/src/icu/icu4c/source/i18n/ucol_res.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2016 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | /* |
4 | | ******************************************************************************* |
5 | | * Copyright (C) 1996-2016, International Business Machines |
6 | | * Corporation and others. All Rights Reserved. |
7 | | ******************************************************************************* |
8 | | * file name: ucol_res.cpp |
9 | | * encoding: UTF-8 |
10 | | * tab size: 8 (not used) |
11 | | * indentation:4 |
12 | | * |
13 | | * Description: |
14 | | * This file contains dependencies that the collation run-time doesn't normally |
15 | | * need. This mainly contains resource bundle usage and collation meta information |
16 | | * |
17 | | * Modification history |
18 | | * Date Name Comments |
19 | | * 1996-1999 various members of ICU team maintained C API for collation framework |
20 | | * 02/16/2001 synwee Added internal method getPrevSpecialCE |
21 | | * 03/01/2001 synwee Added maxexpansion functionality. |
22 | | * 03/16/2001 weiv Collation framework is rewritten in C and made UCA compliant |
23 | | * 12/08/2004 grhoten Split part of ucol.cpp into ucol_res.cpp |
24 | | * 2012-2014 markus Rewritten in C++ again. |
25 | | */ |
26 | | |
27 | | #include "unicode/utypes.h" |
28 | | |
29 | | #if !UCONFIG_NO_COLLATION |
30 | | |
31 | | #include "unicode/coll.h" |
32 | | #include "unicode/localpointer.h" |
33 | | #include "unicode/locid.h" |
34 | | #include "unicode/tblcoll.h" |
35 | | #include "unicode/ucol.h" |
36 | | #include "unicode/uloc.h" |
37 | | #include "unicode/unistr.h" |
38 | | #include "unicode/ures.h" |
39 | | #include "charstr.h" |
40 | | #include "cmemory.h" |
41 | | #include "cstring.h" |
42 | | #include "collationdatareader.h" |
43 | | #include "collationroot.h" |
44 | | #include "collationtailoring.h" |
45 | | #include "resource.h" |
46 | | #include "putilimp.h" |
47 | | #include "uassert.h" |
48 | | #include "ucln_in.h" |
49 | | #include "ucol_imp.h" |
50 | | #include "uenumimp.h" |
51 | | #include "ulist.h" |
52 | | #include "umutex.h" |
53 | | #include "unifiedcache.h" |
54 | | #include "uresimp.h" |
55 | | #include "ustrenum.h" |
56 | | #include "utracimp.h" |
57 | | |
58 | | U_NAMESPACE_BEGIN |
59 | | |
60 | | namespace { |
61 | | |
62 | | const char16_t* rootRules = nullptr; |
63 | | int32_t rootRulesLength = 0; |
64 | | UResourceBundle* rootBundle = nullptr; |
65 | | UInitOnce gInitOnceUcolRes{}; |
66 | | |
67 | | } // namespace |
68 | | |
69 | | U_CDECL_BEGIN |
70 | | |
71 | | static UBool U_CALLCONV |
72 | 0 | ucol_res_cleanup() { |
73 | 0 | rootRules = nullptr; |
74 | 0 | rootRulesLength = 0; |
75 | 0 | ures_close(rootBundle); |
76 | 0 | rootBundle = nullptr; |
77 | 0 | gInitOnceUcolRes.reset(); |
78 | 0 | return true; |
79 | 0 | } |
80 | | |
81 | | void U_CALLCONV |
82 | 0 | CollationLoader::loadRootRules(UErrorCode &errorCode) { |
83 | 0 | if(U_FAILURE(errorCode)) { return; } |
84 | 0 | rootBundle = ures_open(U_ICUDATA_COLL, kRootLocaleName, &errorCode); |
85 | 0 | if(U_FAILURE(errorCode)) { return; } |
86 | 0 | rootRules = ures_getStringByKey(rootBundle, "UCARules", &rootRulesLength, &errorCode); |
87 | 0 | if(U_FAILURE(errorCode)) { |
88 | 0 | ures_close(rootBundle); |
89 | 0 | rootBundle = nullptr; |
90 | 0 | return; |
91 | 0 | } |
92 | 0 | ucln_i18n_registerCleanup(UCLN_I18N_UCOL_RES, ucol_res_cleanup); |
93 | 0 | } |
94 | | |
95 | | U_CDECL_END |
96 | | |
97 | | void |
98 | 0 | CollationLoader::appendRootRules(UnicodeString &s) { |
99 | 0 | UErrorCode errorCode = U_ZERO_ERROR; |
100 | 0 | umtx_initOnce(gInitOnceUcolRes, CollationLoader::loadRootRules, errorCode); |
101 | 0 | if(U_SUCCESS(errorCode)) { |
102 | 0 | s.append(rootRules, rootRulesLength); |
103 | 0 | } |
104 | 0 | } |
105 | | |
106 | | void |
107 | | CollationLoader::loadRules(const char *localeID, const char *collationType, |
108 | 4.87k | UnicodeString &rules, UErrorCode &errorCode) { |
109 | 4.87k | if(U_FAILURE(errorCode)) { return; } |
110 | 4.87k | U_ASSERT(collationType != nullptr && *collationType != 0); |
111 | | // Copy the type for lowercasing. |
112 | 4.87k | char type[16]; |
113 | 4.87k | int32_t typeLength = static_cast<int32_t>(uprv_strlen(collationType)); |
114 | 4.87k | if(typeLength >= UPRV_LENGTHOF(type)) { |
115 | 21 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
116 | 21 | return; |
117 | 21 | } |
118 | 4.87k | uprv_memcpy(type, collationType, typeLength + 1); |
119 | 4.85k | T_CString_toLowerCase(type); |
120 | | |
121 | 4.85k | LocalUResourceBundlePointer bundle(ures_open(U_ICUDATA_COLL, localeID, &errorCode)); |
122 | 4.85k | LocalUResourceBundlePointer collations( |
123 | 4.85k | ures_getByKey(bundle.getAlias(), "collations", nullptr, &errorCode)); |
124 | 4.85k | LocalUResourceBundlePointer data( |
125 | 4.85k | ures_getByKeyWithFallback(collations.getAlias(), type, nullptr, &errorCode)); |
126 | 4.85k | int32_t length; |
127 | 4.85k | const char16_t *s = ures_getStringByKey(data.getAlias(), "Sequence", &length, &errorCode); |
128 | 4.85k | if(U_FAILURE(errorCode)) { return; } |
129 | | |
130 | | // No string pointer aliasing so that we need not hold onto the resource bundle. |
131 | 4.81k | rules.setTo(s, length); |
132 | 4.81k | if(rules.isBogus()) { |
133 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
134 | 0 | } |
135 | 4.81k | } |
136 | | |
137 | | template<> U_I18N_API |
138 | | const CollationCacheEntry * |
139 | | LocaleCacheKey<CollationCacheEntry>::createObject(const void *creationContext, |
140 | 6.59k | UErrorCode &errorCode) const { |
141 | 6.59k | CollationLoader *loader = |
142 | 6.59k | reinterpret_cast<CollationLoader *>( |
143 | 6.59k | const_cast<void *>(creationContext)); |
144 | 6.59k | return loader->createCacheEntry(errorCode); |
145 | 6.59k | } |
146 | | |
147 | | const CollationCacheEntry * |
148 | 14.3k | CollationLoader::loadTailoring(const Locale &locale, UErrorCode &errorCode) { |
149 | 14.3k | const CollationCacheEntry *rootEntry = CollationRoot::getRootCacheEntry(errorCode); |
150 | 14.3k | if(U_FAILURE(errorCode)) { return nullptr; } |
151 | 14.3k | const char *name = locale.getName(); |
152 | 14.3k | if(*name == 0 || uprv_strcmp(name, "root") == 0) { |
153 | | |
154 | | // Have to add a ref. |
155 | 546 | rootEntry->addRef(); |
156 | 546 | return rootEntry; |
157 | 546 | } |
158 | | |
159 | | // Clear warning codes before loading where they get cached. |
160 | 13.8k | errorCode = U_ZERO_ERROR; |
161 | 13.8k | CollationLoader loader(rootEntry, locale, errorCode); |
162 | | |
163 | | // getCacheEntry adds a ref for us. |
164 | 13.8k | return loader.getCacheEntry(errorCode); |
165 | 14.3k | } |
166 | | |
167 | | CollationLoader::CollationLoader(const CollationCacheEntry *re, const Locale &requested, |
168 | | UErrorCode &errorCode) |
169 | 13.8k | : cache(UnifiedCache::getInstance(errorCode)), rootEntry(re), |
170 | 13.8k | validLocale(re->validLocale), locale(requested), |
171 | 13.8k | typesTried(0), typeFallback(false), |
172 | 13.8k | bundle(nullptr), collations(nullptr), data(nullptr) { |
173 | 13.8k | type[0] = 0; |
174 | 13.8k | defaultType[0] = 0; |
175 | 13.8k | if(U_FAILURE(errorCode)) { return; } |
176 | | |
177 | 13.8k | if (locale.isBogus()) { |
178 | 0 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
179 | 0 | return; |
180 | 0 | } |
181 | | // Canonicalize the locale ID: Ignore all irrelevant keywords. |
182 | 13.8k | const char *baseName = locale.getBaseName(); |
183 | 13.8k | if(uprv_strcmp(locale.getName(), baseName) != 0) { |
184 | 2.24k | locale = Locale(baseName); |
185 | | // Due to ICU-22416, we may have bogus locale constructed from |
186 | | // a string of getBaseName(). |
187 | 2.24k | if (locale.isBogus()) { |
188 | 1 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
189 | 1 | return; |
190 | 1 | } |
191 | | |
192 | | // Fetch the collation type from the locale ID. |
193 | 2.24k | int32_t typeLength = requested.getKeywordValue("collation", |
194 | 2.24k | type, UPRV_LENGTHOF(type) - 1, errorCode); |
195 | 2.24k | if(U_FAILURE(errorCode)) { |
196 | 387 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
197 | 387 | return; |
198 | 387 | } |
199 | 1.85k | type[typeLength] = 0; // in case of U_NOT_TERMINATED_WARNING |
200 | 1.85k | if(typeLength == 0) { |
201 | | // No collation type. |
202 | 1.39k | } else if(uprv_stricmp(type, "default") == 0) { |
203 | | // Ignore "default" (case-insensitive). |
204 | 1 | type[0] = 0; |
205 | 465 | } else { |
206 | | // Copy the collation type. |
207 | 465 | T_CString_toLowerCase(type); |
208 | 465 | locale.setKeywordValue("collation", type, errorCode); |
209 | 465 | } |
210 | 1.85k | } |
211 | 13.8k | } |
212 | | |
213 | 13.8k | CollationLoader::~CollationLoader() { |
214 | 13.8k | ures_close(data); |
215 | 13.8k | ures_close(collations); |
216 | 13.8k | ures_close(bundle); |
217 | 13.8k | } |
218 | | |
219 | | const CollationCacheEntry * |
220 | 6.59k | CollationLoader::createCacheEntry(UErrorCode &errorCode) { |
221 | | // This is a linear lookup and fallback flow turned into a state machine. |
222 | | // Most local variables have been turned into instance fields. |
223 | | // In a cache miss, cache.get() calls CacheKey::createObject(), |
224 | | // which means that we progress via recursion. |
225 | | // loadFromCollations() will recurse to itself as well for collation type fallback. |
226 | 6.59k | if(bundle == nullptr) { |
227 | 5.87k | return loadFromLocale(errorCode); |
228 | 5.87k | } else if(collations == nullptr) { |
229 | 463 | return loadFromBundle(errorCode); |
230 | 463 | } else if(data == nullptr) { |
231 | 249 | return loadFromCollations(errorCode); |
232 | 249 | } else { |
233 | 14 | return loadFromData(errorCode); |
234 | 14 | } |
235 | 6.59k | } |
236 | | |
237 | | const CollationCacheEntry * |
238 | 5.87k | CollationLoader::loadFromLocale(UErrorCode &errorCode) { |
239 | 5.87k | if(U_FAILURE(errorCode)) { return nullptr; } |
240 | 5.87k | U_ASSERT(bundle == nullptr); |
241 | 5.87k | bundle = ures_openNoDefault(U_ICUDATA_COLL, locale.getBaseName(), &errorCode); |
242 | 5.87k | if(errorCode == U_MISSING_RESOURCE_ERROR) { |
243 | 0 | errorCode = U_USING_DEFAULT_WARNING; |
244 | | |
245 | | // Have to add that ref that we promise. |
246 | 0 | rootEntry->addRef(); |
247 | 0 | return rootEntry; |
248 | 0 | } |
249 | 5.87k | Locale requestedLocale(locale); |
250 | 5.87k | const char *vLocale = ures_getLocaleByType(bundle, ULOC_ACTUAL_LOCALE, &errorCode); |
251 | 5.87k | if(U_FAILURE(errorCode)) { return nullptr; } |
252 | 5.86k | locale = validLocale = Locale(vLocale); // no type until loadFromCollations() |
253 | 5.86k | if(type[0] != 0) { |
254 | 446 | locale.setKeywordValue("collation", type, errorCode); |
255 | 446 | } |
256 | 5.86k | if(locale != requestedLocale) { |
257 | 5.33k | return getCacheEntry(errorCode); |
258 | 5.33k | } else { |
259 | 537 | return loadFromBundle(errorCode); |
260 | 537 | } |
261 | 5.86k | } |
262 | | |
263 | | const CollationCacheEntry * |
264 | 1.00k | CollationLoader::loadFromBundle(UErrorCode &errorCode) { |
265 | 1.00k | if(U_FAILURE(errorCode)) { return nullptr; } |
266 | 1.00k | U_ASSERT(collations == nullptr); |
267 | | // There are zero or more tailorings in the collations table. |
268 | 1.00k | collations = ures_getByKey(bundle, "collations", nullptr, &errorCode); |
269 | 1.00k | if(errorCode == U_MISSING_RESOURCE_ERROR) { |
270 | 0 | errorCode = U_USING_DEFAULT_WARNING; |
271 | | // Return the root tailoring with the validLocale, without collation type. |
272 | 0 | return makeCacheEntryFromRoot(validLocale, errorCode); |
273 | 0 | } |
274 | 1.00k | if(U_FAILURE(errorCode)) { return nullptr; } |
275 | | |
276 | | // Fetch the default type from the data. |
277 | 1.00k | { |
278 | 1.00k | UErrorCode internalErrorCode = U_ZERO_ERROR; |
279 | 1.00k | LocalUResourceBundlePointer def( |
280 | 1.00k | ures_getByKeyWithFallback(collations, "default", nullptr, &internalErrorCode)); |
281 | 1.00k | int32_t length; |
282 | 1.00k | const char16_t *s = ures_getString(def.getAlias(), &length, &internalErrorCode); |
283 | 1.00k | if(U_SUCCESS(internalErrorCode) && 0 < length && length < UPRV_LENGTHOF(defaultType)) { |
284 | 1.00k | u_UCharsToChars(s, defaultType, length + 1); |
285 | 1.00k | } else { |
286 | 0 | uprv_strcpy(defaultType, "standard"); |
287 | 0 | } |
288 | 1.00k | } |
289 | | |
290 | | // Record which collation types we have looked for already, |
291 | | // so that we do not deadlock in the cache. |
292 | | // |
293 | | // If there is no explicit type, then we look in the cache |
294 | | // for the entry with the default type. |
295 | | // If the explicit type is the default type, then we do not look in the cache |
296 | | // for the entry with an empty type. |
297 | | // Otherwise, two concurrent requests with opposite fallbacks would deadlock each other. |
298 | | // Also, it is easier to always enter the next method with a non-empty type. |
299 | 1.00k | if(type[0] == 0) { |
300 | 652 | uprv_strcpy(type, defaultType); |
301 | 652 | typesTried |= TRIED_DEFAULT; |
302 | 652 | if(uprv_strcmp(type, "search") == 0) { |
303 | 0 | typesTried |= TRIED_SEARCH; |
304 | 0 | } |
305 | 652 | if(uprv_strcmp(type, "standard") == 0) { |
306 | 612 | typesTried |= TRIED_STANDARD; |
307 | 612 | } |
308 | 652 | locale.setKeywordValue("collation", type, errorCode); |
309 | 652 | return getCacheEntry(errorCode); |
310 | 652 | } else { |
311 | 348 | if(uprv_strcmp(type, defaultType) == 0) { |
312 | 0 | typesTried |= TRIED_DEFAULT; |
313 | 0 | } |
314 | 348 | if(uprv_strcmp(type, "search") == 0) { |
315 | 4 | typesTried |= TRIED_SEARCH; |
316 | 4 | } |
317 | 348 | if(uprv_strcmp(type, "standard") == 0) { |
318 | 1 | typesTried |= TRIED_STANDARD; |
319 | 1 | } |
320 | 348 | return loadFromCollations(errorCode); |
321 | 348 | } |
322 | 1.00k | } |
323 | | |
324 | | const CollationCacheEntry * |
325 | 597 | CollationLoader::loadFromCollations(UErrorCode &errorCode) { |
326 | 597 | if(U_FAILURE(errorCode)) { return nullptr; } |
327 | 597 | U_ASSERT(data == nullptr); |
328 | | // Load the collations/type tailoring, with type fallback. |
329 | 597 | LocalUResourceBundlePointer localData( |
330 | 597 | ures_getByKeyWithFallback(collations, type, nullptr, &errorCode)); |
331 | 597 | int32_t typeLength = static_cast<int32_t>(uprv_strlen(type)); |
332 | 597 | if(errorCode == U_MISSING_RESOURCE_ERROR) { |
333 | 322 | errorCode = U_USING_DEFAULT_WARNING; |
334 | 322 | typeFallback = true; |
335 | 322 | if((typesTried & TRIED_SEARCH) == 0 && |
336 | 322 | typeLength > 6 && uprv_strncmp(type, "search", 6) == 0) { |
337 | | // fall back from something like "searchjl" to "search" |
338 | 4 | typesTried |= TRIED_SEARCH; |
339 | 4 | type[6] = 0; |
340 | 318 | } else if((typesTried & TRIED_DEFAULT) == 0) { |
341 | | // fall back to the default type |
342 | 318 | typesTried |= TRIED_DEFAULT; |
343 | 318 | uprv_strcpy(type, defaultType); |
344 | 318 | } else if((typesTried & TRIED_STANDARD) == 0) { |
345 | | // fall back to the "standard" type |
346 | 0 | typesTried |= TRIED_STANDARD; |
347 | 0 | uprv_strcpy(type, "standard"); |
348 | 0 | } else { |
349 | | // Return the root tailoring with the validLocale, without collation type. |
350 | 0 | return makeCacheEntryFromRoot(validLocale, errorCode); |
351 | 0 | } |
352 | 322 | locale.setKeywordValue("collation", type, errorCode); |
353 | 322 | return getCacheEntry(errorCode); |
354 | 322 | } |
355 | 275 | if(U_FAILURE(errorCode)) { return nullptr; } |
356 | | |
357 | 275 | data = localData.orphan(); |
358 | 275 | const char *actualLocale = ures_getLocaleByType(data, ULOC_ACTUAL_LOCALE, &errorCode); |
359 | 275 | if(U_FAILURE(errorCode)) { return nullptr; } |
360 | 275 | const char *vLocale = validLocale.getBaseName(); |
361 | 275 | UBool actualAndValidLocalesAreDifferent = Locale(actualLocale) != Locale(vLocale); |
362 | | |
363 | | // Set the collation types on the informational locales, |
364 | | // except when they match the default types (for brevity and backwards compatibility). |
365 | | // For the valid locale, suppress the default type. |
366 | 275 | if(uprv_strcmp(type, defaultType) != 0) { |
367 | 30 | validLocale.setKeywordValue("collation", type, errorCode); |
368 | 30 | if(U_FAILURE(errorCode)) { return nullptr; } |
369 | 30 | } |
370 | | |
371 | | // Is this the same as the root collator? If so, then use that instead. |
372 | 275 | if((*actualLocale == 0 || uprv_strcmp(actualLocale, "root") == 0) && |
373 | 275 | uprv_strcmp(type, "standard") == 0) { |
374 | 36 | if(typeFallback) { |
375 | 0 | errorCode = U_USING_DEFAULT_WARNING; |
376 | 0 | } |
377 | 36 | return makeCacheEntryFromRoot(validLocale, errorCode); |
378 | 36 | } |
379 | | |
380 | 239 | locale = Locale(actualLocale); |
381 | 239 | if(actualAndValidLocalesAreDifferent) { |
382 | 70 | locale.setKeywordValue("collation", type, errorCode); |
383 | 70 | const CollationCacheEntry *entry = getCacheEntry(errorCode); |
384 | 70 | return makeCacheEntry(validLocale, entry, errorCode); |
385 | 169 | } else { |
386 | 169 | return loadFromData(errorCode); |
387 | 169 | } |
388 | 239 | } |
389 | | |
390 | | const CollationCacheEntry * |
391 | 183 | CollationLoader::loadFromData(UErrorCode &errorCode) { |
392 | 183 | if(U_FAILURE(errorCode)) { return nullptr; } |
393 | 183 | LocalPointer<CollationTailoring> t(new CollationTailoring(rootEntry->tailoring->settings)); |
394 | 183 | if(t.isNull() || t->isBogus()) { |
395 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
396 | 0 | return nullptr; |
397 | 0 | } |
398 | | |
399 | | // deserialize |
400 | 183 | LocalUResourceBundlePointer binary(ures_getByKey(data, "%%CollationBin", nullptr, &errorCode)); |
401 | | // Note: U_MISSING_RESOURCE_ERROR --> The old code built from rules if available |
402 | | // but that created undesirable dependencies. |
403 | 183 | int32_t length = 0; |
404 | 183 | const uint8_t *inBytes = ures_getBinary(binary.getAlias(), &length, &errorCode); |
405 | 183 | CollationDataReader::read(rootEntry->tailoring, inBytes, length, *t, errorCode); |
406 | | // Note: U_COLLATOR_VERSION_MISMATCH --> The old code built from rules if available |
407 | | // but that created undesirable dependencies. |
408 | 183 | if(U_FAILURE(errorCode)) { return nullptr; } |
409 | | |
410 | | // Try to fetch the optional rules string. |
411 | 180 | { |
412 | 180 | UErrorCode internalErrorCode = U_ZERO_ERROR; |
413 | 180 | int32_t len; |
414 | 180 | const char16_t *s = ures_getStringByKey(data, "Sequence", &len, |
415 | 180 | &internalErrorCode); |
416 | 180 | if(U_SUCCESS(internalErrorCode)) { |
417 | 180 | t->rules.setTo(true, s, len); |
418 | 180 | } |
419 | 180 | } |
420 | | |
421 | 180 | const char *actualLocale = locale.getBaseName(); // without type |
422 | 180 | const char *vLocale = validLocale.getBaseName(); |
423 | 180 | UBool actualAndValidLocalesAreDifferent = Locale(actualLocale) != Locale(vLocale); |
424 | | |
425 | | // For the actual locale, suppress the default type *according to the actual locale*. |
426 | | // For example, zh has default=pinyin and contains all of the Chinese tailorings. |
427 | | // zh_Hant has default=stroke but has no other data. |
428 | | // For the valid locale "zh_Hant" we need to suppress stroke. |
429 | | // For the actual locale "zh" we need to suppress pinyin instead. |
430 | 180 | if(actualAndValidLocalesAreDifferent) { |
431 | | // Opening a bundle for the actual locale should always succeed. |
432 | 13 | LocalUResourceBundlePointer actualBundle( |
433 | 13 | ures_open(U_ICUDATA_COLL, actualLocale, &errorCode)); |
434 | 13 | if(U_FAILURE(errorCode)) { return nullptr; } |
435 | 13 | UErrorCode internalErrorCode = U_ZERO_ERROR; |
436 | 13 | LocalUResourceBundlePointer def( |
437 | 13 | ures_getByKeyWithFallback(actualBundle.getAlias(), "collations/default", nullptr, |
438 | 13 | &internalErrorCode)); |
439 | 13 | int32_t len; |
440 | 13 | const char16_t *s = ures_getString(def.getAlias(), &len, &internalErrorCode); |
441 | 13 | if(U_SUCCESS(internalErrorCode) && len < UPRV_LENGTHOF(defaultType)) { |
442 | 13 | u_UCharsToChars(s, defaultType, len + 1); |
443 | 13 | } else { |
444 | 0 | uprv_strcpy(defaultType, "standard"); |
445 | 0 | } |
446 | 13 | } |
447 | 180 | t->actualLocale = locale; |
448 | 180 | if(uprv_strcmp(type, defaultType) != 0) { |
449 | 19 | t->actualLocale.setKeywordValue("collation", type, errorCode); |
450 | 161 | } else if(uprv_strcmp(locale.getName(), locale.getBaseName()) != 0) { |
451 | | // Remove the collation keyword if it was set. |
452 | 4 | t->actualLocale.setKeywordValue("collation", nullptr, errorCode); |
453 | 4 | } |
454 | 180 | if(U_FAILURE(errorCode)) { return nullptr; } |
455 | | |
456 | 180 | if(typeFallback) { |
457 | 4 | errorCode = U_USING_DEFAULT_WARNING; |
458 | 4 | } |
459 | 180 | t->bundle = bundle; |
460 | 180 | bundle = nullptr; |
461 | 180 | const CollationCacheEntry *entry = new CollationCacheEntry(validLocale, t.getAlias()); |
462 | 180 | if(entry == nullptr) { |
463 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
464 | 0 | return nullptr; |
465 | 180 | } else { |
466 | 180 | t.orphan(); |
467 | 180 | } |
468 | | // Have to add that reference that we promise. |
469 | 180 | entry->addRef(); |
470 | 180 | return entry; |
471 | 180 | } |
472 | | |
473 | | const CollationCacheEntry * |
474 | 20.2k | CollationLoader::getCacheEntry(UErrorCode &errorCode) { |
475 | 20.2k | LocaleCacheKey<CollationCacheEntry> key(locale); |
476 | 20.2k | const CollationCacheEntry *entry = nullptr; |
477 | 20.2k | cache->get(key, this, entry, errorCode); |
478 | 20.2k | return entry; |
479 | 20.2k | } |
480 | | |
481 | | const CollationCacheEntry * |
482 | | CollationLoader::makeCacheEntryFromRoot( |
483 | | const Locale &/*loc*/, |
484 | 36 | UErrorCode &errorCode) const { |
485 | 36 | if (U_FAILURE(errorCode)) { |
486 | 0 | return nullptr; |
487 | 0 | } |
488 | 36 | rootEntry->addRef(); |
489 | 36 | return makeCacheEntry(validLocale, rootEntry, errorCode); |
490 | 36 | } |
491 | | |
492 | | const CollationCacheEntry * |
493 | | CollationLoader::makeCacheEntry( |
494 | | const Locale &loc, |
495 | | const CollationCacheEntry *entryFromCache, |
496 | 106 | UErrorCode &errorCode) { |
497 | 106 | if(U_FAILURE(errorCode) || loc == entryFromCache->validLocale) { |
498 | 40 | return entryFromCache; |
499 | 40 | } |
500 | 66 | CollationCacheEntry *entry = new CollationCacheEntry(loc, entryFromCache->tailoring); |
501 | 66 | if(entry == nullptr) { |
502 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
503 | 0 | entryFromCache->removeRef(); |
504 | 0 | return nullptr; |
505 | 0 | } |
506 | 66 | entry->addRef(); |
507 | 66 | entryFromCache->removeRef(); |
508 | 66 | return entry; |
509 | 66 | } |
510 | | |
511 | | U_NAMESPACE_END |
512 | | |
513 | | U_NAMESPACE_USE |
514 | | |
515 | | U_CAPI UCollator* |
516 | | ucol_open(const char *loc, |
517 | | UErrorCode *status) |
518 | 0 | { |
519 | 0 | UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN); |
520 | 0 | UTRACE_DATA1(UTRACE_INFO, "locale = \"%s\"", loc); |
521 | 0 | UCollator *result = nullptr; |
522 | |
|
523 | 0 | Collator *coll = Collator::createInstance(loc, *status); |
524 | 0 | if(U_SUCCESS(*status)) { |
525 | 0 | result = coll->toUCollator(); |
526 | 0 | } |
527 | 0 | UTRACE_EXIT_PTR_STATUS(result, *status); |
528 | 0 | return result; |
529 | 0 | } |
530 | | |
531 | | |
532 | | U_CAPI int32_t U_EXPORT2 |
533 | | ucol_getDisplayName( const char *objLoc, |
534 | | const char *dispLoc, |
535 | | char16_t *result, |
536 | | int32_t resultLength, |
537 | | UErrorCode *status) |
538 | 0 | { |
539 | 0 | if(U_FAILURE(*status)) return -1; |
540 | 0 | UnicodeString dst; |
541 | 0 | if(!(result==nullptr && resultLength==0)) { |
542 | | // nullptr destination for pure preflighting: empty dummy string |
543 | | // otherwise, alias the destination buffer |
544 | 0 | dst.setTo(result, 0, resultLength); |
545 | 0 | } |
546 | 0 | Collator::getDisplayName(Locale(objLoc), Locale(dispLoc), dst); |
547 | 0 | return dst.extract(result, resultLength, *status); |
548 | 0 | } |
549 | | |
550 | | U_CAPI const char* U_EXPORT2 |
551 | | ucol_getAvailable(int32_t index) |
552 | 0 | { |
553 | 0 | int32_t count = 0; |
554 | 0 | const Locale *loc = Collator::getAvailableLocales(count); |
555 | 0 | if (loc != nullptr && index < count) { |
556 | 0 | return loc[index].getName(); |
557 | 0 | } |
558 | 0 | return nullptr; |
559 | 0 | } |
560 | | |
561 | | U_CAPI int32_t U_EXPORT2 |
562 | | ucol_countAvailable() |
563 | 0 | { |
564 | 0 | int32_t count = 0; |
565 | 0 | Collator::getAvailableLocales(count); |
566 | 0 | return count; |
567 | 0 | } |
568 | | |
569 | | #if !UCONFIG_NO_SERVICE |
570 | | U_CAPI UEnumeration* U_EXPORT2 |
571 | 0 | ucol_openAvailableLocales(UErrorCode *status) { |
572 | | // This is a wrapper over Collator::getAvailableLocales() |
573 | 0 | if (U_FAILURE(*status)) { |
574 | 0 | return nullptr; |
575 | 0 | } |
576 | 0 | StringEnumeration *s = icu::Collator::getAvailableLocales(); |
577 | 0 | if (s == nullptr) { |
578 | 0 | *status = U_MEMORY_ALLOCATION_ERROR; |
579 | 0 | return nullptr; |
580 | 0 | } |
581 | 0 | return uenum_openFromStringEnumeration(s, status); |
582 | 0 | } |
583 | | #endif |
584 | | |
585 | | // Note: KEYWORDS[0] != RESOURCE_NAME - alan |
586 | | |
587 | | static const char RESOURCE_NAME[] = "collations"; |
588 | | |
589 | | static const char* const KEYWORDS[] = { "collation" }; |
590 | | |
591 | 0 | #define KEYWORD_COUNT UPRV_LENGTHOF(KEYWORDS) |
592 | | |
593 | | U_CAPI UEnumeration* U_EXPORT2 |
594 | 0 | ucol_getKeywords(UErrorCode *status) { |
595 | 0 | UEnumeration *result = nullptr; |
596 | 0 | if (U_SUCCESS(*status)) { |
597 | 0 | return uenum_openCharStringsEnumeration(KEYWORDS, KEYWORD_COUNT, status); |
598 | 0 | } |
599 | 0 | return result; |
600 | 0 | } |
601 | | |
602 | | U_CAPI UEnumeration* U_EXPORT2 |
603 | 0 | ucol_getKeywordValues(const char *keyword, UErrorCode *status) { |
604 | 0 | if (U_FAILURE(*status)) { |
605 | 0 | return nullptr; |
606 | 0 | } |
607 | | // hard-coded to accept exactly one collation keyword |
608 | | // modify if additional collation keyword is added later |
609 | 0 | if (keyword==nullptr || uprv_strcmp(keyword, KEYWORDS[0])!=0) |
610 | 0 | { |
611 | 0 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
612 | 0 | return nullptr; |
613 | 0 | } |
614 | 0 | return ures_getKeywordValues(U_ICUDATA_COLL, RESOURCE_NAME, status); |
615 | 0 | } |
616 | | |
617 | | static const UEnumeration defaultKeywordValues = { |
618 | | nullptr, |
619 | | nullptr, |
620 | | ulist_close_keyword_values_iterator, |
621 | | ulist_count_keyword_values, |
622 | | uenum_unextDefault, |
623 | | ulist_next_keyword_value, |
624 | | ulist_reset_keyword_values_iterator |
625 | | }; |
626 | | |
627 | | namespace { |
628 | | |
629 | | struct KeywordsSink : public ResourceSink { |
630 | | public: |
631 | | KeywordsSink(UErrorCode &errorCode) : |
632 | 0 | values(ulist_createEmptyList(&errorCode)), hasDefault(false) {} |
633 | | virtual ~KeywordsSink(); |
634 | | |
635 | | virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/, |
636 | 0 | UErrorCode &errorCode) override { |
637 | 0 | if (U_FAILURE(errorCode)) { return; } |
638 | 0 | ResourceTable collations = value.getTable(errorCode); |
639 | 0 | for (int32_t i = 0; collations.getKeyAndValue(i, key, value); ++i) { |
640 | 0 | UResType type = value.getType(); |
641 | 0 | if (type == URES_STRING) { |
642 | 0 | if (!hasDefault && uprv_strcmp(key, "default") == 0) { |
643 | 0 | CharString defcoll; |
644 | 0 | defcoll.appendInvariantChars(value.getUnicodeString(errorCode), errorCode); |
645 | 0 | if (U_SUCCESS(errorCode) && !defcoll.isEmpty()) { |
646 | 0 | char *ownedDefault = uprv_strdup(defcoll.data()); |
647 | 0 | if (ownedDefault == nullptr) { |
648 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
649 | 0 | return; |
650 | 0 | } |
651 | 0 | ulist_removeString(values, defcoll.data()); |
652 | 0 | ulist_addItemBeginList(values, ownedDefault, true, &errorCode); |
653 | 0 | hasDefault = true; |
654 | 0 | } |
655 | 0 | } |
656 | 0 | } else if (type == URES_TABLE && uprv_strncmp(key, "private-", 8) != 0) { |
657 | 0 | if (!ulist_containsString(values, key, static_cast<int32_t>(uprv_strlen(key)))) { |
658 | 0 | ulist_addItemEndList(values, key, false, &errorCode); |
659 | 0 | } |
660 | 0 | } |
661 | 0 | if (U_FAILURE(errorCode)) { return; } |
662 | 0 | } |
663 | 0 | } |
664 | | |
665 | | UList *values; |
666 | | UBool hasDefault; |
667 | | }; |
668 | | |
669 | 0 | KeywordsSink::~KeywordsSink() { |
670 | 0 | ulist_deleteList(values); |
671 | 0 | } |
672 | | |
673 | | } // namespace |
674 | | |
675 | | U_CAPI UEnumeration* U_EXPORT2 |
676 | | ucol_getKeywordValuesForLocale(const char* /*key*/, const char* locale, |
677 | 0 | UBool /*commonlyUsed*/, UErrorCode* status) { |
678 | | // Note: The parameter commonlyUsed is not used. |
679 | | // The switch is in the method signature for consistency |
680 | | // with other locale services. |
681 | | |
682 | | // Read available collation values from collation bundles. |
683 | 0 | LocalUResourceBundlePointer bundle(ures_open(U_ICUDATA_COLL, locale, status)); |
684 | 0 | KeywordsSink sink(*status); |
685 | 0 | ures_getAllItemsWithFallback(bundle.getAlias(), RESOURCE_NAME, sink, *status); |
686 | 0 | if (U_FAILURE(*status)) { return nullptr; } |
687 | | |
688 | 0 | UEnumeration *en = (UEnumeration *)uprv_malloc(sizeof(UEnumeration)); |
689 | 0 | if (en == nullptr) { |
690 | 0 | *status = U_MEMORY_ALLOCATION_ERROR; |
691 | 0 | return nullptr; |
692 | 0 | } |
693 | 0 | memcpy(en, &defaultKeywordValues, sizeof(UEnumeration)); |
694 | 0 | ulist_resetList(sink.values); // Initialize the iterator. |
695 | 0 | en->context = sink.values; |
696 | 0 | sink.values = nullptr; // Avoid deletion in the sink destructor. |
697 | 0 | return en; |
698 | 0 | } |
699 | | |
700 | | U_CAPI int32_t U_EXPORT2 |
701 | | ucol_getFunctionalEquivalent(char* result, int32_t resultCapacity, |
702 | | const char* keyword, const char* locale, |
703 | | UBool* isAvailable, UErrorCode* status) |
704 | 0 | { |
705 | | // N.B.: Resource name is "collations" but keyword is "collation" |
706 | 0 | return ures_getFunctionalEquivalent(result, resultCapacity, U_ICUDATA_COLL, |
707 | 0 | "collations", keyword, locale, |
708 | 0 | isAvailable, true, status); |
709 | 0 | } |
710 | | |
711 | | #endif /* #if !UCONFIG_NO_COLLATION */ |