.length();
1173     for (patPos = 0; patPos < patLen; patPos++) {
1174         UChar patChr = patternForm.charAt(patPos);
1175         if (patChr == SINGLE_QUOTE) {
1176             inQuoted = !inQuoted;
1177         } else if (!inQuoted) {
1178             // Handle special mappings for 'j' and 'C' in which fields lengths
1179             // 1,3,5 => hour field length 1
1180             // 2,4,6 => hour field length 2
1181             // 1,2 => abbreviated dayPeriod (field length 1..3)
1182             // 3,4 => long dayPeriod (field length 4)
1183             // 5,6 => narrow dayPeriod (field length 5)
1184             if (patChr == LOW_J || patChr == CAP_C) {
1185                 int32_t extraLen = 0; // 1 less than total field length
1186                 while (patPos+1 < patLen && patternForm.charAt(patPos+1)==patChr) {
1187                     extraLen++;
1188                     patPos++;
1189                 }
1190                 int32_t hourLen = 1 + (extraLen & 1);
1191                 int32_t dayPeriodLen = (extraLen < 2)? 1: 3 + (extraLen >> 1);
1192                 UChar hourChar = LOW_H;
1193                 UChar dayPeriodChar = LOW_A;
1194                 if (patChr == LOW_J) {
1195                     hourChar = fDefaultHourFormatChar;
1196                 } else {
1197                     AllowedHourFormat bestAllowed;
1198                     if (fAllowedHourFormats[0] != ALLOWED_HOUR_FORMAT_UNKNOWN) {
1199                         bestAllowed = (AllowedHourFormat)fAllowedHourFormats[0];
1200                     } else {
1201                         status = U_INVALID_FORMAT_ERROR;
1202                         return UnicodeString();
1203                     }
1204                     if (bestAllowed == ALLOWED_HOUR_FORMAT_H || bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_Hb) {
1205                         hourChar = CAP_H;
1206                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_K || bestAllowed == ALLOWED_HOUR_FORMAT_KB || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) {
1207                         hourChar = CAP_K;
1208                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_k) {
1209                         hourChar = LOW_K;
1210                     }
1211                     // in #13183 just add b/B to skeleton, no longer need to set special flags
1212                     if (bestAllowed == ALLOWED_HOUR_FORMAT_HB || bestAllowed == ALLOWED_HOUR_FORMAT_hB || bestAllowed == ALLOWED_HOUR_FORMAT_KB) {
1213                         dayPeriodChar = CAP_B;
1214                     } else if (bestAllowed == ALLOWED_HOUR_FORMAT_Hb || bestAllowed == ALLOWED_HOUR_FORMAT_hb || bestAllowed == ALLOWED_HOUR_FORMAT_Kb) {
1215                         dayPeriodChar = LOW_B;
1216                     }
1217                 }
1218                 if (hourChar==CAP_H || hourChar==LOW_K) {
1219                     dayPeriodLen = 0;
1220                 }
1221                 while (dayPeriodLen-- > 0) {
1222                     patternFormMapped.append(dayPeriodChar);
1223                 }
1224                 while (hourLen-- > 0) {
1225                     patternFormMapped.append(hourChar);
1226                 }
1227             } else if (patChr == CAP_J) {
1228                 // Get pattern for skeleton with H, then replace H or k
1229                 // with fDefaultHourFormatChar (if different)
1230                 patternFormMapped.append(CAP_H);
1231                 *flags |= kDTPGSkeletonUsesCapJ;
1232             } else {
1233                 patternFormMapped.append(patChr);
1234             }
1235         }
1236     }
1237     return patternFormMapped;
1238 }
1239 
1240 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UErrorCode & status)1241 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1242                                             const UnicodeString& skeleton,
1243                                             UErrorCode& status) {
1244     return replaceFieldTypes(pattern, skeleton, UDATPG_MATCH_NO_OPTIONS, status);
1245 }
1246 
1247 UnicodeString
replaceFieldTypes(const UnicodeString & pattern,const UnicodeString & skeleton,UDateTimePatternMatchOptions options,UErrorCode & status)1248 DateTimePatternGenerator::replaceFieldTypes(const UnicodeString& pattern,
1249                                             const UnicodeString& skeleton,
1250                                             UDateTimePatternMatchOptions options,
1251                                             UErrorCode& status) {
1252     if (U_FAILURE(status)) {
1253         return UnicodeString();
1254     }
1255     if (U_FAILURE(internalErrorCode)) {
1256         status = internalErrorCode;
1257         return UnicodeString();
1258     }
1259     dtMatcher->set(skeleton, fp);
1260     UnicodeString result = adjustFieldTypes(pattern, nullptr, kDTPGNoFlags, options);
1261     return result;
1262 }
1263 
1264 void
setDecimal(const UnicodeString & newDecimal)1265 DateTimePatternGenerator::setDecimal(const UnicodeString& newDecimal) {
1266     this->decimal = newDecimal;
1267     // NUL-terminate for the C API.
1268     this->decimal.getTerminatedBuffer();
1269 }
1270 
1271 const UnicodeString&
getDecimal() const1272 DateTimePatternGenerator::getDecimal() const {
1273     return decimal;
1274 }
1275 
1276 void
addCanonicalItems(UErrorCode & status)1277 DateTimePatternGenerator::addCanonicalItems(UErrorCode& status) {
1278     if (U_FAILURE(status)) { return; }
1279     UnicodeString  conflictingPattern;
1280 
1281     for (int32_t i=0; i<UDATPG_FIELD_COUNT; i++) {
1282         if (Canonical_Items[i] > 0) {
1283             addPattern(UnicodeString(Canonical_Items[i]), FALSE, conflictingPattern, status);
1284         }
1285         if (U_FAILURE(status)) { return; }
1286     }
1287 }
1288 
1289 void
setDateTimeFormat(const UnicodeString & dtFormat)1290 DateTimePatternGenerator::setDateTimeFormat(const UnicodeString& dtFormat) {
1291     dateTimeFormat = dtFormat;
1292     // NUL-terminate for the C API.
1293     dateTimeFormat.getTerminatedBuffer();
1294 }
1295 
1296 const UnicodeString&
getDateTimeFormat() const1297 DateTimePatternGenerator::getDateTimeFormat() const {
1298     return dateTimeFormat;
1299 }
1300 
1301 void
setDateTimeFromCalendar(const Locale & locale,UErrorCode & status)1302 DateTimePatternGenerator::setDateTimeFromCalendar(const Locale& locale, UErrorCode& status) {
1303     if (U_FAILURE(status)) { return; }
1304 
1305     const UChar *resStr;
1306     int32_t resStrLen = 0;
1307 
1308     LocalPointer<Calendar> fCalendar(Calendar::createInstance(locale, status), status);
1309     if (U_FAILURE(status)) { return; }
1310 
1311     LocalUResourceBundlePointer calData(ures_open(nullptr, locale.getBaseName(), &status));
1312     if (U_FAILURE(status)) { return; }
1313     ures_getByKey(calData.getAlias(), DT_DateTimeCalendarTag, calData.getAlias(), &status);
1314     if (U_FAILURE(status)) { return; }
1315 
1316     LocalUResourceBundlePointer dateTimePatterns;
1317     if (fCalendar->getType() != nullptr && *fCalendar->getType() != '\0'
1318             && uprv_strcmp(fCalendar->getType(), DT_DateTimeGregorianTag) != 0) {
1319         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), fCalendar->getType(),
1320                                                                 nullptr, &status));
1321         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1322                                   dateTimePatterns.getAlias(), &status);
1323     }
1324 
1325     if (dateTimePatterns.isNull() || status == U_MISSING_RESOURCE_ERROR) {
1326         status = U_ZERO_ERROR;
1327         dateTimePatterns.adoptInstead(ures_getByKeyWithFallback(calData.getAlias(), DT_DateTimeGregorianTag,
1328                                                                 dateTimePatterns.orphan(), &status));
1329         ures_getByKeyWithFallback(dateTimePatterns.getAlias(), DT_DateTimePatternsTag,
1330                                   dateTimePatterns.getAlias(), &status);
1331     }
1332     if (U_FAILURE(status)) { return; }
1333 
1334     if (ures_getSize(dateTimePatterns.getAlias()) <= DateFormat::kDateTime)
1335     {
1336         status = U_INVALID_FORMAT_ERROR;
1337         return;
1338     }
1339     resStr = ures_getStringByIndex(dateTimePatterns.getAlias(), (int32_t)DateFormat::kDateTime, &resStrLen, &status);
1340     setDateTimeFormat(UnicodeString(TRUE, resStr, resStrLen));
1341 }
1342 
1343 void
setDecimalSymbols(const Locale & locale,UErrorCode & status)1344 DateTimePatternGenerator::setDecimalSymbols(const Locale& locale, UErrorCode& status) {
1345     DecimalFormatSymbols dfs = DecimalFormatSymbols(locale, status);
1346     if(U_SUCCESS(status)) {
1347         decimal = dfs.getSymbol(DecimalFormatSymbols::kDecimalSeparatorSymbol);
1348         // NUL-terminate for the C API.
1349         decimal.getTerminatedBuffer();
1350     }
1351 }
1352 
1353 UDateTimePatternConflict
addPattern(const UnicodeString & pattern,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1354 DateTimePatternGenerator::addPattern(
1355     const UnicodeString& pattern,
1356     UBool override,
1357     UnicodeString &conflictingPattern,
1358     UErrorCode& status)
1359 {
1360     if (U_FAILURE(internalErrorCode)) {
1361         status = internalErrorCode;
1362         return UDATPG_NO_CONFLICT;
1363     }
1364 
1365     return addPatternWithSkeleton(pattern, nullptr, override, conflictingPattern, status);
1366 }
1367 
1368 // For DateTimePatternGenerator::addPatternWithSkeleton -
1369 // If skeletonToUse is specified, then an availableFormats entry is being added. In this case:
1370 // 1. We pass that skeleton to matcher.set instead of having it derive a skeleton from the pattern.
1371 // 2. If the new entry's skeleton or basePattern does match an existing entry but that entry also had a skeleton specified
1372 // (i.e. it was also from availableFormats), then the new entry does not override it regardless of the value of the override
1373 // parameter. This prevents later availableFormats entries from a parent locale overriding earlier ones from the actual
1374 // specified locale. However, availableFormats entries *should* override entries with matching skeleton whose skeleton was
1375 // derived (i.e. entries derived from the standard date/time patters for the specified locale).
1376 // 3. When adding the pattern (patternMap->add), we set a new boolean to indicate that the added entry had a
1377 // specified skeleton (which sets a new field in the PtnElem in the PatternMap).
1378 UDateTimePatternConflict
addPatternWithSkeleton(const UnicodeString & pattern,const UnicodeString * skeletonToUse,UBool override,UnicodeString & conflictingPattern,UErrorCode & status)1379 DateTimePatternGenerator::addPatternWithSkeleton(
1380     const UnicodeString& pattern,
1381     const UnicodeString* skeletonToUse,
1382     UBool override,
1383     UnicodeString& conflictingPattern,
1384     UErrorCode& status)
1385 {
1386     if (U_FAILURE(internalErrorCode)) {
1387         status = internalErrorCode;
1388         return UDATPG_NO_CONFLICT;
1389     }
1390 
1391     UnicodeString basePattern;
1392     PtnSkeleton   skeleton;
1393     UDateTimePatternConflict conflictingStatus = UDATPG_NO_CONFLICT;
1394 
1395     DateTimeMatcher matcher;
1396     if ( skeletonToUse == nullptr ) {
1397         matcher.set(pattern, fp, skeleton);
1398         matcher.getBasePattern(basePattern);
1399     } else {
1400         matcher.set(*skeletonToUse, fp, skeleton); // no longer trims skeleton fields to max len 3, per #7930
1401         matcher.getBasePattern(basePattern); // or perhaps instead: basePattern = *skeletonToUse;
1402     }
1403     // We only care about base conflicts - and replacing the pattern associated with a base - if:
1404     // 1. the conflicting previous base pattern did *not* have an explicit skeleton; in that case the previous
1405     // base + pattern combination was derived from either (a) a canonical item, (b) a standard format, or
1406     // (c) a pattern specified programmatically with a previous call to addPattern (which would only happen
1407     // if we are getting here from a subsequent call to addPattern).
1408     // 2. a skeleton is specified for the current pattern, but override=false; in that case we are checking
1409     // availableFormats items from root, which should not override any previous entry with the same base.
1410     UBool entryHadSpecifiedSkeleton;
1411     const UnicodeString *duplicatePattern = patternMap->getPatternFromBasePattern(basePattern, entryHadSpecifiedSkeleton);
1412     if (duplicatePattern != nullptr && (!entryHadSpecifiedSkeleton || (skeletonToUse != nullptr && !override))) {
1413         conflictingStatus = UDATPG_BASE_CONFLICT;
1414         conflictingPattern = *duplicatePattern;
1415         if (!override) {
1416             return conflictingStatus;
1417         }
1418     }
1419     // The only time we get here with override=true and skeletonToUse!=null is when adding availableFormats
1420     // items from CLDR data. In that case, we don't want an item from a parent locale to replace an item with
1421     // same skeleton from the specified locale, so skip the current item if skeletonWasSpecified is true for
1422     // the previously-specified conflicting item.
1423     const PtnSkeleton* entrySpecifiedSkeleton = nullptr;
1424     duplicatePattern = patternMap->getPatternFromSkeleton(skeleton, &entrySpecifiedSkeleton);
1425     if (duplicatePattern != nullptr ) {
1426         conflictingStatus = UDATPG_CONFLICT;
1427         conflictingPattern = *duplicatePattern;
1428         if (!override || (skeletonToUse != nullptr && entrySpecifiedSkeleton != nullptr)) {
1429             return conflictingStatus;
1430         }
1431     }
1432     patternMap->add(basePattern, skeleton, pattern, skeletonToUse != nullptr, status);
1433     if(U_FAILURE(status)) {
1434         return conflictingStatus;
1435     }
1436 
1437     return UDATPG_NO_CONFLICT;
1438 }
1439 
1440 
1441 UDateTimePatternField
getAppendFormatNumber(const char * field) const1442 DateTimePatternGenerator::getAppendFormatNumber(const char* field) const {
1443     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1444         if (uprv_strcmp(CLDR_FIELD_APPEND[i], field)==0) {
1445             return (UDateTimePatternField)i;
1446         }
1447     }
1448     return UDATPG_FIELD_COUNT;
1449 }
1450 
1451 UDateTimePatternField
getFieldAndWidthIndices(const char * key,UDateTimePGDisplayWidth * widthP) const1452 DateTimePatternGenerator::getFieldAndWidthIndices(const char* key, UDateTimePGDisplayWidth* widthP) const {
1453     char cldrFieldKey[UDATPG_FIELD_KEY_MAX + 1];
1454     uprv_strncpy(cldrFieldKey, key, UDATPG_FIELD_KEY_MAX);
1455     cldrFieldKey[UDATPG_FIELD_KEY_MAX]=0; // ensure termination
1456     *widthP = UDATPG_WIDE;
1457     char* hyphenPtr = uprv_strchr(cldrFieldKey, '-');
1458     if (hyphenPtr) {
1459         for (int32_t i=UDATPG_WIDTH_COUNT-1; i>0; --i) {
1460             if (uprv_strcmp(CLDR_FIELD_WIDTH[i], hyphenPtr)==0) {
1461                 *widthP=(UDateTimePGDisplayWidth)i;
1462                 break;
1463             }
1464         }
1465         *hyphenPtr = 0; // now delete width portion of key
1466     }
1467     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
1468         if (uprv_strcmp(CLDR_FIELD_NAME[i],cldrFieldKey)==0) {
1469             return (UDateTimePatternField)i;
1470         }
1471     }
1472     return UDATPG_FIELD_COUNT;
1473 }
1474 
1475 const UnicodeString*
getBestRaw(DateTimeMatcher & source,int32_t includeMask,DistanceInfo * missingFields,UErrorCode & status,const PtnSkeleton ** specifiedSkeletonPtr)1476 DateTimePatternGenerator::getBestRaw(DateTimeMatcher& source,
1477                                      int32_t includeMask,
1478                                      DistanceInfo* missingFields,
1479                                      UErrorCode &status,
1480                                      const PtnSkeleton** specifiedSkeletonPtr) {
1481     int32_t bestDistance = 0x7fffffff;
1482     int32_t bestMissingFieldMask = -1;
1483     DistanceInfo tempInfo;
1484     const UnicodeString *bestPattern=nullptr;
1485     const PtnSkeleton* specifiedSkeleton=nullptr;
1486 
1487     PatternMapIterator it(status);
1488     if (U_FAILURE(status)) { return nullptr; }
1489 
1490     for (it.set(*patternMap); it.hasNext(); ) {
1491         DateTimeMatcher trial = it.next();
1492         if (trial.equals(skipMatcher)) {
1493             continue;
1494         }
1495         int32_t distance=source.getDistance(trial, includeMask, tempInfo);
1496         // Because we iterate over a map the order is undefined. Can change between implementations,
1497         // versions, and will very likely be different between Java and C/C++.
1498         // So if we have patterns with the same distance we also look at the missingFieldMask,
1499         // and we favour the smallest one. Because the field is a bitmask this technically means we
1500         // favour differences in the "least significant fields". For example we prefer the one with differences
1501         // in seconds field vs one with difference in the hours field.
1502         if (distance<bestDistance || (distance==bestDistance && bestMissingFieldMask<tempInfo.missingFieldMask)) {
1503             bestDistance=distance;
1504             bestMissingFieldMask=tempInfo.missingFieldMask;
1505             bestPattern=patternMap->getPatternFromSkeleton(*trial.getSkeletonPtr(), &specifiedSkeleton);
1506             missingFields->setTo(tempInfo);
1507             if (distance==0) {
1508                 break;
1509             }
1510         }
1511     }
1512 
1513     // If the best raw match had a specified skeleton and that skeleton was requested by the caller,
1514     // then return it too. This generally happens when the caller needs to pass that skeleton
1515     // through to adjustFieldTypes so the latter can do a better job.
1516     if (bestPattern && specifiedSkeletonPtr) {
1517         *specifiedSkeletonPtr = specifiedSkeleton;
1518     }
1519     return bestPattern;
1520 }
1521 
1522 UnicodeString
adjustFieldTypes(const UnicodeString & pattern,const PtnSkeleton * specifiedSkeleton,int32_t flags,UDateTimePatternMatchOptions options)1523 DateTimePatternGenerator::adjustFieldTypes(const UnicodeString& pattern,
1524                                            const PtnSkeleton* specifiedSkeleton,
1525                                            int32_t flags,
1526                                            UDateTimePatternMatchOptions options) {
1527     UnicodeString newPattern;
1528     fp->set(pattern);
1529     for (int32_t i=0; i < fp->itemNumber; i++) {
1530         UnicodeString field = fp->items[i];
1531         if ( fp->isQuoteLiteral(field) ) {
1532 
1533             UnicodeString quoteLiteral;
1534             fp->getQuoteLiteral(quoteLiteral, &i);
1535             newPattern += quoteLiteral;
1536         }
1537         else {
1538             if (fp->isPatternSeparator(field)) {
1539                 newPattern+=field;
1540                 continue;
1541             }
1542             int32_t canonicalIndex = fp->getCanonicalIndex(field);
1543             if (canonicalIndex < 0) {
1544                 newPattern+=field;
1545                 continue;  // don't adjust
1546             }
1547             const dtTypeElem *row = &dtTypes[canonicalIndex];
1548             int32_t typeValue = row->field;
1549 
1550             // handle day periods - with #13183, no longer need special handling here, integrated with normal types
1551 
1552             if ((flags & kDTPGFixFractionalSeconds) != 0 && typeValue == UDATPG_SECOND_FIELD) {
1553                 field += decimal;
1554                 dtMatcher->skeleton.original.appendFieldTo(UDATPG_FRACTIONAL_SECOND_FIELD, field);
1555             } else if (dtMatcher->skeleton.type[typeValue]!=0) {
1556                     // Here:
1557                     // - "reqField" is the field from the originally requested skeleton, with length
1558                     // "reqFieldLen".
1559                     // - "field" is the field from the found pattern.
1560                     //
1561                     // The adjusted field should consist of characters from the originally requested
1562                     // skeleton, except in the case of UDATPG_HOUR_FIELD or UDATPG_MONTH_FIELD or
1563                     // UDATPG_WEEKDAY_FIELD or UDATPG_YEAR_FIELD, in which case it should consist
1564                     // of characters from the  found pattern.
1565                     //
1566                     // The length of the adjusted field (adjFieldLen) should match that in the originally
1567                     // requested skeleton, except that in the following cases the length of the adjusted field
1568                     // should match that in the found pattern (i.e. the length of this pattern field should
1569                     // not be adjusted):
1570                     // 1. typeValue is UDATPG_HOUR_FIELD/MINUTE/SECOND and the corresponding bit in options is
1571                     //    not set (ticket #7180). Note, we may want to implement a similar change for other
1572                     //    numeric fields (MM, dd, etc.) so the default behavior is to get locale preference for
1573                     //    field length, but options bits can be used to override this.
1574                     // 2. There is a specified skeleton for the found pattern and one of the following is true:
1575                     //    a) The length of the field in the skeleton (skelFieldLen) is equal to reqFieldLen.
1576                     //    b) The pattern field is numeric and the skeleton field is not, or vice versa.
1577 
1578                     UChar reqFieldChar = dtMatcher->skeleton.original.getFieldChar(typeValue);
1579                     int32_t reqFieldLen = dtMatcher->skeleton.original.getFieldLength(typeValue);
1580                     if (reqFieldChar == CAP_E && reqFieldLen < 3)
1581                         reqFieldLen = 3; // 1-3 for E are equivalent to 3 for c,e
1582                     int32_t adjFieldLen = reqFieldLen;
1583                     if ( (typeValue==UDATPG_HOUR_FIELD && (options & UDATPG_MATCH_HOUR_FIELD_LENGTH)==0) ||
1584                          (typeValue==UDATPG_MINUTE_FIELD && (options & UDATPG_MATCH_MINUTE_FIELD_LENGTH)==0) ||
1585                          (typeValue==UDATPG_SECOND_FIELD && (options & UDATPG_MATCH_SECOND_FIELD_LENGTH)==0) ) {
1586                          adjFieldLen = field.length();
1587                     } else if (specifiedSkeleton) {
1588                         int32_t skelFieldLen = specifiedSkeleton->original.getFieldLength(typeValue);
1589                         UBool patFieldIsNumeric = (row->type > 0);
1590                         UBool skelFieldIsNumeric = (specifiedSkeleton->type[typeValue] > 0);
1591                         if (skelFieldLen == reqFieldLen || (patFieldIsNumeric && !skelFieldIsNumeric) || (skelFieldIsNumeric && !patFieldIsNumeric)) {
1592                             // don't adjust the field length in the found pattern
1593                             adjFieldLen = field.length();
1594                         }
1595                     }
1596                     UChar c = (typeValue!= UDATPG_HOUR_FIELD
1597                             && typeValue!= UDATPG_MONTH_FIELD
1598                             && typeValue!= UDATPG_WEEKDAY_FIELD
1599                             && (typeValue!= UDATPG_YEAR_FIELD || reqFieldChar==CAP_Y))
1600                             ? reqFieldChar
1601                             : field.charAt(0);
1602                     if (typeValue == UDATPG_HOUR_FIELD && (flags & kDTPGSkeletonUsesCapJ) != 0) {
1603                         c = fDefaultHourFormatChar;
1604                     }
1605                     field.remove();
1606                     for (int32_t j=adjFieldLen; j>0; --j) {
1607                         field += c;
1608                     }
1609             }
1610             newPattern+=field;
1611         }
1612     }
1613     return newPattern;
1614 }
1615 
1616 UnicodeString
getBestAppending(int32_t missingFields,int32_t flags,UErrorCode & status,UDateTimePatternMatchOptions options)1617 DateTimePatternGenerator::getBestAppending(int32_t missingFields, int32_t flags, UErrorCode &status, UDateTimePatternMatchOptions options) {
1618     if (U_FAILURE(status)) {
1619         return UnicodeString();
1620     }
1621     UnicodeString  resultPattern, tempPattern;
1622     const UnicodeString* tempPatternPtr;
1623     int32_t lastMissingFieldMask=0;
1624     if (missingFields!=0) {
1625         resultPattern=UnicodeString();
1626         const PtnSkeleton* specifiedSkeleton=nullptr;
1627         tempPatternPtr = getBestRaw(*dtMatcher, missingFields, distanceInfo, status, &specifiedSkeleton);
1628         if (U_FAILURE(status)) {
1629             return UnicodeString();
1630         }
1631         tempPattern = *tempPatternPtr;
1632         resultPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1633         if ( distanceInfo->missingFieldMask==0 ) {
1634             return resultPattern;
1635         }
1636         while (distanceInfo->missingFieldMask!=0) { // precondition: EVERY single field must work!
1637             if ( lastMissingFieldMask == distanceInfo->missingFieldMask ) {
1638                 break;  // cannot find the proper missing field
1639             }
1640             if (((distanceInfo->missingFieldMask & UDATPG_SECOND_AND_FRACTIONAL_MASK)==UDATPG_FRACTIONAL_MASK) &&
1641                 ((missingFields & UDATPG_SECOND_AND_FRACTIONAL_MASK) == UDATPG_SECOND_AND_FRACTIONAL_MASK)) {
1642                 resultPattern = adjustFieldTypes(resultPattern, specifiedSkeleton, flags | kDTPGFixFractionalSeconds, options);
1643                 distanceInfo->missingFieldMask &= ~UDATPG_FRACTIONAL_MASK;
1644                 continue;
1645             }
1646             int32_t startingMask = distanceInfo->missingFieldMask;
1647             tempPatternPtr = getBestRaw(*dtMatcher, distanceInfo->missingFieldMask, distanceInfo, status, &specifiedSkeleton);
1648             if (U_FAILURE(status)) {
1649                 return UnicodeString();
1650             }
1651             tempPattern = *tempPatternPtr;
1652             tempPattern = adjustFieldTypes(tempPattern, specifiedSkeleton, flags, options);
1653             int32_t foundMask=startingMask& ~distanceInfo->missingFieldMask;
1654             int32_t topField=getTopBitNumber(foundMask);
1655 
1656             if (appendItemFormats[topField].length() != 0) {
1657                 UnicodeString appendName;
1658                 getAppendName((UDateTimePatternField)topField, appendName);
1659                 const UnicodeString *values[3] = {
1660                     &resultPattern,
1661                     &tempPattern,
1662                     &appendName
1663                 };
1664                 SimpleFormatter(appendItemFormats[topField], 2, 3, status).
1665                     formatAndReplace(values, 3, resultPattern, nullptr, 0, status);
1666             }
1667             lastMissingFieldMask = distanceInfo->missingFieldMask;
1668         }
1669     }
1670     return resultPattern;
1671 }
1672 
1673 int32_t
getTopBitNumber(int32_t foundMask) const1674 DateTimePatternGenerator::getTopBitNumber(int32_t foundMask) const {
1675     if ( foundMask==0 ) {
1676         return 0;
1677     }
1678     int32_t i=0;
1679     while (foundMask!=0) {
1680         foundMask >>=1;
1681         ++i;
1682     }
1683     if (i-1 >UDATPG_ZONE_FIELD) {
1684         return UDATPG_ZONE_FIELD;
1685     }
1686     else
1687         return i-1;
1688 }
1689 
1690 void
setAvailableFormat(const UnicodeString & key,UErrorCode & err)1691 DateTimePatternGenerator::setAvailableFormat(const UnicodeString &key, UErrorCode& err)
1692 {
1693     fAvailableFormatKeyHash->puti(key, 1, err);
1694 }
1695 
1696 UBool
isAvailableFormatSet(const UnicodeString & key) const1697 DateTimePatternGenerator::isAvailableFormatSet(const UnicodeString &key) const {
1698     return (UBool)(fAvailableFormatKeyHash->geti(key) == 1);
1699 }
1700 
1701 void
copyHashtable(Hashtable * other,UErrorCode & status)1702 DateTimePatternGenerator::copyHashtable(Hashtable *other, UErrorCode &status) {
1703     if (other == nullptr || U_FAILURE(status)) {
1704         return;
1705     }
1706     if (fAvailableFormatKeyHash != nullptr) {
1707         delete fAvailableFormatKeyHash;
1708         fAvailableFormatKeyHash = nullptr;
1709     }
1710     initHashtable(status);
1711     if(U_FAILURE(status)){
1712         return;
1713     }
1714     int32_t pos = UHASH_FIRST;
1715     const UHashElement* elem = nullptr;
1716     // walk through the hash table and create a deep clone
1717     while((elem = other->nextElement(pos))!= nullptr){
1718         const UHashTok otherKeyTok = elem->key;
1719         UnicodeString* otherKey = (UnicodeString*)otherKeyTok.pointer;
1720         fAvailableFormatKeyHash->puti(*otherKey, 1, status);
1721         if(U_FAILURE(status)){
1722             return;
1723         }
1724     }
1725 }
1726 
1727 StringEnumeration*
getSkeletons(UErrorCode & status) const1728 DateTimePatternGenerator::getSkeletons(UErrorCode& status) const {
1729     if (U_FAILURE(status)) {
1730         return nullptr;
1731     }
1732     if (U_FAILURE(internalErrorCode)) {
1733         status = internalErrorCode;
1734         return nullptr;
1735     }
1736     LocalPointer<StringEnumeration> skeletonEnumerator(
1737         new DTSkeletonEnumeration(*patternMap, DT_SKELETON, status), status);
1738 
1739     return U_SUCCESS(status) ? skeletonEnumerator.orphan() : nullptr;
1740 }
1741 
1742 const UnicodeString&
getPatternForSkeleton(const UnicodeString & skeleton) const1743 DateTimePatternGenerator::getPatternForSkeleton(const UnicodeString& skeleton) const {
1744     PtnElem *curElem;
1745 
1746     if (skeleton.length() ==0) {
1747         return emptyString;
1748     }
1749     curElem = patternMap->getHeader(skeleton.charAt(0));
1750     while ( curElem != nullptr ) {
1751         if ( curElem->skeleton->getSkeleton()==skeleton ) {
1752             return curElem->pattern;
1753         }
1754         curElem = curElem->next.getAlias();
1755     }
1756     return emptyString;
1757 }
1758 
1759 StringEnumeration*
getBaseSkeletons(UErrorCode & status) const1760 DateTimePatternGenerator::getBaseSkeletons(UErrorCode& status) const {
1761     if (U_FAILURE(status)) {
1762         return nullptr;
1763     }
1764     if (U_FAILURE(internalErrorCode)) {
1765         status = internalErrorCode;
1766         return nullptr;
1767     }
1768     LocalPointer<StringEnumeration> baseSkeletonEnumerator(
1769         new DTSkeletonEnumeration(*patternMap, DT_BASESKELETON, status), status);
1770 
1771     return U_SUCCESS(status) ? baseSkeletonEnumerator.orphan() : nullptr;
1772 }
1773 
1774 StringEnumeration*
getRedundants(UErrorCode & status)1775 DateTimePatternGenerator::getRedundants(UErrorCode& status) {
1776     if (U_FAILURE(status)) { return nullptr; }
1777     if (U_FAILURE(internalErrorCode)) {
1778         status = internalErrorCode;
1779         return nullptr;
1780     }
1781     LocalPointer<StringEnumeration> output(new DTRedundantEnumeration(), status);
1782     if (U_FAILURE(status)) { return nullptr; }
1783     const UnicodeString *pattern;
1784     PatternMapIterator it(status);
1785     if (U_FAILURE(status)) { return nullptr; }
1786 
1787     for (it.set(*patternMap); it.hasNext(); ) {
1788         DateTimeMatcher current = it.next();
1789         pattern = patternMap->getPatternFromSkeleton(*(it.getSkeleton()));
1790         if ( isCanonicalItem(*pattern) ) {
1791             continue;
1792         }
1793         if ( skipMatcher == nullptr ) {
1794             skipMatcher = new DateTimeMatcher(current);
1795             if (skipMatcher == nullptr) {
1796                 status = U_MEMORY_ALLOCATION_ERROR;
1797                 return nullptr;
1798             }
1799         }
1800         else {
1801             *skipMatcher = current;
1802         }
1803         UnicodeString trial = getBestPattern(current.getPattern(), status);
1804         if (U_FAILURE(status)) { return nullptr; }
1805         if (trial == *pattern) {
1806             ((DTRedundantEnumeration *)output.getAlias())->add(*pattern, status);
1807             if (U_FAILURE(status)) { return nullptr; }
1808         }
1809         if (current.equals(skipMatcher)) {
1810             continue;
1811         }
1812     }
1813     return output.orphan();
1814 }
1815 
1816 UBool
isCanonicalItem(const UnicodeString & item) const1817 DateTimePatternGenerator::isCanonicalItem(const UnicodeString& item) const {
1818     if ( item.length() != 1 ) {
1819         return FALSE;
1820     }
1821     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
1822         if (item.charAt(0)==Canonical_Items[i]) {
1823             return TRUE;
1824         }
1825     }
1826     return FALSE;
1827 }
1828 
1829 
1830 DateTimePatternGenerator*
clone() const1831 DateTimePatternGenerator::clone() const {
1832     return new DateTimePatternGenerator(*this);
1833 }
1834 
PatternMap()1835 PatternMap::PatternMap() {
1836    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1837        boot[i] = nullptr;
1838    }
1839    isDupAllowed = TRUE;
1840 }
1841 
1842 void
copyFrom(const PatternMap & other,UErrorCode & status)1843 PatternMap::copyFrom(const PatternMap& other, UErrorCode& status) {
1844     if (U_FAILURE(status)) {
1845         return;
1846     }
1847     this->isDupAllowed = other.isDupAllowed;
1848     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
1849         PtnElem *curElem, *otherElem, *prevElem=nullptr;
1850         otherElem = other.boot[bootIndex];
1851         while (otherElem != nullptr) {
1852             LocalPointer<PtnElem> newElem(new PtnElem(otherElem->basePattern, otherElem->pattern), status);
1853             if (U_FAILURE(status)) {
1854                 return; // out of memory
1855             }
1856             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(*(otherElem->skeleton)), status);
1857             if (U_FAILURE(status)) {
1858                 return; // out of memory
1859             }
1860             newElem->skeletonWasSpecified = otherElem->skeletonWasSpecified;
1861 
1862             // Release ownership from the LocalPointer of the PtnElem object.
1863             // The PtnElem will now be owned by either the boot (for the first entry in the linked-list)
1864             // or owned by the previous PtnElem object in the linked-list.
1865             curElem = newElem.orphan();
1866 
1867             if (this->boot[bootIndex] == nullptr) {
1868                 this->boot[bootIndex] = curElem;
1869             } else {
1870                 if (prevElem != nullptr) {
1871                     prevElem->next.adoptInstead(curElem);
1872                 } else {
1873                     UPRV_UNREACHABLE;
1874                 }
1875             }
1876             prevElem = curElem;
1877             otherElem = otherElem->next.getAlias();
1878         }
1879 
1880     }
1881 }
1882 
1883 PtnElem*
getHeader(UChar baseChar) const1884 PatternMap::getHeader(UChar baseChar) const {
1885     PtnElem* curElem;
1886 
1887     if ( (baseChar >= CAP_A) && (baseChar <= CAP_Z) ) {
1888          curElem = boot[baseChar-CAP_A];
1889     }
1890     else {
1891         if ( (baseChar >=LOW_A) && (baseChar <= LOW_Z) ) {
1892             curElem = boot[26+baseChar-LOW_A];
1893         }
1894         else {
1895             return nullptr;
1896         }
1897     }
1898     return curElem;
1899 }
1900 
~PatternMap()1901 PatternMap::~PatternMap() {
1902    for (int32_t i=0; i < MAX_PATTERN_ENTRIES; ++i ) {
1903        if (boot[i] != nullptr ) {
1904            delete boot[i];
1905            boot[i] = nullptr;
1906        }
1907    }
1908 }  // PatternMap destructor
1909 
1910 void
add(const UnicodeString & basePattern,const PtnSkeleton & skeleton,const UnicodeString & value,UBool skeletonWasSpecified,UErrorCode & status)1911 PatternMap::add(const UnicodeString& basePattern,
1912                 const PtnSkeleton& skeleton,
1913                 const UnicodeString& value,// mapped pattern value
1914                 UBool skeletonWasSpecified,
1915                 UErrorCode &status) {
1916     UChar baseChar = basePattern.charAt(0);
1917     PtnElem *curElem, *baseElem;
1918     status = U_ZERO_ERROR;
1919 
1920     // the baseChar must be A-Z or a-z
1921     if ((baseChar >= CAP_A) && (baseChar <= CAP_Z)) {
1922         baseElem = boot[baseChar-CAP_A];
1923     }
1924     else {
1925         if ((baseChar >=LOW_A) && (baseChar <= LOW_Z)) {
1926             baseElem = boot[26+baseChar-LOW_A];
1927          }
1928          else {
1929              status = U_ILLEGAL_CHARACTER;
1930              return;
1931          }
1932     }
1933 
1934     if (baseElem == nullptr) {
1935         LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1936         if (U_FAILURE(status)) {
1937             return; // out of memory
1938         }
1939         newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1940         if (U_FAILURE(status)) {
1941             return; // out of memory
1942         }
1943         newElem->skeletonWasSpecified = skeletonWasSpecified;
1944         if (baseChar >= LOW_A) {
1945             boot[26 + (baseChar - LOW_A)] = newElem.orphan(); // the boot array now owns the PtnElem.
1946         }
1947         else {
1948             boot[baseChar - CAP_A] = newElem.orphan(); // the boot array now owns the PtnElem.
1949         }
1950     }
1951     if ( baseElem != nullptr ) {
1952         curElem = getDuplicateElem(basePattern, skeleton, baseElem);
1953 
1954         if (curElem == nullptr) {
1955             // add new element to the list.
1956             curElem = baseElem;
1957             while( curElem -> next != nullptr )
1958             {
1959                 curElem = curElem->next.getAlias();
1960             }
1961 
1962             LocalPointer<PtnElem> newElem(new PtnElem(basePattern, value), status);
1963             if (U_FAILURE(status)) {
1964                 return; // out of memory
1965             }
1966             newElem->skeleton.adoptInsteadAndCheckErrorCode(new PtnSkeleton(skeleton), status);
1967             if (U_FAILURE(status)) {
1968                 return; // out of memory
1969             }
1970             newElem->skeletonWasSpecified = skeletonWasSpecified;
1971             curElem->next.adoptInstead(newElem.orphan());
1972             curElem = curElem->next.getAlias();
1973         }
1974         else {
1975             // Pattern exists in the list already.
1976             if ( !isDupAllowed ) {
1977                 return;
1978             }
1979             // Overwrite the value.
1980             curElem->pattern = value;
1981             // It was a bug that we were not doing the following previously,
1982             // though that bug hid other problems by making things partly work.
1983             curElem->skeletonWasSpecified = skeletonWasSpecified;
1984         }
1985     }
1986 }  // PatternMap::add
1987 
1988 // Find the pattern from the given basePattern string.
1989 const UnicodeString *
getPatternFromBasePattern(const UnicodeString & basePattern,UBool & skeletonWasSpecified) const1990 PatternMap::getPatternFromBasePattern(const UnicodeString& basePattern, UBool& skeletonWasSpecified) const { // key to search for
1991    PtnElem *curElem;
1992 
1993    if ((curElem=getHeader(basePattern.charAt(0)))==nullptr) {
1994        return nullptr;  // no match
1995    }
1996 
1997    do  {
1998        if ( basePattern.compare(curElem->basePattern)==0 ) {
1999           skeletonWasSpecified = curElem->skeletonWasSpecified;
2000           return &(curElem->pattern);
2001        }
2002        curElem = curElem->next.getAlias();
2003    } while (curElem != nullptr);
2004 
2005    return nullptr;
2006 }  // PatternMap::getFromBasePattern
2007 
2008 
2009 // Find the pattern from the given skeleton.
2010 // At least when this is called from getBestRaw & addPattern (in which case specifiedSkeletonPtr is non-NULL),
2011 // the comparison should be based on skeleton.original (which is unique and tied to the distance measurement in bestRaw)
2012 // and not skeleton.baseOriginal (which is not unique); otherwise we may pick a different skeleton than the one with the
2013 // optimum distance value in getBestRaw. When this is called from public getRedundants (specifiedSkeletonPtr is NULL),
2014 // for now it will continue to compare based on baseOriginal so as not to change the behavior unnecessarily.
2015 const UnicodeString *
getPatternFromSkeleton(const PtnSkeleton & skeleton,const PtnSkeleton ** specifiedSkeletonPtr) const2016 PatternMap::getPatternFromSkeleton(const PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr) const { // key to search for
2017    PtnElem *curElem;
2018 
2019    if (specifiedSkeletonPtr) {
2020        *specifiedSkeletonPtr = nullptr;
2021    }
2022 
2023    // find boot entry
2024    UChar baseChar = skeleton.getFirstChar();
2025    if ((curElem=getHeader(baseChar))==nullptr) {
2026        return nullptr;  // no match
2027    }
2028 
2029    do  {
2030        UBool equal;
2031        if (specifiedSkeletonPtr != nullptr) { // called from DateTimePatternGenerator::getBestRaw or addPattern, use original
2032            equal = curElem->skeleton->original == skeleton.original;
2033        } else { // called from DateTimePatternGenerator::getRedundants, use baseOriginal
2034            equal = curElem->skeleton->baseOriginal == skeleton.baseOriginal;
2035        }
2036        if (equal) {
2037            if (specifiedSkeletonPtr && curElem->skeletonWasSpecified) {
2038                *specifiedSkeletonPtr = curElem->skeleton.getAlias();
2039            }
2040            return &(curElem->pattern);
2041        }
2042        curElem = curElem->next.getAlias();
2043    } while (curElem != nullptr);
2044 
2045    return nullptr;
2046 }
2047 
2048 UBool
equals(const PatternMap & other) const2049 PatternMap::equals(const PatternMap& other) const {
2050     if ( this==&other ) {
2051         return TRUE;
2052     }
2053     for (int32_t bootIndex = 0; bootIndex < MAX_PATTERN_ENTRIES; ++bootIndex) {
2054         if (boot[bootIndex] == other.boot[bootIndex]) {
2055             continue;
2056         }
2057         if ((boot[bootIndex] == nullptr) || (other.boot[bootIndex] == nullptr)) {
2058             return FALSE;
2059         }
2060         PtnElem *otherElem = other.boot[bootIndex];
2061         PtnElem *myElem = boot[bootIndex];
2062         while ((otherElem != nullptr) || (myElem != nullptr)) {
2063             if ( myElem == otherElem ) {
2064                 break;
2065             }
2066             if ((otherElem == nullptr) || (myElem == nullptr)) {
2067                 return FALSE;
2068             }
2069             if ( (myElem->basePattern != otherElem->basePattern) ||
2070                  (myElem->pattern != otherElem->pattern) ) {
2071                 return FALSE;
2072             }
2073             if ((myElem->skeleton.getAlias() != otherElem->skeleton.getAlias()) &&
2074                 !myElem->skeleton->equals(*(otherElem->skeleton))) {
2075                 return FALSE;
2076             }
2077             myElem = myElem->next.getAlias();
2078             otherElem = otherElem->next.getAlias();
2079         }
2080     }
2081     return TRUE;
2082 }
2083 
2084 // find any key existing in the mapping table already.
2085 // return TRUE if there is an existing key, otherwise return FALSE.
2086 PtnElem*
getDuplicateElem(const UnicodeString & basePattern,const PtnSkeleton & skeleton,PtnElem * baseElem)2087 PatternMap::getDuplicateElem(
2088             const UnicodeString &basePattern,
2089             const PtnSkeleton &skeleton,
2090             PtnElem *baseElem) {
2091    PtnElem *curElem;
2092 
2093    if ( baseElem == nullptr ) {
2094          return nullptr;
2095    }
2096    else {
2097          curElem = baseElem;
2098    }
2099    do {
2100      if ( basePattern.compare(curElem->basePattern)==0 ) {
2101          UBool isEqual = TRUE;
2102          for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2103             if (curElem->skeleton->type[i] != skeleton.type[i] ) {
2104                 isEqual = FALSE;
2105                 break;
2106             }
2107         }
2108         if (isEqual) {
2109             return curElem;
2110         }
2111      }
2112      curElem = curElem->next.getAlias();
2113    } while( curElem != nullptr );
2114 
2115    // end of the list
2116    return nullptr;
2117 
2118 }  // PatternMap::getDuplicateElem
2119 
DateTimeMatcher(void)2120 DateTimeMatcher::DateTimeMatcher(void) {
2121 }
2122 
~DateTimeMatcher()2123 DateTimeMatcher::~DateTimeMatcher() {}
2124 
DateTimeMatcher(const DateTimeMatcher & other)2125 DateTimeMatcher::DateTimeMatcher(const DateTimeMatcher& other) {
2126     copyFrom(other.skeleton);
2127 }
2128 
2129 
2130 void
set(const UnicodeString & pattern,FormatParser * fp)2131 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp) {
2132     PtnSkeleton localSkeleton;
2133     return set(pattern, fp, localSkeleton);
2134 }
2135 
2136 void
set(const UnicodeString & pattern,FormatParser * fp,PtnSkeleton & skeletonResult)2137 DateTimeMatcher::set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeletonResult) {
2138     int32_t i;
2139     for (i=0; i<UDATPG_FIELD_COUNT; ++i) {
2140         skeletonResult.type[i] = NONE;
2141     }
2142     skeletonResult.original.clear();
2143     skeletonResult.baseOriginal.clear();
2144     skeletonResult.addedDefaultDayPeriod = FALSE;
2145 
2146     fp->set(pattern);
2147     for (i=0; i < fp->itemNumber; i++) {
2148         const UnicodeString& value = fp->items[i];
2149         // don't skip 'a' anymore, dayPeriod handled specially below
2150 
2151         if ( fp->isQuoteLiteral(value) ) {
2152             UnicodeString quoteLiteral;
2153             fp->getQuoteLiteral(quoteLiteral, &i);
2154             continue;
2155         }
2156         int32_t canonicalIndex = fp->getCanonicalIndex(value);
2157         if (canonicalIndex < 0) {
2158             continue;
2159         }
2160         const dtTypeElem *row = &dtTypes[canonicalIndex];
2161         int32_t field = row->field;
2162         skeletonResult.original.populate(field, value);
2163         UChar repeatChar = row->patternChar;
2164         int32_t repeatCount = row->minLen;
2165         skeletonResult.baseOriginal.populate(field, repeatChar, repeatCount);
2166         int16_t subField = row->type;
2167         if (row->type > 0) {
2168             U_ASSERT(value.length() < INT16_MAX);
2169             subField += static_cast<int16_t>(value.length());
2170         }
2171         skeletonResult.type[field] = subField;
2172     }
2173 
2174     // #20739, we have a skeleton with minutes and milliseconds, but no seconds
2175     //
2176     // Theoretically we would need to check and fix all fields with "gaps":
2177     // for example year-day (no month), month-hour (no day), and so on, All the possible field combinations.
2178     // Plus some smartness: year + hour => should we add month, or add day-of-year?
2179     // What about month + day-of-week, or month + am/pm indicator.
2180     // I think beyond a certain point we should not try to fix bad developer input and try guessing what they mean.
2181     // Garbage in, garbage out.
2182     if (!skeletonResult.original.isFieldEmpty(UDATPG_MINUTE_FIELD)
2183         && !skeletonResult.original.isFieldEmpty(UDATPG_FRACTIONAL_SECOND_FIELD)
2184         && skeletonResult.original.isFieldEmpty(UDATPG_SECOND_FIELD)) {
2185         // Force the use of seconds
2186         for (i = 0; dtTypes[i].patternChar != 0; i++) {
2187             if (dtTypes[i].field == UDATPG_SECOND_FIELD) {
2188                 // first entry for UDATPG_SECOND_FIELD
2189                 skeletonResult.original.populate(UDATPG_SECOND_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2190                 skeletonResult.baseOriginal.populate(UDATPG_SECOND_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2191                 // We add value.length, same as above, when type is first initialized.
2192                 // The value we want to "fake" here is "s", and 1 means "s".length()
2193                 int16_t subField = dtTypes[i].type;
2194                 skeletonResult.type[UDATPG_SECOND_FIELD] = (subField > 0) ? subField + 1 : subField;
2195                 break;
2196             }
2197         }
2198     }
2199 
2200     // #13183, handle special behavior for day period characters (a, b, B)
2201     if (!skeletonResult.original.isFieldEmpty(UDATPG_HOUR_FIELD)) {
2202         if (skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==LOW_H || skeletonResult.original.getFieldChar(UDATPG_HOUR_FIELD)==CAP_K) {
2203             // We have a skeleton with 12-hour-cycle format
2204             if (skeletonResult.original.isFieldEmpty(UDATPG_DAYPERIOD_FIELD)) {
2205                 // But we do not have a day period in the skeleton; add the default DAYPERIOD (currently "a")
2206                 for (i = 0; dtTypes[i].patternChar != 0; i++) {
2207                     if ( dtTypes[i].field == UDATPG_DAYPERIOD_FIELD ) {
2208                         // first entry for UDATPG_DAYPERIOD_FIELD
2209                         skeletonResult.original.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2210                         skeletonResult.baseOriginal.populate(UDATPG_DAYPERIOD_FIELD, dtTypes[i].patternChar, dtTypes[i].minLen);
2211                         skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = dtTypes[i].type;
2212                         skeletonResult.addedDefaultDayPeriod = TRUE;
2213                         break;
2214                     }
2215                 }
2216             }
2217         } else {
2218             // Skeleton has 24-hour-cycle hour format and has dayPeriod, delete dayPeriod (i.e. ignore it)
2219             skeletonResult.original.clearField(UDATPG_DAYPERIOD_FIELD);
2220             skeletonResult.baseOriginal.clearField(UDATPG_DAYPERIOD_FIELD);
2221             skeletonResult.type[UDATPG_DAYPERIOD_FIELD] = NONE;
2222         }
2223     }
2224     copyFrom(skeletonResult);
2225 }
2226 
2227 void
getBasePattern(UnicodeString & result)2228 DateTimeMatcher::getBasePattern(UnicodeString &result ) {
2229     result.remove(); // Reset the result first.
2230     skeleton.baseOriginal.appendTo(result);
2231 }
2232 
2233 UnicodeString
getPattern()2234 DateTimeMatcher::getPattern() {
2235     UnicodeString result;
2236     return skeleton.original.appendTo(result);
2237 }
2238 
2239 int32_t
getDistance(const DateTimeMatcher & other,int32_t includeMask,DistanceInfo & distanceInfo) const2240 DateTimeMatcher::getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo) const {
2241     int32_t result = 0;
2242     distanceInfo.clear();
2243     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i ) {
2244         int32_t myType = (includeMask&(1<<i))==0 ? 0 : skeleton.type[i];
2245         int32_t otherType = other.skeleton.type[i];
2246         if (myType==otherType) {
2247             continue;
2248         }
2249         if (myType==0) {// and other is not
2250             result += EXTRA_FIELD;
2251             distanceInfo.addExtra(i);
2252         }
2253         else {
2254             if (otherType==0) {
2255                 result += MISSING_FIELD;
2256                 distanceInfo.addMissing(i);
2257             }
2258             else {
2259                 result += abs(myType - otherType);
2260             }
2261         }
2262 
2263     }
2264     return result;
2265 }
2266 
2267 void
copyFrom(const PtnSkeleton & newSkeleton)2268 DateTimeMatcher::copyFrom(const PtnSkeleton& newSkeleton) {
2269     skeleton.copyFrom(newSkeleton);
2270 }
2271 
2272 void
copyFrom()2273 DateTimeMatcher::copyFrom() {
2274     // same as clear
2275     skeleton.clear();
2276 }
2277 
2278 UBool
equals(const DateTimeMatcher * other) const2279 DateTimeMatcher::equals(const DateTimeMatcher* other) const {
2280     if (other==nullptr) { return FALSE; }
2281     return skeleton.original == other->skeleton.original;
2282 }
2283 
2284 int32_t
getFieldMask() const2285 DateTimeMatcher::getFieldMask() const {
2286     int32_t result = 0;
2287 
2288     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2289         if (skeleton.type[i]!=0) {
2290             result |= (1<<i);
2291         }
2292     }
2293     return result;
2294 }
2295 
2296 PtnSkeleton*
getSkeletonPtr()2297 DateTimeMatcher::getSkeletonPtr() {
2298     return &skeleton;
2299 }
2300 
FormatParser()2301 FormatParser::FormatParser () {
2302     status = START;
2303     itemNumber = 0;
2304 }
2305 
2306 
~FormatParser()2307 FormatParser::~FormatParser () {
2308 }
2309 
2310 
2311 // Find the next token with the starting position and length
2312 // Note: the startPos may
2313 FormatParser::TokenStatus
setTokens(const UnicodeString & pattern,int32_t startPos,int32_t * len)2314 FormatParser::setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len) {
2315     int32_t curLoc = startPos;
2316     if ( curLoc >= pattern.length()) {
2317         return DONE;
2318     }
2319     // check the current char is between A-Z or a-z
2320     do {
2321         UChar c=pattern.charAt(curLoc);
2322         if ( (c>=CAP_A && c<=CAP_Z) || (c>=LOW_A && c<=LOW_Z) ) {
2323            curLoc++;
2324         }
2325         else {
2326                startPos = curLoc;
2327                *len=1;
2328                return ADD_TOKEN;
2329         }
2330 
2331         if ( pattern.charAt(curLoc)!= pattern.charAt(startPos) ) {
2332             break;  // not the same token
2333         }
2334     } while(curLoc <= pattern.length());
2335     *len = curLoc-startPos;
2336     return ADD_TOKEN;
2337 }
2338 
2339 void
set(const UnicodeString & pattern)2340 FormatParser::set(const UnicodeString& pattern) {
2341     int32_t startPos = 0;
2342     TokenStatus result = START;
2343     int32_t len = 0;
2344     itemNumber = 0;
2345 
2346     do {
2347         result = setTokens( pattern, startPos, &len );
2348         if ( result == ADD_TOKEN )
2349         {
2350             items[itemNumber++] = UnicodeString(pattern, startPos, len );
2351             startPos += len;
2352         }
2353         else {
2354             break;
2355         }
2356     } while (result==ADD_TOKEN && itemNumber < MAX_DT_TOKEN);
2357 }
2358 
2359 int32_t
getCanonicalIndex(const UnicodeString & s,UBool strict)2360 FormatParser::getCanonicalIndex(const UnicodeString& s, UBool strict) {
2361     int32_t len = s.length();
2362     if (len == 0) {
2363         return -1;
2364     }
2365     UChar ch = s.charAt(0);
2366 
2367     // Verify that all are the same character.
2368     for (int32_t l = 1; l < len; l++) {
2369         if (ch != s.charAt(l)) {
2370             return -1;
2371         }
2372     }
2373     int32_t i = 0;
2374     int32_t bestRow = -1;
2375     while (dtTypes[i].patternChar != 0x0000) {
2376         if ( dtTypes[i].patternChar != ch ) {
2377             ++i;
2378             continue;
2379         }
2380         bestRow = i;
2381         if (dtTypes[i].patternChar != dtTypes[i+1].patternChar) {
2382             return i;
2383         }
2384         if (dtTypes[i+1].minLen <= len) {
2385             ++i;
2386             continue;
2387         }
2388         return i;
2389     }
2390     return strict ? -1 : bestRow;
2391 }
2392 
2393 UBool
isQuoteLiteral(const UnicodeString & s)2394 FormatParser::isQuoteLiteral(const UnicodeString& s) {
2395     return (UBool)(s.charAt(0) == SINGLE_QUOTE);
2396 }
2397 
2398 // This function assumes the current itemIndex points to the quote literal.
2399 // Please call isQuoteLiteral prior to this function.
2400 void
getQuoteLiteral(UnicodeString & quote,int32_t * itemIndex)2401 FormatParser::getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex) {
2402     int32_t i = *itemIndex;
2403 
2404     quote.remove();
2405     if (items[i].charAt(0)==SINGLE_QUOTE) {
2406         quote += items[i];
2407         ++i;
2408     }
2409     while ( i < itemNumber ) {
2410         if ( items[i].charAt(0)==SINGLE_QUOTE ) {
2411             if ( (i+1<itemNumber) && (items[i+1].charAt(0)==SINGLE_QUOTE)) {
2412                 // two single quotes e.g. 'o''clock'
2413                 quote += items[i++];
2414                 quote += items[i++];
2415                 continue;
2416             }
2417             else {
2418                 quote += items[i];
2419                 break;
2420             }
2421         }
2422         else {
2423             quote += items[i];
2424         }
2425         ++i;
2426     }
2427     *itemIndex=i;
2428 }
2429 
2430 UBool
isPatternSeparator(const UnicodeString & field) const2431 FormatParser::isPatternSeparator(const UnicodeString& field) const {
2432     for (int32_t i=0; i<field.length(); ++i ) {
2433         UChar c= field.charAt(i);
2434         if ( (c==SINGLE_QUOTE) || (c==BACKSLASH) || (c==SPACE) || (c==COLON) ||
2435              (c==QUOTATION_MARK) || (c==COMMA) || (c==HYPHEN) ||(items[i].charAt(0)==DOT) ) {
2436             continue;
2437         }
2438         else {
2439             return FALSE;
2440         }
2441     }
2442     return TRUE;
2443 }
2444 
~DistanceInfo()2445 DistanceInfo::~DistanceInfo() {}
2446 
2447 void
setTo(const DistanceInfo & other)2448 DistanceInfo::setTo(const DistanceInfo& other) {
2449     missingFieldMask = other.missingFieldMask;
2450     extraFieldMask= other.extraFieldMask;
2451 }
2452 
PatternMapIterator(UErrorCode & status)2453 PatternMapIterator::PatternMapIterator(UErrorCode& status) :
2454     bootIndex(0), nodePtr(nullptr), matcher(nullptr), patternMap(nullptr)
2455 {
2456     if (U_FAILURE(status)) { return; }
2457     matcher.adoptInsteadAndCheckErrorCode(new DateTimeMatcher(), status);
2458 }
2459 
~PatternMapIterator()2460 PatternMapIterator::~PatternMapIterator() {
2461 }
2462 
2463 void
set(PatternMap & newPatternMap)2464 PatternMapIterator::set(PatternMap& newPatternMap) {
2465     this->patternMap=&newPatternMap;
2466 }
2467 
2468 PtnSkeleton*
getSkeleton() const2469 PatternMapIterator::getSkeleton() const {
2470     if ( nodePtr == nullptr ) {
2471         return nullptr;
2472     }
2473     else {
2474         return nodePtr->skeleton.getAlias();
2475     }
2476 }
2477 
2478 UBool
hasNext() const2479 PatternMapIterator::hasNext() const {
2480     int32_t headIndex = bootIndex;
2481     PtnElem *curPtr = nodePtr;
2482 
2483     if (patternMap==nullptr) {
2484         return FALSE;
2485     }
2486     while ( headIndex < MAX_PATTERN_ENTRIES ) {
2487         if ( curPtr != nullptr ) {
2488             if ( curPtr->next != nullptr ) {
2489                 return TRUE;
2490             }
2491             else {
2492                 headIndex++;
2493                 curPtr=nullptr;
2494                 continue;
2495             }
2496         }
2497         else {
2498             if ( patternMap->boot[headIndex] != nullptr ) {
2499                 return TRUE;
2500             }
2501             else {
2502                 headIndex++;
2503                 continue;
2504             }
2505         }
2506     }
2507     return FALSE;
2508 }
2509 
2510 DateTimeMatcher&
next()2511 PatternMapIterator::next() {
2512     while ( bootIndex < MAX_PATTERN_ENTRIES ) {
2513         if ( nodePtr != nullptr ) {
2514             if ( nodePtr->next != nullptr ) {
2515                 nodePtr = nodePtr->next.getAlias();
2516                 break;
2517             }
2518             else {
2519                 bootIndex++;
2520                 nodePtr=nullptr;
2521                 continue;
2522             }
2523         }
2524         else {
2525             if ( patternMap->boot[bootIndex] != nullptr ) {
2526                 nodePtr = patternMap->boot[bootIndex];
2527                 break;
2528             }
2529             else {
2530                 bootIndex++;
2531                 continue;
2532             }
2533         }
2534     }
2535     if (nodePtr!=nullptr) {
2536         matcher->copyFrom(*nodePtr->skeleton);
2537     }
2538     else {
2539         matcher->copyFrom();
2540     }
2541     return *matcher;
2542 }
2543 
2544 
SkeletonFields()2545 SkeletonFields::SkeletonFields() {
2546     // Set initial values to zero
2547     clear();
2548 }
2549 
clear()2550 void SkeletonFields::clear() {
2551     uprv_memset(chars, 0, sizeof(chars));
2552     uprv_memset(lengths, 0, sizeof(lengths));
2553 }
2554 
copyFrom(const SkeletonFields & other)2555 void SkeletonFields::copyFrom(const SkeletonFields& other) {
2556     uprv_memcpy(chars, other.chars, sizeof(chars));
2557     uprv_memcpy(lengths, other.lengths, sizeof(lengths));
2558 }
2559 
clearField(int32_t field)2560 void SkeletonFields::clearField(int32_t field) {
2561     chars[field] = 0;
2562     lengths[field] = 0;
2563 }
2564 
getFieldChar(int32_t field) const2565 UChar SkeletonFields::getFieldChar(int32_t field) const {
2566     return chars[field];
2567 }
2568 
getFieldLength(int32_t field) const2569 int32_t SkeletonFields::getFieldLength(int32_t field) const {
2570     return lengths[field];
2571 }
2572 
populate(int32_t field,const UnicodeString & value)2573 void SkeletonFields::populate(int32_t field, const UnicodeString& value) {
2574     populate(field, value.charAt(0), value.length());
2575 }
2576 
populate(int32_t field,UChar ch,int32_t length)2577 void SkeletonFields::populate(int32_t field, UChar ch, int32_t length) {
2578     chars[field] = (int8_t) ch;
2579     lengths[field] = (int8_t) length;
2580 }
2581 
isFieldEmpty(int32_t field) const2582 UBool SkeletonFields::isFieldEmpty(int32_t field) const {
2583     return lengths[field] == 0;
2584 }
2585 
appendTo(UnicodeString & string) const2586 UnicodeString& SkeletonFields::appendTo(UnicodeString& string) const {
2587     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2588         appendFieldTo(i, string);
2589     }
2590     return string;
2591 }
2592 
appendFieldTo(int32_t field,UnicodeString & string) const2593 UnicodeString& SkeletonFields::appendFieldTo(int32_t field, UnicodeString& string) const {
2594     UChar ch(chars[field]);
2595     int32_t length = (int32_t) lengths[field];
2596 
2597     for (int32_t i=0; i<length; i++) {
2598         string += ch;
2599     }
2600     return string;
2601 }
2602 
getFirstChar() const2603 UChar SkeletonFields::getFirstChar() const {
2604     for (int32_t i = 0; i < UDATPG_FIELD_COUNT; ++i) {
2605         if (lengths[i] != 0) {
2606             return chars[i];
2607         }
2608     }
2609     return '\0';
2610 }
2611 
2612 
PtnSkeleton()2613 PtnSkeleton::PtnSkeleton()
2614     : addedDefaultDayPeriod(FALSE) {
2615 }
2616 
PtnSkeleton(const PtnSkeleton & other)2617 PtnSkeleton::PtnSkeleton(const PtnSkeleton& other) {
2618     copyFrom(other);
2619 }
2620 
copyFrom(const PtnSkeleton & other)2621 void PtnSkeleton::copyFrom(const PtnSkeleton& other) {
2622     uprv_memcpy(type, other.type, sizeof(type));
2623     original.copyFrom(other.original);
2624     baseOriginal.copyFrom(other.baseOriginal);
2625     addedDefaultDayPeriod = other.addedDefaultDayPeriod;
2626 }
2627 
clear()2628 void PtnSkeleton::clear() {
2629     uprv_memset(type, 0, sizeof(type));
2630     original.clear();
2631     baseOriginal.clear();
2632 }
2633 
2634 UBool
equals(const PtnSkeleton & other) const2635 PtnSkeleton::equals(const PtnSkeleton& other) const  {
2636     return (original == other.original)
2637         && (baseOriginal == other.baseOriginal)
2638         && (uprv_memcmp(type, other.type, sizeof(type)) == 0);
2639 }
2640 
2641 UnicodeString
getSkeleton() const2642 PtnSkeleton::getSkeleton() const {
2643     UnicodeString result;
2644     result = original.appendTo(result);
2645     int32_t pos;
2646     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2647         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2648         // was not in the provided skeleton, remove it here before returning skeleton.
2649         result.remove(pos, 1);
2650     }
2651     return result;
2652 }
2653 
2654 UnicodeString
getBaseSkeleton() const2655 PtnSkeleton::getBaseSkeleton() const {
2656     UnicodeString result;
2657     result = baseOriginal.appendTo(result);
2658     int32_t pos;
2659     if (addedDefaultDayPeriod && (pos = result.indexOf(LOW_A)) >= 0) {
2660         // for backward compatibility: if DateTimeMatcher.set added a single 'a' that
2661         // was not in the provided skeleton, remove it here before returning skeleton.
2662         result.remove(pos, 1);
2663     }
2664     return result;
2665 }
2666 
2667 UChar
getFirstChar() const2668 PtnSkeleton::getFirstChar() const {
2669     return baseOriginal.getFirstChar();
2670 }
2671 
~PtnSkeleton()2672 PtnSkeleton::~PtnSkeleton() {
2673 }
2674 
PtnElem(const UnicodeString & basePat,const UnicodeString & pat)2675 PtnElem::PtnElem(const UnicodeString &basePat, const UnicodeString &pat) :
2676     basePattern(basePat), skeleton(nullptr), pattern(pat), next(nullptr)
2677 {
2678 }
2679 
~PtnElem()2680 PtnElem::~PtnElem() {
2681 }
2682 
DTSkeletonEnumeration(PatternMap & patternMap,dtStrEnum type,UErrorCode & status)2683 DTSkeletonEnumeration::DTSkeletonEnumeration(PatternMap& patternMap, dtStrEnum type, UErrorCode& status) : fSkeletons(nullptr) {
2684     PtnElem  *curElem;
2685     PtnSkeleton *curSkeleton;
2686     UnicodeString s;
2687     int32_t bootIndex;
2688 
2689     pos=0;
2690     fSkeletons.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2691     if (U_FAILURE(status)) {
2692         return;
2693     }
2694 
2695     for (bootIndex=0; bootIndex<MAX_PATTERN_ENTRIES; ++bootIndex ) {
2696         curElem = patternMap.boot[bootIndex];
2697         while (curElem!=nullptr) {
2698             switch(type) {
2699                 case DT_BASESKELETON:
2700                     s=curElem->basePattern;
2701                     break;
2702                 case DT_PATTERN:
2703                     s=curElem->pattern;
2704                     break;
2705                 case DT_SKELETON:
2706                     curSkeleton=curElem->skeleton.getAlias();
2707                     s=curSkeleton->getSkeleton();
2708                     break;
2709             }
2710             if ( !isCanonicalItem(s) ) {
2711                 LocalPointer<UnicodeString> newElem(new UnicodeString(s), status);
2712                 if (U_FAILURE(status)) {
2713                     return;
2714                 }
2715                 fSkeletons->addElement(newElem.getAlias(), status);
2716                 if (U_FAILURE(status)) {
2717                     fSkeletons.adoptInstead(nullptr);
2718                     return;
2719                 }
2720                 newElem.orphan(); // fSkeletons vector now owns the UnicodeString.
2721             }
2722             curElem = curElem->next.getAlias();
2723         }
2724     }
2725     if ((bootIndex==MAX_PATTERN_ENTRIES) && (curElem!=nullptr) ) {
2726         status = U_BUFFER_OVERFLOW_ERROR;
2727     }
2728 }
2729 
2730 const UnicodeString*
snext(UErrorCode & status)2731 DTSkeletonEnumeration::snext(UErrorCode& status) {
2732     if (U_SUCCESS(status) && fSkeletons.isValid() && pos < fSkeletons->size()) {
2733         return (const UnicodeString*)fSkeletons->elementAt(pos++);
2734     }
2735     return nullptr;
2736 }
2737 
2738 void
reset(UErrorCode &)2739 DTSkeletonEnumeration::reset(UErrorCode& /*status*/) {
2740     pos=0;
2741 }
2742 
2743 int32_t
count(UErrorCode &) const2744 DTSkeletonEnumeration::count(UErrorCode& /*status*/) const {
2745    return (fSkeletons.isNull()) ? 0 : fSkeletons->size();
2746 }
2747 
2748 UBool
isCanonicalItem(const UnicodeString & item)2749 DTSkeletonEnumeration::isCanonicalItem(const UnicodeString& item) {
2750     if ( item.length() != 1 ) {
2751         return FALSE;
2752     }
2753     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2754         if (item.charAt(0)==Canonical_Items[i]) {
2755             return TRUE;
2756         }
2757     }
2758     return FALSE;
2759 }
2760 
~DTSkeletonEnumeration()2761 DTSkeletonEnumeration::~DTSkeletonEnumeration() {
2762     UnicodeString *s;
2763     if (fSkeletons.isValid()) {
2764         for (int32_t i = 0; i < fSkeletons->size(); ++i) {
2765             if ((s = (UnicodeString *)fSkeletons->elementAt(i)) != nullptr) {
2766                 delete s;
2767             }
2768         }
2769     }
2770 }
2771 
DTRedundantEnumeration()2772 DTRedundantEnumeration::DTRedundantEnumeration() : pos(0), fPatterns(nullptr) {
2773 }
2774 
2775 void
add(const UnicodeString & pattern,UErrorCode & status)2776 DTRedundantEnumeration::add(const UnicodeString& pattern, UErrorCode& status) {
2777     if (U_FAILURE(status)) { return; }
2778     if (fPatterns.isNull())  {
2779         fPatterns.adoptInsteadAndCheckErrorCode(new UVector(status), status);
2780         if (U_FAILURE(status)) {
2781             return;
2782        }
2783     }
2784     LocalPointer<UnicodeString> newElem(new UnicodeString(pattern), status);
2785     if (U_FAILURE(status)) {
2786         return;
2787     }
2788     fPatterns->addElement(newElem.getAlias(), status);
2789     if (U_FAILURE(status)) {
2790         fPatterns.adoptInstead(nullptr);
2791         return;
2792     }
2793     newElem.orphan(); // fPatterns now owns the string.
2794 }
2795 
2796 const UnicodeString*
snext(UErrorCode & status)2797 DTRedundantEnumeration::snext(UErrorCode& status) {
2798     if (U_SUCCESS(status) && fPatterns.isValid() && pos < fPatterns->size()) {
2799         return (const UnicodeString*)fPatterns->elementAt(pos++);
2800     }
2801     return nullptr;
2802 }
2803 
2804 void
reset(UErrorCode &)2805 DTRedundantEnumeration::reset(UErrorCode& /*status*/) {
2806     pos=0;
2807 }
2808 
2809 int32_t
count(UErrorCode &) const2810 DTRedundantEnumeration::count(UErrorCode& /*status*/) const {
2811     return (fPatterns.isNull()) ? 0 : fPatterns->size();
2812 }
2813 
2814 UBool
isCanonicalItem(const UnicodeString & item) const2815 DTRedundantEnumeration::isCanonicalItem(const UnicodeString& item) const {
2816     if ( item.length() != 1 ) {
2817         return FALSE;
2818     }
2819     for (int32_t i=0; i<UDATPG_FIELD_COUNT; ++i) {
2820         if (item.charAt(0)==Canonical_Items[i]) {
2821             return TRUE;
2822         }
2823     }
2824     return FALSE;
2825 }
2826 
~DTRedundantEnumeration()2827 DTRedundantEnumeration::~DTRedundantEnumeration() {
2828     UnicodeString *s;
2829     if (fPatterns.isValid()) {
2830         for (int32_t i = 0; i < fPatterns->size(); ++i) {
2831             if ((s = (UnicodeString *)fPatterns->elementAt(i)) != nullptr) {
2832                 delete s;
2833             }
2834         }
2835     }
2836 }
2837 
2838 U_NAMESPACE_END
2839 
2840 
2841 #endif /* #if !UCONFIG_NO_FORMATTING */
2842 
2843 //eof
2844