]>
Commit | Line | Data |
---|---|---|
1 | // © 2016 and later: Unicode, Inc. and others. | |
2 | // License & terms of use: http://www.unicode.org/copyright.html | |
3 | /******************************************************************** | |
4 | * COPYRIGHT: | |
5 | * Copyright (c) 1997-2015, International Business Machines Corporation and | |
6 | * others. All Rights Reserved. | |
7 | ******************************************************************** | |
8 | * | |
9 | * File MSGFMT.CPP | |
10 | * | |
11 | * Modification History: | |
12 | * | |
13 | * Date Name Description | |
14 | * 02/19/97 aliu Converted from java. | |
15 | * 03/20/97 helena Finished first cut of implementation. | |
16 | * 04/10/97 aliu Made to work on AIX. Added stoi to replace wtoi. | |
17 | * 06/11/97 helena Fixed addPattern to take the pattern correctly. | |
18 | * 06/17/97 helena Fixed the getPattern to return the correct pattern. | |
19 | * 07/09/97 helena Made ParsePosition into a class. | |
20 | * 02/22/99 stephen Removed character literals for EBCDIC safety | |
21 | * 11/01/09 kirtig Added SelectFormat | |
22 | ********************************************************************/ | |
23 | ||
24 | #include "unicode/utypes.h" | |
25 | ||
26 | #if !UCONFIG_NO_FORMATTING | |
27 | ||
28 | #include "unicode/appendable.h" | |
29 | #include "unicode/choicfmt.h" | |
30 | #include "unicode/datefmt.h" | |
31 | #include "unicode/decimfmt.h" | |
32 | #include "unicode/localpointer.h" | |
33 | #include "unicode/msgfmt.h" | |
34 | #include "unicode/numberformatter.h" | |
35 | #include "unicode/plurfmt.h" | |
36 | #include "unicode/rbnf.h" | |
37 | #include "unicode/selfmt.h" | |
38 | #include "unicode/smpdtfmt.h" | |
39 | #include "unicode/umsg.h" | |
40 | #include "unicode/ustring.h" | |
41 | #include "cmemory.h" | |
42 | #include "patternprops.h" | |
43 | #include "messageimpl.h" | |
44 | #include "msgfmt_impl.h" | |
45 | #include "plurrule_impl.h" | |
46 | #include "uassert.h" | |
47 | #include "uelement.h" | |
48 | #include "uhash.h" | |
49 | #include "ustrfmt.h" | |
50 | #include "util.h" | |
51 | #include "uvector.h" | |
52 | #include "number_decimalquantity.h" | |
53 | ||
54 | // ***************************************************************************** | |
55 | // class MessageFormat | |
56 | // ***************************************************************************** | |
57 | ||
58 | #define SINGLE_QUOTE ((UChar)0x0027) | |
59 | #define COMMA ((UChar)0x002C) | |
60 | #define LEFT_CURLY_BRACE ((UChar)0x007B) | |
61 | #define RIGHT_CURLY_BRACE ((UChar)0x007D) | |
62 | ||
63 | //--------------------------------------- | |
64 | // static data | |
65 | ||
66 | static const UChar ID_NUMBER[] = { | |
67 | 0x6E, 0x75, 0x6D, 0x62, 0x65, 0x72, 0 /* "number" */ | |
68 | }; | |
69 | static const UChar ID_DATE[] = { | |
70 | 0x64, 0x61, 0x74, 0x65, 0 /* "date" */ | |
71 | }; | |
72 | static const UChar ID_TIME[] = { | |
73 | 0x74, 0x69, 0x6D, 0x65, 0 /* "time" */ | |
74 | }; | |
75 | static const UChar ID_SPELLOUT[] = { | |
76 | 0x73, 0x70, 0x65, 0x6c, 0x6c, 0x6f, 0x75, 0x74, 0 /* "spellout" */ | |
77 | }; | |
78 | static const UChar ID_ORDINAL[] = { | |
79 | 0x6f, 0x72, 0x64, 0x69, 0x6e, 0x61, 0x6c, 0 /* "ordinal" */ | |
80 | }; | |
81 | static const UChar ID_DURATION[] = { | |
82 | 0x64, 0x75, 0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0 /* "duration" */ | |
83 | }; | |
84 | ||
85 | // MessageFormat Type List Number, Date, Time or Choice | |
86 | static const UChar * const TYPE_IDS[] = { | |
87 | ID_NUMBER, | |
88 | ID_DATE, | |
89 | ID_TIME, | |
90 | ID_SPELLOUT, | |
91 | ID_ORDINAL, | |
92 | ID_DURATION, | |
93 | NULL, | |
94 | }; | |
95 | ||
96 | static const UChar ID_EMPTY[] = { | |
97 | 0 /* empty string, used for default so that null can mark end of list */ | |
98 | }; | |
99 | static const UChar ID_CURRENCY[] = { | |
100 | 0x63, 0x75, 0x72, 0x72, 0x65, 0x6E, 0x63, 0x79, 0 /* "currency" */ | |
101 | }; | |
102 | static const UChar ID_PERCENT[] = { | |
103 | 0x70, 0x65, 0x72, 0x63, 0x65, 0x6E, 0x74, 0 /* "percent" */ | |
104 | }; | |
105 | static const UChar ID_INTEGER[] = { | |
106 | 0x69, 0x6E, 0x74, 0x65, 0x67, 0x65, 0x72, 0 /* "integer" */ | |
107 | }; | |
108 | ||
109 | // NumberFormat modifier list, default, currency, percent or integer | |
110 | static const UChar * const NUMBER_STYLE_IDS[] = { | |
111 | ID_EMPTY, | |
112 | ID_CURRENCY, | |
113 | ID_PERCENT, | |
114 | ID_INTEGER, | |
115 | NULL, | |
116 | }; | |
117 | ||
118 | static const UChar ID_SHORT[] = { | |
119 | 0x73, 0x68, 0x6F, 0x72, 0x74, 0 /* "short" */ | |
120 | }; | |
121 | static const UChar ID_MEDIUM[] = { | |
122 | 0x6D, 0x65, 0x64, 0x69, 0x75, 0x6D, 0 /* "medium" */ | |
123 | }; | |
124 | static const UChar ID_LONG[] = { | |
125 | 0x6C, 0x6F, 0x6E, 0x67, 0 /* "long" */ | |
126 | }; | |
127 | static const UChar ID_FULL[] = { | |
128 | 0x66, 0x75, 0x6C, 0x6C, 0 /* "full" */ | |
129 | }; | |
130 | ||
131 | // DateFormat modifier list, default, short, medium, long or full | |
132 | static const UChar * const DATE_STYLE_IDS[] = { | |
133 | ID_EMPTY, | |
134 | ID_SHORT, | |
135 | ID_MEDIUM, | |
136 | ID_LONG, | |
137 | ID_FULL, | |
138 | NULL, | |
139 | }; | |
140 | ||
141 | static const icu::DateFormat::EStyle DATE_STYLES[] = { | |
142 | icu::DateFormat::kDefault, | |
143 | icu::DateFormat::kShort, | |
144 | icu::DateFormat::kMedium, | |
145 | icu::DateFormat::kLong, | |
146 | icu::DateFormat::kFull, | |
147 | }; | |
148 | ||
149 | static const int32_t DEFAULT_INITIAL_CAPACITY = 10; | |
150 | ||
151 | static const UChar NULL_STRING[] = { | |
152 | 0x6E, 0x75, 0x6C, 0x6C, 0 // "null" | |
153 | }; | |
154 | ||
155 | static const UChar OTHER_STRING[] = { | |
156 | 0x6F, 0x74, 0x68, 0x65, 0x72, 0 // "other" | |
157 | }; | |
158 | ||
159 | U_CDECL_BEGIN | |
160 | static UBool U_CALLCONV equalFormatsForHash(const UHashTok key1, | |
161 | const UHashTok key2) { | |
162 | return icu::MessageFormat::equalFormats(key1.pointer, key2.pointer); | |
163 | } | |
164 | ||
165 | U_CDECL_END | |
166 | ||
167 | U_NAMESPACE_BEGIN | |
168 | ||
169 | // ------------------------------------- | |
170 | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(MessageFormat) | |
171 | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(FormatNameEnumeration) | |
172 | ||
173 | //-------------------------------------------------------------------- | |
174 | ||
175 | /** | |
176 | * Convert an integer value to a string and append the result to | |
177 | * the given UnicodeString. | |
178 | */ | |
179 | static UnicodeString& itos(int32_t i, UnicodeString& appendTo) { | |
180 | UChar temp[16]; | |
181 | uprv_itou(temp,16,i,10,0); // 10 == radix | |
182 | appendTo.append(temp, -1); | |
183 | return appendTo; | |
184 | } | |
185 | ||
186 | ||
187 | // AppendableWrapper: encapsulates the result of formatting, keeping track | |
188 | // of the string and its length. | |
189 | class AppendableWrapper : public UMemory { | |
190 | public: | |
191 | AppendableWrapper(Appendable& appendable) : app(appendable), len(0) { | |
192 | } | |
193 | void append(const UnicodeString& s) { | |
194 | app.appendString(s.getBuffer(), s.length()); | |
195 | len += s.length(); | |
196 | } | |
197 | void append(const UChar* s, const int32_t sLength) { | |
198 | app.appendString(s, sLength); | |
199 | len += sLength; | |
200 | } | |
201 | void append(const UnicodeString& s, int32_t start, int32_t length) { | |
202 | append(s.tempSubString(start, length)); | |
203 | } | |
204 | void formatAndAppend(const Format* formatter, const Formattable& arg, UErrorCode& ec) { | |
205 | UnicodeString s; | |
206 | formatter->format(arg, s, ec); | |
207 | if (U_SUCCESS(ec)) { | |
208 | append(s); | |
209 | } | |
210 | } | |
211 | void formatAndAppend(const Format* formatter, const Formattable& arg, | |
212 | const UnicodeString &argString, UErrorCode& ec) { | |
213 | if (!argString.isEmpty()) { | |
214 | if (U_SUCCESS(ec)) { | |
215 | append(argString); | |
216 | } | |
217 | } else { | |
218 | formatAndAppend(formatter, arg, ec); | |
219 | } | |
220 | } | |
221 | int32_t length() { | |
222 | return len; | |
223 | } | |
224 | private: | |
225 | Appendable& app; | |
226 | int32_t len; | |
227 | }; | |
228 | ||
229 | ||
230 | // ------------------------------------- | |
231 | // Creates a MessageFormat instance based on the pattern. | |
232 | ||
233 | MessageFormat::MessageFormat(const UnicodeString& pattern, | |
234 | UErrorCode& success) | |
235 | : fLocale(Locale::getDefault()), // Uses the default locale | |
236 | msgPattern(success), | |
237 | formatAliases(NULL), | |
238 | formatAliasesCapacity(0), | |
239 | argTypes(NULL), | |
240 | argTypeCount(0), | |
241 | argTypeCapacity(0), | |
242 | hasArgTypeConflicts(FALSE), | |
243 | defaultNumberFormat(NULL), | |
244 | defaultDateFormat(NULL), | |
245 | cachedFormatters(NULL), | |
246 | customFormatArgStarts(NULL), | |
247 | pluralProvider(*this, UPLURAL_TYPE_CARDINAL), | |
248 | ordinalProvider(*this, UPLURAL_TYPE_ORDINAL) | |
249 | { | |
250 | setLocaleIDs(fLocale.getName(), fLocale.getName()); | |
251 | applyPattern(pattern, success); | |
252 | } | |
253 | ||
254 | MessageFormat::MessageFormat(const UnicodeString& pattern, | |
255 | const Locale& newLocale, | |
256 | UErrorCode& success) | |
257 | : fLocale(newLocale), | |
258 | msgPattern(success), | |
259 | formatAliases(NULL), | |
260 | formatAliasesCapacity(0), | |
261 | argTypes(NULL), | |
262 | argTypeCount(0), | |
263 | argTypeCapacity(0), | |
264 | hasArgTypeConflicts(FALSE), | |
265 | defaultNumberFormat(NULL), | |
266 | defaultDateFormat(NULL), | |
267 | cachedFormatters(NULL), | |
268 | customFormatArgStarts(NULL), | |
269 | pluralProvider(*this, UPLURAL_TYPE_CARDINAL), | |
270 | ordinalProvider(*this, UPLURAL_TYPE_ORDINAL) | |
271 | { | |
272 | setLocaleIDs(fLocale.getName(), fLocale.getName()); | |
273 | applyPattern(pattern, success); | |
274 | } | |
275 | ||
276 | MessageFormat::MessageFormat(const UnicodeString& pattern, | |
277 | const Locale& newLocale, | |
278 | UParseError& parseError, | |
279 | UErrorCode& success) | |
280 | : fLocale(newLocale), | |
281 | msgPattern(success), | |
282 | formatAliases(NULL), | |
283 | formatAliasesCapacity(0), | |
284 | argTypes(NULL), | |
285 | argTypeCount(0), | |
286 | argTypeCapacity(0), | |
287 | hasArgTypeConflicts(FALSE), | |
288 | defaultNumberFormat(NULL), | |
289 | defaultDateFormat(NULL), | |
290 | cachedFormatters(NULL), | |
291 | customFormatArgStarts(NULL), | |
292 | pluralProvider(*this, UPLURAL_TYPE_CARDINAL), | |
293 | ordinalProvider(*this, UPLURAL_TYPE_ORDINAL) | |
294 | { | |
295 | setLocaleIDs(fLocale.getName(), fLocale.getName()); | |
296 | applyPattern(pattern, parseError, success); | |
297 | } | |
298 | ||
299 | MessageFormat::MessageFormat(const MessageFormat& that) | |
300 | : | |
301 | Format(that), | |
302 | fLocale(that.fLocale), | |
303 | msgPattern(that.msgPattern), | |
304 | formatAliases(NULL), | |
305 | formatAliasesCapacity(0), | |
306 | argTypes(NULL), | |
307 | argTypeCount(0), | |
308 | argTypeCapacity(0), | |
309 | hasArgTypeConflicts(that.hasArgTypeConflicts), | |
310 | defaultNumberFormat(NULL), | |
311 | defaultDateFormat(NULL), | |
312 | cachedFormatters(NULL), | |
313 | customFormatArgStarts(NULL), | |
314 | pluralProvider(*this, UPLURAL_TYPE_CARDINAL), | |
315 | ordinalProvider(*this, UPLURAL_TYPE_ORDINAL) | |
316 | { | |
317 | // This will take care of creating the hash tables (since they are NULL). | |
318 | UErrorCode ec = U_ZERO_ERROR; | |
319 | copyObjects(that, ec); | |
320 | if (U_FAILURE(ec)) { | |
321 | resetPattern(); | |
322 | } | |
323 | } | |
324 | ||
325 | MessageFormat::~MessageFormat() | |
326 | { | |
327 | uhash_close(cachedFormatters); | |
328 | uhash_close(customFormatArgStarts); | |
329 | ||
330 | uprv_free(argTypes); | |
331 | uprv_free(formatAliases); | |
332 | delete defaultNumberFormat; | |
333 | delete defaultDateFormat; | |
334 | } | |
335 | ||
336 | //-------------------------------------------------------------------- | |
337 | // Variable-size array management | |
338 | ||
339 | /** | |
340 | * Allocate argTypes[] to at least the given capacity and return | |
341 | * TRUE if successful. If not, leave argTypes[] unchanged. | |
342 | * | |
343 | * If argTypes is NULL, allocate it. If it is not NULL, enlarge it | |
344 | * if necessary to be at least as large as specified. | |
345 | */ | |
346 | UBool MessageFormat::allocateArgTypes(int32_t capacity, UErrorCode& status) { | |
347 | if (U_FAILURE(status)) { | |
348 | return FALSE; | |
349 | } | |
350 | if (argTypeCapacity >= capacity) { | |
351 | return TRUE; | |
352 | } | |
353 | if (capacity < DEFAULT_INITIAL_CAPACITY) { | |
354 | capacity = DEFAULT_INITIAL_CAPACITY; | |
355 | } else if (capacity < 2*argTypeCapacity) { | |
356 | capacity = 2*argTypeCapacity; | |
357 | } | |
358 | Formattable::Type* a = (Formattable::Type*) | |
359 | uprv_realloc(argTypes, sizeof(*argTypes) * capacity); | |
360 | if (a == NULL) { | |
361 | status = U_MEMORY_ALLOCATION_ERROR; | |
362 | return FALSE; | |
363 | } | |
364 | argTypes = a; | |
365 | argTypeCapacity = capacity; | |
366 | return TRUE; | |
367 | } | |
368 | ||
369 | // ------------------------------------- | |
370 | // assignment operator | |
371 | ||
372 | const MessageFormat& | |
373 | MessageFormat::operator=(const MessageFormat& that) | |
374 | { | |
375 | if (this != &that) { | |
376 | // Calls the super class for assignment first. | |
377 | Format::operator=(that); | |
378 | ||
379 | setLocale(that.fLocale); | |
380 | msgPattern = that.msgPattern; | |
381 | hasArgTypeConflicts = that.hasArgTypeConflicts; | |
382 | ||
383 | UErrorCode ec = U_ZERO_ERROR; | |
384 | copyObjects(that, ec); | |
385 | if (U_FAILURE(ec)) { | |
386 | resetPattern(); | |
387 | } | |
388 | } | |
389 | return *this; | |
390 | } | |
391 | ||
392 | UBool | |
393 | MessageFormat::operator==(const Format& rhs) const | |
394 | { | |
395 | if (this == &rhs) return TRUE; | |
396 | ||
397 | MessageFormat& that = (MessageFormat&)rhs; | |
398 | ||
399 | // Check class ID before checking MessageFormat members | |
400 | if (!Format::operator==(rhs) || | |
401 | msgPattern != that.msgPattern || | |
402 | fLocale != that.fLocale) { | |
403 | return FALSE; | |
404 | } | |
405 | ||
406 | // Compare hashtables. | |
407 | if ((customFormatArgStarts == NULL) != (that.customFormatArgStarts == NULL)) { | |
408 | return FALSE; | |
409 | } | |
410 | if (customFormatArgStarts == NULL) { | |
411 | return TRUE; | |
412 | } | |
413 | ||
414 | UErrorCode ec = U_ZERO_ERROR; | |
415 | const int32_t count = uhash_count(customFormatArgStarts); | |
416 | const int32_t rhs_count = uhash_count(that.customFormatArgStarts); | |
417 | if (count != rhs_count) { | |
418 | return FALSE; | |
419 | } | |
420 | int32_t idx = 0, rhs_idx = 0, pos = UHASH_FIRST, rhs_pos = UHASH_FIRST; | |
421 | for (; idx < count && rhs_idx < rhs_count && U_SUCCESS(ec); ++idx, ++rhs_idx) { | |
422 | const UHashElement* cur = uhash_nextElement(customFormatArgStarts, &pos); | |
423 | const UHashElement* rhs_cur = uhash_nextElement(that.customFormatArgStarts, &rhs_pos); | |
424 | if (cur->key.integer != rhs_cur->key.integer) { | |
425 | return FALSE; | |
426 | } | |
427 | const Format* format = (const Format*)uhash_iget(cachedFormatters, cur->key.integer); | |
428 | const Format* rhs_format = (const Format*)uhash_iget(that.cachedFormatters, rhs_cur->key.integer); | |
429 | if (*format != *rhs_format) { | |
430 | return FALSE; | |
431 | } | |
432 | } | |
433 | return TRUE; | |
434 | } | |
435 | ||
436 | // ------------------------------------- | |
437 | // Creates a copy of this MessageFormat, the caller owns the copy. | |
438 | ||
439 | Format* | |
440 | MessageFormat::clone() const | |
441 | { | |
442 | return new MessageFormat(*this); | |
443 | } | |
444 | ||
445 | // ------------------------------------- | |
446 | // Sets the locale of this MessageFormat object to theLocale. | |
447 | ||
448 | void | |
449 | MessageFormat::setLocale(const Locale& theLocale) | |
450 | { | |
451 | if (fLocale != theLocale) { | |
452 | delete defaultNumberFormat; | |
453 | defaultNumberFormat = NULL; | |
454 | delete defaultDateFormat; | |
455 | defaultDateFormat = NULL; | |
456 | fLocale = theLocale; | |
457 | setLocaleIDs(fLocale.getName(), fLocale.getName()); | |
458 | pluralProvider.reset(); | |
459 | ordinalProvider.reset(); | |
460 | } | |
461 | } | |
462 | ||
463 | // ------------------------------------- | |
464 | // Gets the locale of this MessageFormat object. | |
465 | ||
466 | const Locale& | |
467 | MessageFormat::getLocale() const | |
468 | { | |
469 | return fLocale; | |
470 | } | |
471 | ||
472 | void | |
473 | MessageFormat::applyPattern(const UnicodeString& newPattern, | |
474 | UErrorCode& status) | |
475 | { | |
476 | UParseError parseError; | |
477 | applyPattern(newPattern,parseError,status); | |
478 | } | |
479 | ||
480 | ||
481 | // ------------------------------------- | |
482 | // Applies the new pattern and returns an error if the pattern | |
483 | // is not correct. | |
484 | void | |
485 | MessageFormat::applyPattern(const UnicodeString& pattern, | |
486 | UParseError& parseError, | |
487 | UErrorCode& ec) | |
488 | { | |
489 | if(U_FAILURE(ec)) { | |
490 | return; | |
491 | } | |
492 | msgPattern.parse(pattern, &parseError, ec); | |
493 | cacheExplicitFormats(ec); | |
494 | ||
495 | if (U_FAILURE(ec)) { | |
496 | resetPattern(); | |
497 | } | |
498 | } | |
499 | ||
500 | void MessageFormat::resetPattern() { | |
501 | msgPattern.clear(); | |
502 | uhash_close(cachedFormatters); | |
503 | cachedFormatters = NULL; | |
504 | uhash_close(customFormatArgStarts); | |
505 | customFormatArgStarts = NULL; | |
506 | argTypeCount = 0; | |
507 | hasArgTypeConflicts = FALSE; | |
508 | } | |
509 | ||
510 | void | |
511 | MessageFormat::applyPattern(const UnicodeString& pattern, | |
512 | UMessagePatternApostropheMode aposMode, | |
513 | UParseError* parseError, | |
514 | UErrorCode& status) { | |
515 | if (aposMode != msgPattern.getApostropheMode()) { | |
516 | msgPattern.clearPatternAndSetApostropheMode(aposMode); | |
517 | } | |
518 | applyPattern(pattern, *parseError, status); | |
519 | } | |
520 | ||
521 | // ------------------------------------- | |
522 | // Converts this MessageFormat instance to a pattern. | |
523 | ||
524 | UnicodeString& | |
525 | MessageFormat::toPattern(UnicodeString& appendTo) const { | |
526 | if ((customFormatArgStarts != NULL && 0 != uhash_count(customFormatArgStarts)) || | |
527 | 0 == msgPattern.countParts() | |
528 | ) { | |
529 | appendTo.setToBogus(); | |
530 | return appendTo; | |
531 | } | |
532 | return appendTo.append(msgPattern.getPatternString()); | |
533 | } | |
534 | ||
535 | int32_t MessageFormat::nextTopLevelArgStart(int32_t partIndex) const { | |
536 | if (partIndex != 0) { | |
537 | partIndex = msgPattern.getLimitPartIndex(partIndex); | |
538 | } | |
539 | for (;;) { | |
540 | UMessagePatternPartType type = msgPattern.getPartType(++partIndex); | |
541 | if (type == UMSGPAT_PART_TYPE_ARG_START) { | |
542 | return partIndex; | |
543 | } | |
544 | if (type == UMSGPAT_PART_TYPE_MSG_LIMIT) { | |
545 | return -1; | |
546 | } | |
547 | } | |
548 | } | |
549 | ||
550 | void MessageFormat::setArgStartFormat(int32_t argStart, | |
551 | Format* formatter, | |
552 | UErrorCode& status) { | |
553 | if (U_FAILURE(status)) { | |
554 | delete formatter; | |
555 | return; | |
556 | } | |
557 | if (cachedFormatters == NULL) { | |
558 | cachedFormatters=uhash_open(uhash_hashLong, uhash_compareLong, | |
559 | equalFormatsForHash, &status); | |
560 | if (U_FAILURE(status)) { | |
561 | delete formatter; | |
562 | return; | |
563 | } | |
564 | uhash_setValueDeleter(cachedFormatters, uprv_deleteUObject); | |
565 | } | |
566 | if (formatter == NULL) { | |
567 | formatter = new DummyFormat(); | |
568 | } | |
569 | uhash_iput(cachedFormatters, argStart, formatter, &status); | |
570 | } | |
571 | ||
572 | ||
573 | UBool MessageFormat::argNameMatches(int32_t partIndex, const UnicodeString& argName, int32_t argNumber) { | |
574 | const MessagePattern::Part& part = msgPattern.getPart(partIndex); | |
575 | return part.getType() == UMSGPAT_PART_TYPE_ARG_NAME ? | |
576 | msgPattern.partSubstringMatches(part, argName) : | |
577 | part.getValue() == argNumber; // ARG_NUMBER | |
578 | } | |
579 | ||
580 | // Sets a custom formatter for a MessagePattern ARG_START part index. | |
581 | // "Custom" formatters are provided by the user via setFormat() or similar APIs. | |
582 | void MessageFormat::setCustomArgStartFormat(int32_t argStart, | |
583 | Format* formatter, | |
584 | UErrorCode& status) { | |
585 | setArgStartFormat(argStart, formatter, status); | |
586 | if (customFormatArgStarts == NULL) { | |
587 | customFormatArgStarts=uhash_open(uhash_hashLong, uhash_compareLong, | |
588 | NULL, &status); | |
589 | } | |
590 | uhash_iputi(customFormatArgStarts, argStart, 1, &status); | |
591 | } | |
592 | ||
593 | Format* MessageFormat::getCachedFormatter(int32_t argumentNumber) const { | |
594 | if (cachedFormatters == NULL) { | |
595 | return NULL; | |
596 | } | |
597 | void* ptr = uhash_iget(cachedFormatters, argumentNumber); | |
598 | if (ptr != NULL && dynamic_cast<DummyFormat*>((Format*)ptr) == NULL) { | |
599 | return (Format*) ptr; | |
600 | } else { | |
601 | // Not cached, or a DummyFormat representing setFormat(NULL). | |
602 | return NULL; | |
603 | } | |
604 | } | |
605 | ||
606 | // ------------------------------------- | |
607 | // Adopts the new formats array and updates the array count. | |
608 | // This MessageFormat instance owns the new formats. | |
609 | void | |
610 | MessageFormat::adoptFormats(Format** newFormats, | |
611 | int32_t count) { | |
612 | if (newFormats == NULL || count < 0) { | |
613 | return; | |
614 | } | |
615 | // Throw away any cached formatters. | |
616 | if (cachedFormatters != NULL) { | |
617 | uhash_removeAll(cachedFormatters); | |
618 | } | |
619 | if (customFormatArgStarts != NULL) { | |
620 | uhash_removeAll(customFormatArgStarts); | |
621 | } | |
622 | ||
623 | int32_t formatNumber = 0; | |
624 | UErrorCode status = U_ZERO_ERROR; | |
625 | for (int32_t partIndex = 0; | |
626 | formatNumber < count && U_SUCCESS(status) && | |
627 | (partIndex = nextTopLevelArgStart(partIndex)) >= 0;) { | |
628 | setCustomArgStartFormat(partIndex, newFormats[formatNumber], status); | |
629 | ++formatNumber; | |
630 | } | |
631 | // Delete those that didn't get used (if any). | |
632 | for (; formatNumber < count; ++formatNumber) { | |
633 | delete newFormats[formatNumber]; | |
634 | } | |
635 | ||
636 | } | |
637 | ||
638 | // ------------------------------------- | |
639 | // Sets the new formats array and updates the array count. | |
640 | // This MessageFormat instance maks a copy of the new formats. | |
641 | ||
642 | void | |
643 | MessageFormat::setFormats(const Format** newFormats, | |
644 | int32_t count) { | |
645 | if (newFormats == NULL || count < 0) { | |
646 | return; | |
647 | } | |
648 | // Throw away any cached formatters. | |
649 | if (cachedFormatters != NULL) { | |
650 | uhash_removeAll(cachedFormatters); | |
651 | } | |
652 | if (customFormatArgStarts != NULL) { | |
653 | uhash_removeAll(customFormatArgStarts); | |
654 | } | |
655 | ||
656 | UErrorCode status = U_ZERO_ERROR; | |
657 | int32_t formatNumber = 0; | |
658 | for (int32_t partIndex = 0; | |
659 | formatNumber < count && U_SUCCESS(status) && (partIndex = nextTopLevelArgStart(partIndex)) >= 0;) { | |
660 | Format* newFormat = NULL; | |
661 | if (newFormats[formatNumber] != NULL) { | |
662 | newFormat = newFormats[formatNumber]->clone(); | |
663 | if (newFormat == NULL) { | |
664 | status = U_MEMORY_ALLOCATION_ERROR; | |
665 | } | |
666 | } | |
667 | setCustomArgStartFormat(partIndex, newFormat, status); | |
668 | ++formatNumber; | |
669 | } | |
670 | if (U_FAILURE(status)) { | |
671 | resetPattern(); | |
672 | } | |
673 | } | |
674 | ||
675 | // ------------------------------------- | |
676 | // Adopt a single format by format number. | |
677 | // Do nothing if the format number is not less than the array count. | |
678 | ||
679 | void | |
680 | MessageFormat::adoptFormat(int32_t n, Format *newFormat) { | |
681 | LocalPointer<Format> p(newFormat); | |
682 | if (n >= 0) { | |
683 | int32_t formatNumber = 0; | |
684 | for (int32_t partIndex = 0; (partIndex = nextTopLevelArgStart(partIndex)) >= 0;) { | |
685 | if (n == formatNumber) { | |
686 | UErrorCode status = U_ZERO_ERROR; | |
687 | setCustomArgStartFormat(partIndex, p.orphan(), status); | |
688 | return; | |
689 | } | |
690 | ++formatNumber; | |
691 | } | |
692 | } | |
693 | } | |
694 | ||
695 | // ------------------------------------- | |
696 | // Adopt a single format by format name. | |
697 | // Do nothing if there is no match of formatName. | |
698 | void | |
699 | MessageFormat::adoptFormat(const UnicodeString& formatName, | |
700 | Format* formatToAdopt, | |
701 | UErrorCode& status) { | |
702 | LocalPointer<Format> p(formatToAdopt); | |
703 | if (U_FAILURE(status)) { | |
704 | return; | |
705 | } | |
706 | int32_t argNumber = MessagePattern::validateArgumentName(formatName); | |
707 | if (argNumber < UMSGPAT_ARG_NAME_NOT_NUMBER) { | |
708 | status = U_ILLEGAL_ARGUMENT_ERROR; | |
709 | return; | |
710 | } | |
711 | for (int32_t partIndex = 0; | |
712 | (partIndex = nextTopLevelArgStart(partIndex)) >= 0 && U_SUCCESS(status); | |
713 | ) { | |
714 | if (argNameMatches(partIndex + 1, formatName, argNumber)) { | |
715 | Format* f; | |
716 | if (p.isValid()) { | |
717 | f = p.orphan(); | |
718 | } else if (formatToAdopt == NULL) { | |
719 | f = NULL; | |
720 | } else { | |
721 | f = formatToAdopt->clone(); | |
722 | if (f == NULL) { | |
723 | status = U_MEMORY_ALLOCATION_ERROR; | |
724 | return; | |
725 | } | |
726 | } | |
727 | setCustomArgStartFormat(partIndex, f, status); | |
728 | } | |
729 | } | |
730 | } | |
731 | ||
732 | // ------------------------------------- | |
733 | // Set a single format. | |
734 | // Do nothing if the variable is not less than the array count. | |
735 | void | |
736 | MessageFormat::setFormat(int32_t n, const Format& newFormat) { | |
737 | ||
738 | if (n >= 0) { | |
739 | int32_t formatNumber = 0; | |
740 | for (int32_t partIndex = 0; | |
741 | (partIndex = nextTopLevelArgStart(partIndex)) >= 0;) { | |
742 | if (n == formatNumber) { | |
743 | Format* new_format = newFormat.clone(); | |
744 | if (new_format) { | |
745 | UErrorCode status = U_ZERO_ERROR; | |
746 | setCustomArgStartFormat(partIndex, new_format, status); | |
747 | } | |
748 | return; | |
749 | } | |
750 | ++formatNumber; | |
751 | } | |
752 | } | |
753 | } | |
754 | ||
755 | // ------------------------------------- | |
756 | // Get a single format by format name. | |
757 | // Do nothing if the variable is not less than the array count. | |
758 | Format * | |
759 | MessageFormat::getFormat(const UnicodeString& formatName, UErrorCode& status) { | |
760 | if (U_FAILURE(status) || cachedFormatters == NULL) return NULL; | |
761 | ||
762 | int32_t argNumber = MessagePattern::validateArgumentName(formatName); | |
763 | if (argNumber < UMSGPAT_ARG_NAME_NOT_NUMBER) { | |
764 | status = U_ILLEGAL_ARGUMENT_ERROR; | |
765 | return NULL; | |
766 | } | |
767 | for (int32_t partIndex = 0; (partIndex = nextTopLevelArgStart(partIndex)) >= 0;) { | |
768 | if (argNameMatches(partIndex + 1, formatName, argNumber)) { | |
769 | return getCachedFormatter(partIndex); | |
770 | } | |
771 | } | |
772 | return NULL; | |
773 | } | |
774 | ||
775 | // ------------------------------------- | |
776 | // Set a single format by format name | |
777 | // Do nothing if the variable is not less than the array count. | |
778 | void | |
779 | MessageFormat::setFormat(const UnicodeString& formatName, | |
780 | const Format& newFormat, | |
781 | UErrorCode& status) { | |
782 | if (U_FAILURE(status)) return; | |
783 | ||
784 | int32_t argNumber = MessagePattern::validateArgumentName(formatName); | |
785 | if (argNumber < UMSGPAT_ARG_NAME_NOT_NUMBER) { | |
786 | status = U_ILLEGAL_ARGUMENT_ERROR; | |
787 | return; | |
788 | } | |
789 | for (int32_t partIndex = 0; | |
790 | (partIndex = nextTopLevelArgStart(partIndex)) >= 0 && U_SUCCESS(status); | |
791 | ) { | |
792 | if (argNameMatches(partIndex + 1, formatName, argNumber)) { | |
793 | Format* new_format = newFormat.clone(); | |
794 | if (new_format == NULL) { | |
795 | status = U_MEMORY_ALLOCATION_ERROR; | |
796 | return; | |
797 | } | |
798 | setCustomArgStartFormat(partIndex, new_format, status); | |
799 | } | |
800 | } | |
801 | } | |
802 | ||
803 | // ------------------------------------- | |
804 | // Gets the format array. | |
805 | const Format** | |
806 | MessageFormat::getFormats(int32_t& cnt) const | |
807 | { | |
808 | // This old API returns an array (which we hold) of Format* | |
809 | // pointers. The array is valid up to the next call to any | |
810 | // method on this object. We construct and resize an array | |
811 | // on demand that contains aliases to the subformats[i].format | |
812 | // pointers. | |
813 | ||
814 | // Get total required capacity first (it's refreshed on each call). | |
815 | int32_t totalCapacity = 0; | |
816 | for (int32_t partIndex = 0; (partIndex = nextTopLevelArgStart(partIndex)) >= 0; ++totalCapacity) {}; | |
817 | ||
818 | MessageFormat* t = const_cast<MessageFormat*> (this); | |
819 | cnt = 0; | |
820 | if (formatAliases == nullptr) { | |
821 | t->formatAliasesCapacity = totalCapacity; | |
822 | Format** a = (Format**) | |
823 | uprv_malloc(sizeof(Format*) * formatAliasesCapacity); | |
824 | if (a == nullptr) { | |
825 | t->formatAliasesCapacity = 0; | |
826 | return nullptr; | |
827 | } | |
828 | t->formatAliases = a; | |
829 | } else if (totalCapacity > formatAliasesCapacity) { | |
830 | Format** a = (Format**) | |
831 | uprv_realloc(formatAliases, sizeof(Format*) * totalCapacity); | |
832 | if (a == nullptr) { | |
833 | t->formatAliasesCapacity = 0; | |
834 | return nullptr; | |
835 | } | |
836 | t->formatAliases = a; | |
837 | t->formatAliasesCapacity = totalCapacity; | |
838 | } | |
839 | ||
840 | for (int32_t partIndex = 0; (partIndex = nextTopLevelArgStart(partIndex)) >= 0;) { | |
841 | t->formatAliases[cnt++] = getCachedFormatter(partIndex); | |
842 | } | |
843 | ||
844 | return (const Format**)formatAliases; | |
845 | } | |
846 | ||
847 | ||
848 | UnicodeString MessageFormat::getArgName(int32_t partIndex) { | |
849 | const MessagePattern::Part& part = msgPattern.getPart(partIndex); | |
850 | return msgPattern.getSubstring(part); | |
851 | } | |
852 | ||
853 | StringEnumeration* | |
854 | MessageFormat::getFormatNames(UErrorCode& status) { | |
855 | if (U_FAILURE(status)) return NULL; | |
856 | ||
857 | UVector *fFormatNames = new UVector(status); | |
858 | if (U_FAILURE(status)) { | |
859 | status = U_MEMORY_ALLOCATION_ERROR; | |
860 | return NULL; | |
861 | } | |
862 | fFormatNames->setDeleter(uprv_deleteUObject); | |
863 | ||
864 | for (int32_t partIndex = 0; (partIndex = nextTopLevelArgStart(partIndex)) >= 0;) { | |
865 | fFormatNames->addElement(new UnicodeString(getArgName(partIndex + 1)), status); | |
866 | } | |
867 | ||
868 | StringEnumeration* nameEnumerator = new FormatNameEnumeration(fFormatNames, status); | |
869 | return nameEnumerator; | |
870 | } | |
871 | ||
872 | // ------------------------------------- | |
873 | // Formats the source Formattable array and copy into the result buffer. | |
874 | // Ignore the FieldPosition result for error checking. | |
875 | ||
876 | UnicodeString& | |
877 | MessageFormat::format(const Formattable* source, | |
878 | int32_t cnt, | |
879 | UnicodeString& appendTo, | |
880 | FieldPosition& ignore, | |
881 | UErrorCode& success) const | |
882 | { | |
883 | return format(source, NULL, cnt, appendTo, &ignore, success); | |
884 | } | |
885 | ||
886 | // ------------------------------------- | |
887 | // Internally creates a MessageFormat instance based on the | |
888 | // pattern and formats the arguments Formattable array and | |
889 | // copy into the appendTo buffer. | |
890 | ||
891 | UnicodeString& | |
892 | MessageFormat::format( const UnicodeString& pattern, | |
893 | const Formattable* arguments, | |
894 | int32_t cnt, | |
895 | UnicodeString& appendTo, | |
896 | UErrorCode& success) | |
897 | { | |
898 | MessageFormat temp(pattern, success); | |
899 | return temp.format(arguments, NULL, cnt, appendTo, NULL, success); | |
900 | } | |
901 | ||
902 | // ------------------------------------- | |
903 | // Formats the source Formattable object and copy into the | |
904 | // appendTo buffer. The Formattable object must be an array | |
905 | // of Formattable instances, returns error otherwise. | |
906 | ||
907 | UnicodeString& | |
908 | MessageFormat::format(const Formattable& source, | |
909 | UnicodeString& appendTo, | |
910 | FieldPosition& ignore, | |
911 | UErrorCode& success) const | |
912 | { | |
913 | if (U_FAILURE(success)) | |
914 | return appendTo; | |
915 | if (source.getType() != Formattable::kArray) { | |
916 | success = U_ILLEGAL_ARGUMENT_ERROR; | |
917 | return appendTo; | |
918 | } | |
919 | int32_t cnt; | |
920 | const Formattable* tmpPtr = source.getArray(cnt); | |
921 | return format(tmpPtr, NULL, cnt, appendTo, &ignore, success); | |
922 | } | |
923 | ||
924 | UnicodeString& | |
925 | MessageFormat::format(const UnicodeString* argumentNames, | |
926 | const Formattable* arguments, | |
927 | int32_t count, | |
928 | UnicodeString& appendTo, | |
929 | UErrorCode& success) const { | |
930 | return format(arguments, argumentNames, count, appendTo, NULL, success); | |
931 | } | |
932 | ||
933 | // Does linear search to find the match for an ArgName. | |
934 | const Formattable* MessageFormat::getArgFromListByName(const Formattable* arguments, | |
935 | const UnicodeString *argumentNames, | |
936 | int32_t cnt, UnicodeString& name) const { | |
937 | for (int32_t i = 0; i < cnt; ++i) { | |
938 | if (0 == argumentNames[i].compare(name)) { | |
939 | return arguments + i; | |
940 | } | |
941 | } | |
942 | return NULL; | |
943 | } | |
944 | ||
945 | ||
946 | UnicodeString& | |
947 | MessageFormat::format(const Formattable* arguments, | |
948 | const UnicodeString *argumentNames, | |
949 | int32_t cnt, | |
950 | UnicodeString& appendTo, | |
951 | FieldPosition* pos, | |
952 | UErrorCode& status) const { | |
953 | if (U_FAILURE(status)) { | |
954 | return appendTo; | |
955 | } | |
956 | ||
957 | UnicodeStringAppendable usapp(appendTo); | |
958 | AppendableWrapper app(usapp); | |
959 | format(0, NULL, arguments, argumentNames, cnt, app, pos, status); | |
960 | return appendTo; | |
961 | } | |
962 | ||
963 | namespace { | |
964 | ||
965 | /** | |
966 | * Mutable input/output values for the PluralSelectorProvider. | |
967 | * Separate so that it is possible to make MessageFormat Freezable. | |
968 | */ | |
969 | class PluralSelectorContext { | |
970 | public: | |
971 | PluralSelectorContext(int32_t start, const UnicodeString &name, | |
972 | const Formattable &num, double off, UErrorCode &errorCode) | |
973 | : startIndex(start), argName(name), offset(off), | |
974 | numberArgIndex(-1), formatter(NULL), forReplaceNumber(FALSE) { | |
975 | // number needs to be set even when select() is not called. | |
976 | // Keep it as a Number/Formattable: | |
977 | // For format() methods, and to preserve information (e.g., BigDecimal). | |
978 | if(off == 0) { | |
979 | number = num; | |
980 | } else { | |
981 | number = num.getDouble(errorCode) - off; | |
982 | } | |
983 | } | |
984 | ||
985 | // Input values for plural selection with decimals. | |
986 | int32_t startIndex; | |
987 | const UnicodeString &argName; | |
988 | /** argument number - plural offset */ | |
989 | Formattable number; | |
990 | double offset; | |
991 | // Output values for plural selection with decimals. | |
992 | /** -1 if REPLACE_NUMBER, 0 arg not found, >0 ARG_START index */ | |
993 | int32_t numberArgIndex; | |
994 | const Format *formatter; | |
995 | /** formatted argument number - plural offset */ | |
996 | UnicodeString numberString; | |
997 | /** TRUE if number-offset was formatted with the stock number formatter */ | |
998 | UBool forReplaceNumber; | |
999 | }; | |
1000 | ||
1001 | } // namespace | |
1002 | ||
1003 | // if argumentNames is NULL, this means arguments is a numeric array. | |
1004 | // arguments can not be NULL. | |
1005 | // We use const void *plNumber rather than const PluralSelectorContext *pluralNumber | |
1006 | // so that we need not declare the PluralSelectorContext in the public header file. | |
1007 | void MessageFormat::format(int32_t msgStart, const void *plNumber, | |
1008 | const Formattable* arguments, | |
1009 | const UnicodeString *argumentNames, | |
1010 | int32_t cnt, | |
1011 | AppendableWrapper& appendTo, | |
1012 | FieldPosition* ignore, | |
1013 | UErrorCode& success) const { | |
1014 | if (U_FAILURE(success)) { | |
1015 | return; | |
1016 | } | |
1017 | ||
1018 | const UnicodeString& msgString = msgPattern.getPatternString(); | |
1019 | int32_t prevIndex = msgPattern.getPart(msgStart).getLimit(); | |
1020 | for (int32_t i = msgStart + 1; U_SUCCESS(success) ; ++i) { | |
1021 | const MessagePattern::Part* part = &msgPattern.getPart(i); | |
1022 | const UMessagePatternPartType type = part->getType(); | |
1023 | int32_t index = part->getIndex(); | |
1024 | appendTo.append(msgString, prevIndex, index - prevIndex); | |
1025 | if (type == UMSGPAT_PART_TYPE_MSG_LIMIT) { | |
1026 | return; | |
1027 | } | |
1028 | prevIndex = part->getLimit(); | |
1029 | if (type == UMSGPAT_PART_TYPE_REPLACE_NUMBER) { | |
1030 | const PluralSelectorContext &pluralNumber = | |
1031 | *static_cast<const PluralSelectorContext *>(plNumber); | |
1032 | if(pluralNumber.forReplaceNumber) { | |
1033 | // number-offset was already formatted. | |
1034 | appendTo.formatAndAppend(pluralNumber.formatter, | |
1035 | pluralNumber.number, pluralNumber.numberString, success); | |
1036 | } else { | |
1037 | const NumberFormat* nf = getDefaultNumberFormat(success); | |
1038 | appendTo.formatAndAppend(nf, pluralNumber.number, success); | |
1039 | } | |
1040 | continue; | |
1041 | } | |
1042 | if (type != UMSGPAT_PART_TYPE_ARG_START) { | |
1043 | continue; | |
1044 | } | |
1045 | int32_t argLimit = msgPattern.getLimitPartIndex(i); | |
1046 | UMessagePatternArgType argType = part->getArgType(); | |
1047 | part = &msgPattern.getPart(++i); | |
1048 | const Formattable* arg; | |
1049 | UBool noArg = FALSE; | |
1050 | UnicodeString argName = msgPattern.getSubstring(*part); | |
1051 | if (argumentNames == NULL) { | |
1052 | int32_t argNumber = part->getValue(); // ARG_NUMBER | |
1053 | if (0 <= argNumber && argNumber < cnt) { | |
1054 | arg = arguments + argNumber; | |
1055 | } else { | |
1056 | arg = NULL; | |
1057 | noArg = TRUE; | |
1058 | } | |
1059 | } else { | |
1060 | arg = getArgFromListByName(arguments, argumentNames, cnt, argName); | |
1061 | if (arg == NULL) { | |
1062 | noArg = TRUE; | |
1063 | } | |
1064 | } | |
1065 | ++i; | |
1066 | int32_t prevDestLength = appendTo.length(); | |
1067 | const Format* formatter = NULL; | |
1068 | if (noArg) { | |
1069 | appendTo.append( | |
1070 | UnicodeString(LEFT_CURLY_BRACE).append(argName).append(RIGHT_CURLY_BRACE)); | |
1071 | } else if (arg == NULL) { | |
1072 | appendTo.append(NULL_STRING, 4); | |
1073 | } else if(plNumber!=NULL && | |
1074 | static_cast<const PluralSelectorContext *>(plNumber)->numberArgIndex==(i-2)) { | |
1075 | const PluralSelectorContext &pluralNumber = | |
1076 | *static_cast<const PluralSelectorContext *>(plNumber); | |
1077 | if(pluralNumber.offset == 0) { | |
1078 | // The number was already formatted with this formatter. | |
1079 | appendTo.formatAndAppend(pluralNumber.formatter, pluralNumber.number, | |
1080 | pluralNumber.numberString, success); | |
1081 | } else { | |
1082 | // Do not use the formatted (number-offset) string for a named argument | |
1083 | // that formats the number without subtracting the offset. | |
1084 | appendTo.formatAndAppend(pluralNumber.formatter, *arg, success); | |
1085 | } | |
1086 | } else if ((formatter = getCachedFormatter(i -2)) != 0) { | |
1087 | // Handles all ArgType.SIMPLE, and formatters from setFormat() and its siblings. | |
1088 | if (dynamic_cast<const ChoiceFormat*>(formatter) || | |
1089 | dynamic_cast<const PluralFormat*>(formatter) || | |
1090 | dynamic_cast<const SelectFormat*>(formatter)) { | |
1091 | // We only handle nested formats here if they were provided via | |
1092 | // setFormat() or its siblings. Otherwise they are not cached and instead | |
1093 | // handled below according to argType. | |
1094 | UnicodeString subMsgString; | |
1095 | formatter->format(*arg, subMsgString, success); | |
1096 | if (subMsgString.indexOf(LEFT_CURLY_BRACE) >= 0 || | |
1097 | (subMsgString.indexOf(SINGLE_QUOTE) >= 0 && !MessageImpl::jdkAposMode(msgPattern)) | |
1098 | ) { | |
1099 | MessageFormat subMsgFormat(subMsgString, fLocale, success); | |
1100 | subMsgFormat.format(0, NULL, arguments, argumentNames, cnt, appendTo, ignore, success); | |
1101 | } else { | |
1102 | appendTo.append(subMsgString); | |
1103 | } | |
1104 | } else { | |
1105 | appendTo.formatAndAppend(formatter, *arg, success); | |
1106 | } | |
1107 | } else if (argType == UMSGPAT_ARG_TYPE_NONE || (cachedFormatters && uhash_iget(cachedFormatters, i - 2))) { | |
1108 | // We arrive here if getCachedFormatter returned NULL, but there was actually an element in the hash table. | |
1109 | // This can only happen if the hash table contained a DummyFormat, so the if statement above is a check | |
1110 | // for the hash table containind DummyFormat. | |
1111 | if (arg->isNumeric()) { | |
1112 | const NumberFormat* nf = getDefaultNumberFormat(success); | |
1113 | appendTo.formatAndAppend(nf, *arg, success); | |
1114 | } else if (arg->getType() == Formattable::kDate) { | |
1115 | const DateFormat* df = getDefaultDateFormat(success); | |
1116 | appendTo.formatAndAppend(df, *arg, success); | |
1117 | } else { | |
1118 | appendTo.append(arg->getString(success)); | |
1119 | } | |
1120 | } else if (argType == UMSGPAT_ARG_TYPE_CHOICE) { | |
1121 | if (!arg->isNumeric()) { | |
1122 | success = U_ILLEGAL_ARGUMENT_ERROR; | |
1123 | return; | |
1124 | } | |
1125 | // We must use the Formattable::getDouble() variant with the UErrorCode parameter | |
1126 | // because only this one converts non-double numeric types to double. | |
1127 | const double number = arg->getDouble(success); | |
1128 | int32_t subMsgStart = ChoiceFormat::findSubMessage(msgPattern, i, number); | |
1129 | formatComplexSubMessage(subMsgStart, NULL, arguments, argumentNames, | |
1130 | cnt, appendTo, success); | |
1131 | } else if (UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE(argType)) { | |
1132 | if (!arg->isNumeric()) { | |
1133 | success = U_ILLEGAL_ARGUMENT_ERROR; | |
1134 | return; | |
1135 | } | |
1136 | const PluralSelectorProvider &selector = | |
1137 | argType == UMSGPAT_ARG_TYPE_PLURAL ? pluralProvider : ordinalProvider; | |
1138 | // We must use the Formattable::getDouble() variant with the UErrorCode parameter | |
1139 | // because only this one converts non-double numeric types to double. | |
1140 | double offset = msgPattern.getPluralOffset(i); | |
1141 | PluralSelectorContext context(i, argName, *arg, offset, success); | |
1142 | int32_t subMsgStart = PluralFormat::findSubMessage( | |
1143 | msgPattern, i, selector, &context, arg->getDouble(success), success); | |
1144 | formatComplexSubMessage(subMsgStart, &context, arguments, argumentNames, | |
1145 | cnt, appendTo, success); | |
1146 | } else if (argType == UMSGPAT_ARG_TYPE_SELECT) { | |
1147 | int32_t subMsgStart = SelectFormat::findSubMessage(msgPattern, i, arg->getString(success), success); | |
1148 | formatComplexSubMessage(subMsgStart, NULL, arguments, argumentNames, | |
1149 | cnt, appendTo, success); | |
1150 | } else { | |
1151 | // This should never happen. | |
1152 | success = U_INTERNAL_PROGRAM_ERROR; | |
1153 | return; | |
1154 | } | |
1155 | ignore = updateMetaData(appendTo, prevDestLength, ignore, arg); | |
1156 | prevIndex = msgPattern.getPart(argLimit).getLimit(); | |
1157 | i = argLimit; | |
1158 | } | |
1159 | } | |
1160 | ||
1161 | ||
1162 | void MessageFormat::formatComplexSubMessage(int32_t msgStart, | |
1163 | const void *plNumber, | |
1164 | const Formattable* arguments, | |
1165 | const UnicodeString *argumentNames, | |
1166 | int32_t cnt, | |
1167 | AppendableWrapper& appendTo, | |
1168 | UErrorCode& success) const { | |
1169 | if (U_FAILURE(success)) { | |
1170 | return; | |
1171 | } | |
1172 | ||
1173 | if (!MessageImpl::jdkAposMode(msgPattern)) { | |
1174 | format(msgStart, plNumber, arguments, argumentNames, cnt, appendTo, NULL, success); | |
1175 | return; | |
1176 | } | |
1177 | ||
1178 | // JDK compatibility mode: (see JDK MessageFormat.format() API docs) | |
1179 | // - remove SKIP_SYNTAX; that is, remove half of the apostrophes | |
1180 | // - if the result string contains an open curly brace '{' then | |
1181 | // instantiate a temporary MessageFormat object and format again; | |
1182 | // otherwise just append the result string | |
1183 | const UnicodeString& msgString = msgPattern.getPatternString(); | |
1184 | UnicodeString sb; | |
1185 | int32_t prevIndex = msgPattern.getPart(msgStart).getLimit(); | |
1186 | for (int32_t i = msgStart;;) { | |
1187 | const MessagePattern::Part& part = msgPattern.getPart(++i); | |
1188 | const UMessagePatternPartType type = part.getType(); | |
1189 | int32_t index = part.getIndex(); | |
1190 | if (type == UMSGPAT_PART_TYPE_MSG_LIMIT) { | |
1191 | sb.append(msgString, prevIndex, index - prevIndex); | |
1192 | break; | |
1193 | } else if (type == UMSGPAT_PART_TYPE_REPLACE_NUMBER || type == UMSGPAT_PART_TYPE_SKIP_SYNTAX) { | |
1194 | sb.append(msgString, prevIndex, index - prevIndex); | |
1195 | if (type == UMSGPAT_PART_TYPE_REPLACE_NUMBER) { | |
1196 | const PluralSelectorContext &pluralNumber = | |
1197 | *static_cast<const PluralSelectorContext *>(plNumber); | |
1198 | if(pluralNumber.forReplaceNumber) { | |
1199 | // number-offset was already formatted. | |
1200 | sb.append(pluralNumber.numberString); | |
1201 | } else { | |
1202 | const NumberFormat* nf = getDefaultNumberFormat(success); | |
1203 | sb.append(nf->format(pluralNumber.number, sb, success)); | |
1204 | } | |
1205 | } | |
1206 | prevIndex = part.getLimit(); | |
1207 | } else if (type == UMSGPAT_PART_TYPE_ARG_START) { | |
1208 | sb.append(msgString, prevIndex, index - prevIndex); | |
1209 | prevIndex = index; | |
1210 | i = msgPattern.getLimitPartIndex(i); | |
1211 | index = msgPattern.getPart(i).getLimit(); | |
1212 | MessageImpl::appendReducedApostrophes(msgString, prevIndex, index, sb); | |
1213 | prevIndex = index; | |
1214 | } | |
1215 | } | |
1216 | if (sb.indexOf(LEFT_CURLY_BRACE) >= 0) { | |
1217 | UnicodeString emptyPattern; // gcc 3.3.3 fails with "UnicodeString()" as the first parameter. | |
1218 | MessageFormat subMsgFormat(emptyPattern, fLocale, success); | |
1219 | subMsgFormat.applyPattern(sb, UMSGPAT_APOS_DOUBLE_REQUIRED, NULL, success); | |
1220 | subMsgFormat.format(0, NULL, arguments, argumentNames, cnt, appendTo, NULL, success); | |
1221 | } else { | |
1222 | appendTo.append(sb); | |
1223 | } | |
1224 | } | |
1225 | ||
1226 | ||
1227 | UnicodeString MessageFormat::getLiteralStringUntilNextArgument(int32_t from) const { | |
1228 | const UnicodeString& msgString=msgPattern.getPatternString(); | |
1229 | int32_t prevIndex=msgPattern.getPart(from).getLimit(); | |
1230 | UnicodeString b; | |
1231 | for (int32_t i = from + 1; ; ++i) { | |
1232 | const MessagePattern::Part& part = msgPattern.getPart(i); | |
1233 | const UMessagePatternPartType type=part.getType(); | |
1234 | int32_t index=part.getIndex(); | |
1235 | b.append(msgString, prevIndex, index - prevIndex); | |
1236 | if(type==UMSGPAT_PART_TYPE_ARG_START || type==UMSGPAT_PART_TYPE_MSG_LIMIT) { | |
1237 | return b; | |
1238 | } | |
1239 | // Unexpected Part "part" in parsed message. | |
1240 | U_ASSERT(type==UMSGPAT_PART_TYPE_SKIP_SYNTAX || type==UMSGPAT_PART_TYPE_INSERT_CHAR); | |
1241 | prevIndex=part.getLimit(); | |
1242 | } | |
1243 | } | |
1244 | ||
1245 | ||
1246 | FieldPosition* MessageFormat::updateMetaData(AppendableWrapper& /*dest*/, int32_t /*prevLength*/, | |
1247 | FieldPosition* /*fp*/, const Formattable* /*argId*/) const { | |
1248 | // Unlike in Java, there are no field attributes defined for MessageFormat. Do nothing. | |
1249 | return NULL; | |
1250 | /* | |
1251 | if (fp != NULL && Field.ARGUMENT.equals(fp.getFieldAttribute())) { | |
1252 | fp->setBeginIndex(prevLength); | |
1253 | fp->setEndIndex(dest.get_length()); | |
1254 | return NULL; | |
1255 | } | |
1256 | return fp; | |
1257 | */ | |
1258 | } | |
1259 | ||
1260 | int32_t | |
1261 | MessageFormat::findOtherSubMessage(int32_t partIndex) const { | |
1262 | int32_t count=msgPattern.countParts(); | |
1263 | const MessagePattern::Part *part = &msgPattern.getPart(partIndex); | |
1264 | if(MessagePattern::Part::hasNumericValue(part->getType())) { | |
1265 | ++partIndex; | |
1266 | } | |
1267 | // Iterate over (ARG_SELECTOR [ARG_INT|ARG_DOUBLE] message) tuples | |
1268 | // until ARG_LIMIT or end of plural-only pattern. | |
1269 | UnicodeString other(FALSE, OTHER_STRING, 5); | |
1270 | do { | |
1271 | part=&msgPattern.getPart(partIndex++); | |
1272 | UMessagePatternPartType type=part->getType(); | |
1273 | if(type==UMSGPAT_PART_TYPE_ARG_LIMIT) { | |
1274 | break; | |
1275 | } | |
1276 | U_ASSERT(type==UMSGPAT_PART_TYPE_ARG_SELECTOR); | |
1277 | // part is an ARG_SELECTOR followed by an optional explicit value, and then a message | |
1278 | if(msgPattern.partSubstringMatches(*part, other)) { | |
1279 | return partIndex; | |
1280 | } | |
1281 | if(MessagePattern::Part::hasNumericValue(msgPattern.getPartType(partIndex))) { | |
1282 | ++partIndex; // skip the numeric-value part of "=1" etc. | |
1283 | } | |
1284 | partIndex=msgPattern.getLimitPartIndex(partIndex); | |
1285 | } while(++partIndex<count); | |
1286 | return 0; | |
1287 | } | |
1288 | ||
1289 | int32_t | |
1290 | MessageFormat::findFirstPluralNumberArg(int32_t msgStart, const UnicodeString &argName) const { | |
1291 | for(int32_t i=msgStart+1;; ++i) { | |
1292 | const MessagePattern::Part &part=msgPattern.getPart(i); | |
1293 | UMessagePatternPartType type=part.getType(); | |
1294 | if(type==UMSGPAT_PART_TYPE_MSG_LIMIT) { | |
1295 | return 0; | |
1296 | } | |
1297 | if(type==UMSGPAT_PART_TYPE_REPLACE_NUMBER) { | |
1298 | return -1; | |
1299 | } | |
1300 | if(type==UMSGPAT_PART_TYPE_ARG_START) { | |
1301 | UMessagePatternArgType argType=part.getArgType(); | |
1302 | if(!argName.isEmpty() && (argType==UMSGPAT_ARG_TYPE_NONE || argType==UMSGPAT_ARG_TYPE_SIMPLE)) { | |
1303 | // ARG_NUMBER or ARG_NAME | |
1304 | if(msgPattern.partSubstringMatches(msgPattern.getPart(i+1), argName)) { | |
1305 | return i; | |
1306 | } | |
1307 | } | |
1308 | i=msgPattern.getLimitPartIndex(i); | |
1309 | } | |
1310 | } | |
1311 | } | |
1312 | ||
1313 | void MessageFormat::copyObjects(const MessageFormat& that, UErrorCode& ec) { | |
1314 | // Deep copy pointer fields. | |
1315 | // We need not copy the formatAliases because they are re-filled | |
1316 | // in each getFormats() call. | |
1317 | // The defaultNumberFormat, defaultDateFormat and pluralProvider.rules | |
1318 | // also get created on demand. | |
1319 | argTypeCount = that.argTypeCount; | |
1320 | if (argTypeCount > 0) { | |
1321 | if (!allocateArgTypes(argTypeCount, ec)) { | |
1322 | return; | |
1323 | } | |
1324 | uprv_memcpy(argTypes, that.argTypes, argTypeCount * sizeof(argTypes[0])); | |
1325 | } | |
1326 | if (cachedFormatters != NULL) { | |
1327 | uhash_removeAll(cachedFormatters); | |
1328 | } | |
1329 | if (customFormatArgStarts != NULL) { | |
1330 | uhash_removeAll(customFormatArgStarts); | |
1331 | } | |
1332 | if (that.cachedFormatters) { | |
1333 | if (cachedFormatters == NULL) { | |
1334 | cachedFormatters=uhash_open(uhash_hashLong, uhash_compareLong, | |
1335 | equalFormatsForHash, &ec); | |
1336 | if (U_FAILURE(ec)) { | |
1337 | return; | |
1338 | } | |
1339 | uhash_setValueDeleter(cachedFormatters, uprv_deleteUObject); | |
1340 | } | |
1341 | ||
1342 | const int32_t count = uhash_count(that.cachedFormatters); | |
1343 | int32_t pos, idx; | |
1344 | for (idx = 0, pos = UHASH_FIRST; idx < count && U_SUCCESS(ec); ++idx) { | |
1345 | const UHashElement* cur = uhash_nextElement(that.cachedFormatters, &pos); | |
1346 | Format* newFormat = ((Format*)(cur->value.pointer))->clone(); | |
1347 | if (newFormat) { | |
1348 | uhash_iput(cachedFormatters, cur->key.integer, newFormat, &ec); | |
1349 | } else { | |
1350 | ec = U_MEMORY_ALLOCATION_ERROR; | |
1351 | return; | |
1352 | } | |
1353 | } | |
1354 | } | |
1355 | if (that.customFormatArgStarts) { | |
1356 | if (customFormatArgStarts == NULL) { | |
1357 | customFormatArgStarts=uhash_open(uhash_hashLong, uhash_compareLong, | |
1358 | NULL, &ec); | |
1359 | } | |
1360 | const int32_t count = uhash_count(that.customFormatArgStarts); | |
1361 | int32_t pos, idx; | |
1362 | for (idx = 0, pos = UHASH_FIRST; idx < count && U_SUCCESS(ec); ++idx) { | |
1363 | const UHashElement* cur = uhash_nextElement(that.customFormatArgStarts, &pos); | |
1364 | uhash_iputi(customFormatArgStarts, cur->key.integer, cur->value.integer, &ec); | |
1365 | } | |
1366 | } | |
1367 | } | |
1368 | ||
1369 | ||
1370 | Formattable* | |
1371 | MessageFormat::parse(int32_t msgStart, | |
1372 | const UnicodeString& source, | |
1373 | ParsePosition& pos, | |
1374 | int32_t& count, | |
1375 | UErrorCode& ec) const { | |
1376 | count = 0; | |
1377 | if (U_FAILURE(ec)) { | |
1378 | pos.setErrorIndex(pos.getIndex()); | |
1379 | return NULL; | |
1380 | } | |
1381 | // parse() does not work with named arguments. | |
1382 | if (msgPattern.hasNamedArguments()) { | |
1383 | ec = U_ARGUMENT_TYPE_MISMATCH; | |
1384 | pos.setErrorIndex(pos.getIndex()); | |
1385 | return NULL; | |
1386 | } | |
1387 | LocalArray<Formattable> resultArray(new Formattable[argTypeCount ? argTypeCount : 1]); | |
1388 | const UnicodeString& msgString=msgPattern.getPatternString(); | |
1389 | int32_t prevIndex=msgPattern.getPart(msgStart).getLimit(); | |
1390 | int32_t sourceOffset = pos.getIndex(); | |
1391 | ParsePosition tempStatus(0); | |
1392 | ||
1393 | for(int32_t i=msgStart+1; ; ++i) { | |
1394 | UBool haveArgResult = FALSE; | |
1395 | const MessagePattern::Part* part=&msgPattern.getPart(i); | |
1396 | const UMessagePatternPartType type=part->getType(); | |
1397 | int32_t index=part->getIndex(); | |
1398 | // Make sure the literal string matches. | |
1399 | int32_t len = index - prevIndex; | |
1400 | if (len == 0 || (0 == msgString.compare(prevIndex, len, source, sourceOffset, len))) { | |
1401 | sourceOffset += len; | |
1402 | prevIndex += len; | |
1403 | } else { | |
1404 | pos.setErrorIndex(sourceOffset); | |
1405 | return NULL; // leave index as is to signal error | |
1406 | } | |
1407 | if(type==UMSGPAT_PART_TYPE_MSG_LIMIT) { | |
1408 | // Things went well! Done. | |
1409 | pos.setIndex(sourceOffset); | |
1410 | return resultArray.orphan(); | |
1411 | } | |
1412 | if(type==UMSGPAT_PART_TYPE_SKIP_SYNTAX || type==UMSGPAT_PART_TYPE_INSERT_CHAR) { | |
1413 | prevIndex=part->getLimit(); | |
1414 | continue; | |
1415 | } | |
1416 | // We do not support parsing Plural formats. (No REPLACE_NUMBER here.) | |
1417 | // Unexpected Part "part" in parsed message. | |
1418 | U_ASSERT(type==UMSGPAT_PART_TYPE_ARG_START); | |
1419 | int32_t argLimit=msgPattern.getLimitPartIndex(i); | |
1420 | ||
1421 | UMessagePatternArgType argType=part->getArgType(); | |
1422 | part=&msgPattern.getPart(++i); | |
1423 | int32_t argNumber = part->getValue(); // ARG_NUMBER | |
1424 | UnicodeString key; | |
1425 | ++i; | |
1426 | const Format* formatter = NULL; | |
1427 | Formattable& argResult = resultArray[argNumber]; | |
1428 | ||
1429 | if(cachedFormatters!=NULL && (formatter = getCachedFormatter(i - 2))!=NULL) { | |
1430 | // Just parse using the formatter. | |
1431 | tempStatus.setIndex(sourceOffset); | |
1432 | formatter->parseObject(source, argResult, tempStatus); | |
1433 | if (tempStatus.getIndex() == sourceOffset) { | |
1434 | pos.setErrorIndex(sourceOffset); | |
1435 | return NULL; // leave index as is to signal error | |
1436 | } | |
1437 | sourceOffset = tempStatus.getIndex(); | |
1438 | haveArgResult = TRUE; | |
1439 | } else if( | |
1440 | argType==UMSGPAT_ARG_TYPE_NONE || (cachedFormatters && uhash_iget(cachedFormatters, i -2))) { | |
1441 | // We arrive here if getCachedFormatter returned NULL, but there was actually an element in the hash table. | |
1442 | // This can only happen if the hash table contained a DummyFormat, so the if statement above is a check | |
1443 | // for the hash table containind DummyFormat. | |
1444 | ||
1445 | // Match as a string. | |
1446 | // if at end, use longest possible match | |
1447 | // otherwise uses first match to intervening string | |
1448 | // does NOT recursively try all possibilities | |
1449 | UnicodeString stringAfterArgument = getLiteralStringUntilNextArgument(argLimit); | |
1450 | int32_t next; | |
1451 | if (!stringAfterArgument.isEmpty()) { | |
1452 | next = source.indexOf(stringAfterArgument, sourceOffset); | |
1453 | } else { | |
1454 | next = source.length(); | |
1455 | } | |
1456 | if (next < 0) { | |
1457 | pos.setErrorIndex(sourceOffset); | |
1458 | return NULL; // leave index as is to signal error | |
1459 | } else { | |
1460 | UnicodeString strValue(source.tempSubString(sourceOffset, next - sourceOffset)); | |
1461 | UnicodeString compValue; | |
1462 | compValue.append(LEFT_CURLY_BRACE); | |
1463 | itos(argNumber, compValue); | |
1464 | compValue.append(RIGHT_CURLY_BRACE); | |
1465 | if (0 != strValue.compare(compValue)) { | |
1466 | argResult.setString(strValue); | |
1467 | haveArgResult = TRUE; | |
1468 | } | |
1469 | sourceOffset = next; | |
1470 | } | |
1471 | } else if(argType==UMSGPAT_ARG_TYPE_CHOICE) { | |
1472 | tempStatus.setIndex(sourceOffset); | |
1473 | double choiceResult = ChoiceFormat::parseArgument(msgPattern, i, source, tempStatus); | |
1474 | if (tempStatus.getIndex() == sourceOffset) { | |
1475 | pos.setErrorIndex(sourceOffset); | |
1476 | return NULL; // leave index as is to signal error | |
1477 | } | |
1478 | argResult.setDouble(choiceResult); | |
1479 | haveArgResult = TRUE; | |
1480 | sourceOffset = tempStatus.getIndex(); | |
1481 | } else if(UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE(argType) || argType==UMSGPAT_ARG_TYPE_SELECT) { | |
1482 | // Parsing not supported. | |
1483 | ec = U_UNSUPPORTED_ERROR; | |
1484 | return NULL; | |
1485 | } else { | |
1486 | // This should never happen. | |
1487 | ec = U_INTERNAL_PROGRAM_ERROR; | |
1488 | return NULL; | |
1489 | } | |
1490 | if (haveArgResult && count <= argNumber) { | |
1491 | count = argNumber + 1; | |
1492 | } | |
1493 | prevIndex=msgPattern.getPart(argLimit).getLimit(); | |
1494 | i=argLimit; | |
1495 | } | |
1496 | } | |
1497 | // ------------------------------------- | |
1498 | // Parses the source pattern and returns the Formattable objects array, | |
1499 | // the array count and the ending parse position. The caller of this method | |
1500 | // owns the array. | |
1501 | ||
1502 | Formattable* | |
1503 | MessageFormat::parse(const UnicodeString& source, | |
1504 | ParsePosition& pos, | |
1505 | int32_t& count) const { | |
1506 | UErrorCode ec = U_ZERO_ERROR; | |
1507 | return parse(0, source, pos, count, ec); | |
1508 | } | |
1509 | ||
1510 | // ------------------------------------- | |
1511 | // Parses the source string and returns the array of | |
1512 | // Formattable objects and the array count. The caller | |
1513 | // owns the returned array. | |
1514 | ||
1515 | Formattable* | |
1516 | MessageFormat::parse(const UnicodeString& source, | |
1517 | int32_t& cnt, | |
1518 | UErrorCode& success) const | |
1519 | { | |
1520 | if (msgPattern.hasNamedArguments()) { | |
1521 | success = U_ARGUMENT_TYPE_MISMATCH; | |
1522 | return NULL; | |
1523 | } | |
1524 | ParsePosition status(0); | |
1525 | // Calls the actual implementation method and starts | |
1526 | // from zero offset of the source text. | |
1527 | Formattable* result = parse(source, status, cnt); | |
1528 | if (status.getIndex() == 0) { | |
1529 | success = U_MESSAGE_PARSE_ERROR; | |
1530 | delete[] result; | |
1531 | return NULL; | |
1532 | } | |
1533 | return result; | |
1534 | } | |
1535 | ||
1536 | // ------------------------------------- | |
1537 | // Parses the source text and copy into the result buffer. | |
1538 | ||
1539 | void | |
1540 | MessageFormat::parseObject( const UnicodeString& source, | |
1541 | Formattable& result, | |
1542 | ParsePosition& status) const | |
1543 | { | |
1544 | int32_t cnt = 0; | |
1545 | Formattable* tmpResult = parse(source, status, cnt); | |
1546 | if (tmpResult != NULL) | |
1547 | result.adoptArray(tmpResult, cnt); | |
1548 | } | |
1549 | ||
1550 | UnicodeString | |
1551 | MessageFormat::autoQuoteApostrophe(const UnicodeString& pattern, UErrorCode& status) { | |
1552 | UnicodeString result; | |
1553 | if (U_SUCCESS(status)) { | |
1554 | int32_t plen = pattern.length(); | |
1555 | const UChar* pat = pattern.getBuffer(); | |
1556 | int32_t blen = plen * 2 + 1; // space for null termination, convenience | |
1557 | UChar* buf = result.getBuffer(blen); | |
1558 | if (buf == NULL) { | |
1559 | status = U_MEMORY_ALLOCATION_ERROR; | |
1560 | } else { | |
1561 | int32_t len = umsg_autoQuoteApostrophe(pat, plen, buf, blen, &status); | |
1562 | result.releaseBuffer(U_SUCCESS(status) ? len : 0); | |
1563 | } | |
1564 | } | |
1565 | if (U_FAILURE(status)) { | |
1566 | result.setToBogus(); | |
1567 | } | |
1568 | return result; | |
1569 | } | |
1570 | ||
1571 | // ------------------------------------- | |
1572 | ||
1573 | static Format* makeRBNF(URBNFRuleSetTag tag, const Locale& locale, const UnicodeString& defaultRuleSet, UErrorCode& ec) { | |
1574 | RuleBasedNumberFormat* fmt = new RuleBasedNumberFormat(tag, locale, ec); | |
1575 | if (fmt == NULL) { | |
1576 | ec = U_MEMORY_ALLOCATION_ERROR; | |
1577 | } else if (U_SUCCESS(ec) && defaultRuleSet.length() > 0) { | |
1578 | UErrorCode localStatus = U_ZERO_ERROR; // ignore unrecognized default rule set | |
1579 | fmt->setDefaultRuleSet(defaultRuleSet, localStatus); | |
1580 | } | |
1581 | return fmt; | |
1582 | } | |
1583 | ||
1584 | void MessageFormat::cacheExplicitFormats(UErrorCode& status) { | |
1585 | if (U_FAILURE(status)) { | |
1586 | return; | |
1587 | } | |
1588 | ||
1589 | if (cachedFormatters != NULL) { | |
1590 | uhash_removeAll(cachedFormatters); | |
1591 | } | |
1592 | if (customFormatArgStarts != NULL) { | |
1593 | uhash_removeAll(customFormatArgStarts); | |
1594 | } | |
1595 | ||
1596 | // The last two "parts" can at most be ARG_LIMIT and MSG_LIMIT | |
1597 | // which we need not examine. | |
1598 | int32_t limit = msgPattern.countParts() - 2; | |
1599 | argTypeCount = 0; | |
1600 | // We also need not look at the first two "parts" | |
1601 | // (at most MSG_START and ARG_START) in this loop. | |
1602 | // We determine the argTypeCount first so that we can allocateArgTypes | |
1603 | // so that the next loop can set argTypes[argNumber]. | |
1604 | // (This is for the C API which needs the argTypes to read its va_arg list.) | |
1605 | for (int32_t i = 2; i < limit && U_SUCCESS(status); ++i) { | |
1606 | const MessagePattern::Part& part = msgPattern.getPart(i); | |
1607 | if (part.getType() == UMSGPAT_PART_TYPE_ARG_NUMBER) { | |
1608 | const int argNumber = part.getValue(); | |
1609 | if (argNumber >= argTypeCount) { | |
1610 | argTypeCount = argNumber + 1; | |
1611 | } | |
1612 | } | |
1613 | } | |
1614 | if (!allocateArgTypes(argTypeCount, status)) { | |
1615 | return; | |
1616 | } | |
1617 | // Set all argTypes to kObject, as a "none" value, for lack of any better value. | |
1618 | // We never use kObject for real arguments. | |
1619 | // We use it as "no argument yet" for the check for hasArgTypeConflicts. | |
1620 | for (int32_t i = 0; i < argTypeCount; ++i) { | |
1621 | argTypes[i] = Formattable::kObject; | |
1622 | } | |
1623 | hasArgTypeConflicts = FALSE; | |
1624 | ||
1625 | // This loop starts at part index 1 because we do need to examine | |
1626 | // ARG_START parts. (But we can ignore the MSG_START.) | |
1627 | for (int32_t i = 1; i < limit && U_SUCCESS(status); ++i) { | |
1628 | const MessagePattern::Part* part = &msgPattern.getPart(i); | |
1629 | if (part->getType() != UMSGPAT_PART_TYPE_ARG_START) { | |
1630 | continue; | |
1631 | } | |
1632 | UMessagePatternArgType argType = part->getArgType(); | |
1633 | ||
1634 | int32_t argNumber = -1; | |
1635 | part = &msgPattern.getPart(i + 1); | |
1636 | if (part->getType() == UMSGPAT_PART_TYPE_ARG_NUMBER) { | |
1637 | argNumber = part->getValue(); | |
1638 | } | |
1639 | Formattable::Type formattableType; | |
1640 | ||
1641 | switch (argType) { | |
1642 | case UMSGPAT_ARG_TYPE_NONE: | |
1643 | formattableType = Formattable::kString; | |
1644 | break; | |
1645 | case UMSGPAT_ARG_TYPE_SIMPLE: { | |
1646 | int32_t index = i; | |
1647 | i += 2; | |
1648 | UnicodeString explicitType = msgPattern.getSubstring(msgPattern.getPart(i++)); | |
1649 | UnicodeString style; | |
1650 | if ((part = &msgPattern.getPart(i))->getType() == UMSGPAT_PART_TYPE_ARG_STYLE) { | |
1651 | style = msgPattern.getSubstring(*part); | |
1652 | ++i; | |
1653 | } | |
1654 | UParseError parseError; | |
1655 | Format* formatter = createAppropriateFormat(explicitType, style, formattableType, parseError, status); | |
1656 | setArgStartFormat(index, formatter, status); | |
1657 | break; | |
1658 | } | |
1659 | case UMSGPAT_ARG_TYPE_CHOICE: | |
1660 | case UMSGPAT_ARG_TYPE_PLURAL: | |
1661 | case UMSGPAT_ARG_TYPE_SELECTORDINAL: | |
1662 | formattableType = Formattable::kDouble; | |
1663 | break; | |
1664 | case UMSGPAT_ARG_TYPE_SELECT: | |
1665 | formattableType = Formattable::kString; | |
1666 | break; | |
1667 | default: | |
1668 | status = U_INTERNAL_PROGRAM_ERROR; // Should be unreachable. | |
1669 | formattableType = Formattable::kString; | |
1670 | break; | |
1671 | } | |
1672 | if (argNumber != -1) { | |
1673 | if (argTypes[argNumber] != Formattable::kObject && argTypes[argNumber] != formattableType) { | |
1674 | hasArgTypeConflicts = TRUE; | |
1675 | } | |
1676 | argTypes[argNumber] = formattableType; | |
1677 | } | |
1678 | } | |
1679 | } | |
1680 | ||
1681 | Format* MessageFormat::createAppropriateFormat(UnicodeString& type, UnicodeString& style, | |
1682 | Formattable::Type& formattableType, UParseError& parseError, | |
1683 | UErrorCode& ec) { | |
1684 | if (U_FAILURE(ec)) { | |
1685 | return NULL; | |
1686 | } | |
1687 | Format* fmt = NULL; | |
1688 | int32_t typeID, styleID; | |
1689 | DateFormat::EStyle date_style; | |
1690 | int32_t firstNonSpace; | |
1691 | ||
1692 | switch (typeID = findKeyword(type, TYPE_IDS)) { | |
1693 | case 0: // number | |
1694 | formattableType = Formattable::kDouble; | |
1695 | switch (findKeyword(style, NUMBER_STYLE_IDS)) { | |
1696 | case 0: // default | |
1697 | fmt = NumberFormat::createInstance(fLocale, ec); | |
1698 | break; | |
1699 | case 1: // currency | |
1700 | fmt = NumberFormat::createCurrencyInstance(fLocale, ec); | |
1701 | break; | |
1702 | case 2: // percent | |
1703 | fmt = NumberFormat::createPercentInstance(fLocale, ec); | |
1704 | break; | |
1705 | case 3: // integer | |
1706 | formattableType = Formattable::kLong; | |
1707 | fmt = createIntegerFormat(fLocale, ec); | |
1708 | break; | |
1709 | default: // pattern or skeleton | |
1710 | firstNonSpace = PatternProps::skipWhiteSpace(style, 0); | |
1711 | if (style.compare(firstNonSpace, 2, u"::", 0, 2) == 0) { | |
1712 | // Skeleton | |
1713 | UnicodeString skeleton = style.tempSubString(firstNonSpace + 2); | |
1714 | fmt = number::NumberFormatter::forSkeleton(skeleton, ec).locale(fLocale).toFormat(ec); | |
1715 | } else { | |
1716 | // Pattern | |
1717 | fmt = NumberFormat::createInstance(fLocale, ec); | |
1718 | if (fmt) { | |
1719 | auto* decfmt = dynamic_cast<DecimalFormat*>(fmt); | |
1720 | if (decfmt != nullptr) { | |
1721 | decfmt->applyPattern(style, parseError, ec); | |
1722 | } | |
1723 | } | |
1724 | } | |
1725 | break; | |
1726 | } | |
1727 | break; | |
1728 | ||
1729 | case 1: // date | |
1730 | case 2: // time | |
1731 | formattableType = Formattable::kDate; | |
1732 | firstNonSpace = PatternProps::skipWhiteSpace(style, 0); | |
1733 | if (style.compare(firstNonSpace, 2, u"::", 0, 2) == 0) { | |
1734 | // Skeleton | |
1735 | UnicodeString skeleton = style.tempSubString(firstNonSpace + 2); | |
1736 | fmt = DateFormat::createInstanceForSkeleton(skeleton, fLocale, ec); | |
1737 | } else { | |
1738 | // Pattern | |
1739 | styleID = findKeyword(style, DATE_STYLE_IDS); | |
1740 | date_style = (styleID >= 0) ? DATE_STYLES[styleID] : DateFormat::kDefault; | |
1741 | ||
1742 | if (typeID == 1) { | |
1743 | fmt = DateFormat::createDateInstance(date_style, fLocale); | |
1744 | } else { | |
1745 | fmt = DateFormat::createTimeInstance(date_style, fLocale); | |
1746 | } | |
1747 | ||
1748 | if (styleID < 0 && fmt != NULL) { | |
1749 | SimpleDateFormat* sdtfmt = dynamic_cast<SimpleDateFormat*>(fmt); | |
1750 | if (sdtfmt != NULL) { | |
1751 | sdtfmt->applyPattern(style); | |
1752 | } | |
1753 | } | |
1754 | } | |
1755 | break; | |
1756 | ||
1757 | case 3: // spellout | |
1758 | formattableType = Formattable::kDouble; | |
1759 | fmt = makeRBNF(URBNF_SPELLOUT, fLocale, style, ec); | |
1760 | break; | |
1761 | case 4: // ordinal | |
1762 | formattableType = Formattable::kDouble; | |
1763 | fmt = makeRBNF(URBNF_ORDINAL, fLocale, style, ec); | |
1764 | break; | |
1765 | case 5: // duration | |
1766 | formattableType = Formattable::kDouble; | |
1767 | fmt = makeRBNF(URBNF_DURATION, fLocale, style, ec); | |
1768 | break; | |
1769 | default: | |
1770 | formattableType = Formattable::kString; | |
1771 | ec = U_ILLEGAL_ARGUMENT_ERROR; | |
1772 | break; | |
1773 | } | |
1774 | ||
1775 | return fmt; | |
1776 | } | |
1777 | ||
1778 | ||
1779 | //------------------------------------- | |
1780 | // Finds the string, s, in the string array, list. | |
1781 | int32_t MessageFormat::findKeyword(const UnicodeString& s, | |
1782 | const UChar * const *list) | |
1783 | { | |
1784 | if (s.isEmpty()) { | |
1785 | return 0; // default | |
1786 | } | |
1787 | ||
1788 | int32_t length = s.length(); | |
1789 | const UChar *ps = PatternProps::trimWhiteSpace(s.getBuffer(), length); | |
1790 | UnicodeString buffer(FALSE, ps, length); | |
1791 | // Trims the space characters and turns all characters | |
1792 | // in s to lower case. | |
1793 | buffer.toLower(""); | |
1794 | for (int32_t i = 0; list[i]; ++i) { | |
1795 | if (!buffer.compare(list[i], u_strlen(list[i]))) { | |
1796 | return i; | |
1797 | } | |
1798 | } | |
1799 | return -1; | |
1800 | } | |
1801 | ||
1802 | /** | |
1803 | * Convenience method that ought to be in NumberFormat | |
1804 | */ | |
1805 | NumberFormat* | |
1806 | MessageFormat::createIntegerFormat(const Locale& locale, UErrorCode& status) const { | |
1807 | NumberFormat *temp = NumberFormat::createInstance(locale, status); | |
1808 | DecimalFormat *temp2; | |
1809 | if (temp != NULL && (temp2 = dynamic_cast<DecimalFormat*>(temp)) != NULL) { | |
1810 | temp2->setMaximumFractionDigits(0); | |
1811 | temp2->setDecimalSeparatorAlwaysShown(FALSE); | |
1812 | temp2->setParseIntegerOnly(TRUE); | |
1813 | } | |
1814 | ||
1815 | return temp; | |
1816 | } | |
1817 | ||
1818 | /** | |
1819 | * Return the default number format. Used to format a numeric | |
1820 | * argument when subformats[i].format is NULL. Returns NULL | |
1821 | * on failure. | |
1822 | * | |
1823 | * Semantically const but may modify *this. | |
1824 | */ | |
1825 | const NumberFormat* MessageFormat::getDefaultNumberFormat(UErrorCode& ec) const { | |
1826 | if (defaultNumberFormat == NULL) { | |
1827 | MessageFormat* t = (MessageFormat*) this; | |
1828 | t->defaultNumberFormat = NumberFormat::createInstance(fLocale, ec); | |
1829 | if (U_FAILURE(ec)) { | |
1830 | delete t->defaultNumberFormat; | |
1831 | t->defaultNumberFormat = NULL; | |
1832 | } else if (t->defaultNumberFormat == NULL) { | |
1833 | ec = U_MEMORY_ALLOCATION_ERROR; | |
1834 | } | |
1835 | } | |
1836 | return defaultNumberFormat; | |
1837 | } | |
1838 | ||
1839 | /** | |
1840 | * Return the default date format. Used to format a date | |
1841 | * argument when subformats[i].format is NULL. Returns NULL | |
1842 | * on failure. | |
1843 | * | |
1844 | * Semantically const but may modify *this. | |
1845 | */ | |
1846 | const DateFormat* MessageFormat::getDefaultDateFormat(UErrorCode& ec) const { | |
1847 | if (defaultDateFormat == NULL) { | |
1848 | MessageFormat* t = (MessageFormat*) this; | |
1849 | t->defaultDateFormat = DateFormat::createDateTimeInstance(DateFormat::kShort, DateFormat::kShort, fLocale); | |
1850 | if (t->defaultDateFormat == NULL) { | |
1851 | ec = U_MEMORY_ALLOCATION_ERROR; | |
1852 | } | |
1853 | } | |
1854 | return defaultDateFormat; | |
1855 | } | |
1856 | ||
1857 | UBool | |
1858 | MessageFormat::usesNamedArguments() const { | |
1859 | return msgPattern.hasNamedArguments(); | |
1860 | } | |
1861 | ||
1862 | int32_t | |
1863 | MessageFormat::getArgTypeCount() const { | |
1864 | return argTypeCount; | |
1865 | } | |
1866 | ||
1867 | UBool MessageFormat::equalFormats(const void* left, const void* right) { | |
1868 | return *(const Format*)left==*(const Format*)right; | |
1869 | } | |
1870 | ||
1871 | ||
1872 | UBool MessageFormat::DummyFormat::operator==(const Format&) const { | |
1873 | return TRUE; | |
1874 | } | |
1875 | ||
1876 | Format* MessageFormat::DummyFormat::clone() const { | |
1877 | return new DummyFormat(); | |
1878 | } | |
1879 | ||
1880 | UnicodeString& MessageFormat::DummyFormat::format(const Formattable&, | |
1881 | UnicodeString& appendTo, | |
1882 | UErrorCode& status) const { | |
1883 | if (U_SUCCESS(status)) { | |
1884 | status = U_UNSUPPORTED_ERROR; | |
1885 | } | |
1886 | return appendTo; | |
1887 | } | |
1888 | ||
1889 | UnicodeString& MessageFormat::DummyFormat::format(const Formattable&, | |
1890 | UnicodeString& appendTo, | |
1891 | FieldPosition&, | |
1892 | UErrorCode& status) const { | |
1893 | if (U_SUCCESS(status)) { | |
1894 | status = U_UNSUPPORTED_ERROR; | |
1895 | } | |
1896 | return appendTo; | |
1897 | } | |
1898 | ||
1899 | UnicodeString& MessageFormat::DummyFormat::format(const Formattable&, | |
1900 | UnicodeString& appendTo, | |
1901 | FieldPositionIterator*, | |
1902 | UErrorCode& status) const { | |
1903 | if (U_SUCCESS(status)) { | |
1904 | status = U_UNSUPPORTED_ERROR; | |
1905 | } | |
1906 | return appendTo; | |
1907 | } | |
1908 | ||
1909 | void MessageFormat::DummyFormat::parseObject(const UnicodeString&, | |
1910 | Formattable&, | |
1911 | ParsePosition& ) const { | |
1912 | } | |
1913 | ||
1914 | ||
1915 | FormatNameEnumeration::FormatNameEnumeration(UVector *fNameList, UErrorCode& /*status*/) { | |
1916 | pos=0; | |
1917 | fFormatNames = fNameList; | |
1918 | } | |
1919 | ||
1920 | const UnicodeString* | |
1921 | FormatNameEnumeration::snext(UErrorCode& status) { | |
1922 | if (U_SUCCESS(status) && pos < fFormatNames->size()) { | |
1923 | return (const UnicodeString*)fFormatNames->elementAt(pos++); | |
1924 | } | |
1925 | return NULL; | |
1926 | } | |
1927 | ||
1928 | void | |
1929 | FormatNameEnumeration::reset(UErrorCode& /*status*/) { | |
1930 | pos=0; | |
1931 | } | |
1932 | ||
1933 | int32_t | |
1934 | FormatNameEnumeration::count(UErrorCode& /*status*/) const { | |
1935 | return (fFormatNames==NULL) ? 0 : fFormatNames->size(); | |
1936 | } | |
1937 | ||
1938 | FormatNameEnumeration::~FormatNameEnumeration() { | |
1939 | delete fFormatNames; | |
1940 | } | |
1941 | ||
1942 | MessageFormat::PluralSelectorProvider::PluralSelectorProvider(const MessageFormat &mf, UPluralType t) | |
1943 | : msgFormat(mf), rules(NULL), type(t) { | |
1944 | } | |
1945 | ||
1946 | MessageFormat::PluralSelectorProvider::~PluralSelectorProvider() { | |
1947 | delete rules; | |
1948 | } | |
1949 | ||
1950 | UnicodeString MessageFormat::PluralSelectorProvider::select(void *ctx, double number, | |
1951 | UErrorCode& ec) const { | |
1952 | if (U_FAILURE(ec)) { | |
1953 | return UnicodeString(FALSE, OTHER_STRING, 5); | |
1954 | } | |
1955 | MessageFormat::PluralSelectorProvider* t = const_cast<MessageFormat::PluralSelectorProvider*>(this); | |
1956 | if(rules == NULL) { | |
1957 | t->rules = PluralRules::forLocale(msgFormat.fLocale, type, ec); | |
1958 | if (U_FAILURE(ec)) { | |
1959 | return UnicodeString(FALSE, OTHER_STRING, 5); | |
1960 | } | |
1961 | } | |
1962 | // Select a sub-message according to how the number is formatted, | |
1963 | // which is specified in the selected sub-message. | |
1964 | // We avoid this circle by looking at how | |
1965 | // the number is formatted in the "other" sub-message | |
1966 | // which must always be present and usually contains the number. | |
1967 | // Message authors should be consistent across sub-messages. | |
1968 | PluralSelectorContext &context = *static_cast<PluralSelectorContext *>(ctx); | |
1969 | int32_t otherIndex = msgFormat.findOtherSubMessage(context.startIndex); | |
1970 | context.numberArgIndex = msgFormat.findFirstPluralNumberArg(otherIndex, context.argName); | |
1971 | if(context.numberArgIndex > 0 && msgFormat.cachedFormatters != NULL) { | |
1972 | context.formatter = | |
1973 | (const Format*)uhash_iget(msgFormat.cachedFormatters, context.numberArgIndex); | |
1974 | } | |
1975 | if(context.formatter == NULL) { | |
1976 | context.formatter = msgFormat.getDefaultNumberFormat(ec); | |
1977 | context.forReplaceNumber = TRUE; | |
1978 | } | |
1979 | if (context.number.getDouble(ec) != number) { | |
1980 | ec = U_INTERNAL_PROGRAM_ERROR; | |
1981 | return UnicodeString(FALSE, OTHER_STRING, 5); | |
1982 | } | |
1983 | context.formatter->format(context.number, context.numberString, ec); | |
1984 | auto* decFmt = dynamic_cast<const DecimalFormat *>(context.formatter); | |
1985 | if(decFmt != NULL) { | |
1986 | number::impl::DecimalQuantity dq; | |
1987 | decFmt->formatToDecimalQuantity(context.number, dq, ec); | |
1988 | if (U_FAILURE(ec)) { | |
1989 | return UnicodeString(FALSE, OTHER_STRING, 5); | |
1990 | } | |
1991 | return rules->select(dq); | |
1992 | } else { | |
1993 | return rules->select(number); | |
1994 | } | |
1995 | } | |
1996 | ||
1997 | void MessageFormat::PluralSelectorProvider::reset() { | |
1998 | delete rules; | |
1999 | rules = NULL; | |
2000 | } | |
2001 | ||
2002 | ||
2003 | U_NAMESPACE_END | |
2004 | ||
2005 | #endif /* #if !UCONFIG_NO_FORMATTING */ | |
2006 | ||
2007 | //eof |