1// © 2016 and later: Unicode, Inc. and others.
2// License & terms of use: http://www.unicode.org/copyright.html
3/*
4**********************************************************************
5* Copyright (C) 1997-2013, International Business Machines
6* Corporation and others. All Rights Reserved.
7**********************************************************************
8*
9* File resbund.cpp
10*
11* Modification History:
12*
13* Date Name Description
14* 02/05/97 aliu Fixed bug in chopLocale. Added scanForLocaleInFile
15* based on code taken from scanForLocale. Added
16* constructor which attempts to read resource bundle
17* from a specific file, without searching other files.
18* 02/11/97 aliu Added UErrorCode return values to constructors. Fixed
19* infinite loops in scanForFile and scanForLocale.
20* Modified getRawResourceData to not delete storage in
21* localeData and resourceData which it doesn't own.
22* Added Mac compatibility #ifdefs for tellp() and
23* ios::nocreate.
24* 03/04/97 aliu Modified to use ExpandingDataSink objects instead of
25* the highly inefficient ostrstream objects.
26* 03/13/97 aliu Rewrote to load in entire resource bundle and store
27* it as a Hashtable of ResourceBundleData objects.
28* Added state table to govern parsing of files.
29* Modified to load locale index out of new file distinct
30* from default.txt.
31* 03/25/97 aliu Modified to support 2-d arrays, needed for timezone data.
32* Added support for custom file suffixes. Again, needed
33* to support timezone data. Improved error handling to
34* detect duplicate tags and subtags.
35* 04/07/97 aliu Fixed bug in getHashtableForLocale(). Fixed handling
36* of failing UErrorCode values on entry to API methods.
37* Fixed bugs in getArrayItem() for negative indices.
38* 04/29/97 aliu Update to use new Hashtable deletion protocol.
39* 05/06/97 aliu Flattened kTransitionTable for HP compiler.
40* Fixed usage of CharString.
41* 06/11/99 stephen Removed parsing of .txt files.
42* Reworked to use new binary format.
43* Cleaned up.
44* 06/14/99 stephen Removed methods taking a filename suffix.
45* 06/22/99 stephen Added missing T_FileStream_close in parse()
46* 11/09/99 weiv Added getLocale(), rewritten constructForLocale()
47* March 2000 weiv complete overhaul.
48******************************************************************************
49*/
50
51#include "unicode/utypes.h"
52#include "unicode/resbund.h"
53
54#include "cmemory.h"
55#include "mutex.h"
56#include "uassert.h"
57#include "umutex.h"
58
59#include "uresimp.h"
60
61U_NAMESPACE_BEGIN
62
63/*-----------------------------------------------------------------------------
64 * Implementation Notes
65 *
66 * Resource bundles are read in once, and thereafter cached.
67 * ResourceBundle statically keeps track of which files have been
68 * read, so we are guaranteed that each file is read at most once.
69 * Resource bundles can be loaded from different data directories and
70 * will be treated as distinct, even if they are for the same locale.
71 *
72 * Resource bundles are lightweight objects, which have pointers to
73 * one or more shared Hashtable objects containing all the data.
74 * Copying would be cheap, but there is no copy constructor, since
75 * there wasn't one in the original API.
76 *
77 * The ResourceBundle parsing mechanism is implemented as a transition
78 * network, for easy maintenance and modification. The network is
79 * implemented as a matrix (instead of in code) to make this even
80 * easier. The matrix contains Transition objects. Each Transition
81 * object describes a destination node and an action to take before
82 * moving to the destination node. The source node is encoded by the
83 * index of the object in the array that contains it. The pieces
84 * needed to understand the transition network are the enums for node
85 * IDs and actions, the parse() method, which walks through the
86 * network and implements the actions, and the network itself. The
87 * network guarantees certain conditions, for example, that a new
88 * resource will not be closed until one has been opened first; or
89 * that data will not be stored into a TaggedList until a TaggedList
90 * has been created. Nonetheless, the code in parse() does some
91 * consistency checks as it runs the network, and fails with an
92 * U_INTERNAL_PROGRAM_ERROR if one of these checks fails. If the input
93 * data has a bad format, an U_INVALID_FORMAT_ERROR is returned. If you
94 * see an U_INTERNAL_PROGRAM_ERROR the transition matrix has a bug in
95 * it.
96 *
97 * Old functionality of multiple locales in a single file is still
98 * supported. For this reason, LOCALE names override FILE names. If
99 * data for en_US is located in the en.txt file, once it is loaded,
100 * the code will not care where it came from (other than remembering
101 * which directory it came from). However, if there is an en_US
102 * resource in en_US.txt, that will take precedence. There is no
103 * limit to the number or type of resources that can be stored in a
104 * file, however, files are only searched in a specific way. If
105 * en_US_CA is requested, then first en_US_CA.txt is searched, then
106 * en_US.txt, then en.txt, then default.txt. So it only makes sense
107 * to put certain locales in certain files. In this example, it would
108 * be logical to put en_US_CA, en_US, and en into the en.txt file,
109 * since they would be found there if asked for. The extreme example
110 * is to place all locale resources into default.txt, which should
111 * also work.
112 *
113 * Inheritance is implemented. For example, xx_YY_zz inherits as
114 * follows: xx_YY_zz, xx_YY, xx, default. Inheritance is implemented
115 * as an array of hashtables. There will be from 1 to 4 hashtables in
116 * the array.
117 *
118 * Fallback files are implemented. The fallback pattern is Language
119 * Country Variant (LCV) -> LC -> L. Fallback is first done for the
120 * requested locale. Then it is done for the default locale, as
121 * returned by Locale::getDefault(). Then the special file
122 * default.txt is searched for the default locale. The overall FILE
123 * fallback path is LCV -> LC -> L -> dLCV -> dLC -> dL -> default.
124 *
125 * Note that although file name searching includes the default locale,
126 * once a ResourceBundle object is constructed, the inheritance path
127 * no longer includes the default locale. The path is LCV -> LC -> L
128 * -> default.
129 *
130 * File parsing is lazy. Nothing is parsed unless it is called for by
131 * someone. So when a ResourceBundle for xx_YY_zz is constructed,
132 * only that locale is parsed (along with anything else in the same
133 * file). Later, if the FooBar tag is asked for, and if it isn't
134 * found in xx_YY_zz, then xx_YY.txt will be parsed and checked, and
135 * so forth, until the chain is exhausted or the tag is found.
136 *
137 * Thread-safety is implemented around caches, both the cache that
138 * stores all the resource data, and the cache that stores flags
139 * indicating whether or not a file has been visited. These caches
140 * delete their storage at static cleanup time, when the process
141 * quits.
142 *
143 * ResourceBundle supports TableCollation as a special case. This
144 * involves having special ResourceBundle objects which DO own their
145 * data, since we don't want large collation rule strings in the
146 * ResourceBundle cache (these are already cached in the
147 * TableCollation cache). TableCollation files (.ctx files) have the
148 * same format as normal resource data files, with a different
149 * interpretation, from the standpoint of ResourceBundle. .ctx files
150 * are loaded into otherwise ordinary ResourceBundle objects. They
151 * don't inherit (that's implemented by TableCollation) and they own
152 * their data (as mentioned above). However, they still support
153 * possible multiple locales in a single .ctx file. (This is in
154 * practice a bad idea, since you only want the one locale you're
155 * looking for, and only one tag will be present
156 * ("CollationElements"), so you don't need an inheritance chain of
157 * multiple locales.) Up to 4 locale resources will be loaded from a
158 * .ctx file; everything after the first 4 is ignored (parsed and
159 * deleted). (Normal .txt files have no limit.) Instead of being
160 * loaded into the cache, and then looked up as needed, the locale
161 * resources are read straight into the ResourceBundle object.
162 *
163 * The Index, which used to reside in default.txt, has been moved to a
164 * new file, index.txt. This file contains a slightly modified format
165 * with the addition of the "InstalledLocales" tag; it looks like:
166 *
167 * Index {
168 * InstalledLocales {
169 * ar
170 * ..
171 * zh_TW
172 * }
173 * }
174 */
175//-----------------------------------------------------------------------------
176
177UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ResourceBundle)
178
179ResourceBundle::ResourceBundle(UErrorCode &err)
180 :UObject(), fLocale(nullptr)
181{
182 fResource = ures_open(0, Locale::getDefault().getName(), &err);
183}
184
185ResourceBundle::ResourceBundle(const ResourceBundle &other)
186 :UObject(other), fLocale(nullptr)
187{
188 UErrorCode status = U_ZERO_ERROR;
189
190 if (other.fResource) {
191 fResource = ures_copyResb(0, other.fResource, &status);
192 } else {
193 /* Copying a bad resource bundle */
194 fResource = nullptr;
195 }
196}
197
198ResourceBundle::ResourceBundle(UResourceBundle *res, UErrorCode& err)
199 :UObject(), fLocale(nullptr)
200{
201 if (res) {
202 fResource = ures_copyResb(0, res, &err);
203 } else {
204 /* Copying a bad resource bundle */
205 fResource = nullptr;
206 }
207}
208
209ResourceBundle::ResourceBundle(const char* path, const Locale& locale, UErrorCode& err)
210 :UObject(), fLocale(nullptr)
211{
212 fResource = ures_open(path, locale.getName(), &err);
213}
214
215
216ResourceBundle& ResourceBundle::operator=(const ResourceBundle& other)
217{
218 if(this == &other) {
219 return *this;
220 }
221 if(fResource != 0) {
222 ures_close(fResource);
223 fResource = nullptr;
224 }
225 if (fLocale != nullptr) {
226 delete fLocale;
227 fLocale = nullptr;
228 }
229 UErrorCode status = U_ZERO_ERROR;
230 if (other.fResource) {
231 fResource = ures_copyResb(0, other.fResource, &status);
232 } else {
233 /* Copying a bad resource bundle */
234 fResource = nullptr;
235 }
236 return *this;
237}
238
239ResourceBundle::~ResourceBundle()
240{
241 if(fResource != 0) {
242 ures_close(fResource);
243 }
244 if(fLocale != nullptr) {
245 delete(fLocale);
246 }
247}
248
249ResourceBundle *
250ResourceBundle::clone() const {
251 return new ResourceBundle(*this);
252}
253
254UnicodeString ResourceBundle::getString(UErrorCode& status) const {
255 int32_t len = 0;
256 const char16_t *r = ures_getString(fResource, &len, &status);
257 return UnicodeString(true, r, len);
258}
259
260const uint8_t *ResourceBundle::getBinary(int32_t& len, UErrorCode& status) const {
261 return ures_getBinary(fResource, &len, &status);
262}
263
264const int32_t *ResourceBundle::getIntVector(int32_t& len, UErrorCode& status) const {
265 return ures_getIntVector(fResource, &len, &status);
266}
267
268uint32_t ResourceBundle::getUInt(UErrorCode& status) const {
269 return ures_getUInt(fResource, &status);
270}
271
272int32_t ResourceBundle::getInt(UErrorCode& status) const {
273 return ures_getInt(fResource, &status);
274}
275
276const char *ResourceBundle::getName() const {
277 return ures_getName(fResource);
278}
279
280const char *ResourceBundle::getKey() const {
281 return ures_getKey(fResource);
282}
283
284UResType ResourceBundle::getType() const {
285 return ures_getType(fResource);
286}
287
288int32_t ResourceBundle::getSize() const {
289 return ures_getSize(fResource);
290}
291
292UBool ResourceBundle::hasNext() const {
293 return ures_hasNext(fResource);
294}
295
296void ResourceBundle::resetIterator() {
297 ures_resetIterator(fResource);
298}
299
300ResourceBundle ResourceBundle::getNext(UErrorCode& status) {
301 UResourceBundle r;
302
303 ures_initStackObject(&r);
304 ures_getNextResource(fResource, &r, &status);
305 ResourceBundle res(&r, status);
306 if (U_SUCCESS(status)) {
307 ures_close(&r);
308 }
309 return res;
310}
311
312UnicodeString ResourceBundle::getNextString(UErrorCode& status) {
313 int32_t len = 0;
314 const char16_t* r = ures_getNextString(fResource, &len, 0, &status);
315 return UnicodeString(true, r, len);
316}
317
318UnicodeString ResourceBundle::getNextString(const char ** key, UErrorCode& status) {
319 int32_t len = 0;
320 const char16_t* r = ures_getNextString(fResource, &len, key, &status);
321 return UnicodeString(true, r, len);
322}
323
324ResourceBundle ResourceBundle::get(int32_t indexR, UErrorCode& status) const {
325 UResourceBundle r;
326
327 ures_initStackObject(&r);
328 ures_getByIndex(fResource, indexR, &r, &status);
329 ResourceBundle res(&r, status);
330 if (U_SUCCESS(status)) {
331 ures_close(&r);
332 }
333 return res;
334}
335
336UnicodeString ResourceBundle::getStringEx(int32_t indexS, UErrorCode& status) const {
337 int32_t len = 0;
338 const char16_t* r = ures_getStringByIndex(fResource, indexS, &len, &status);
339 return UnicodeString(true, r, len);
340}
341
342ResourceBundle ResourceBundle::get(const char* key, UErrorCode& status) const {
343 UResourceBundle r;
344
345 ures_initStackObject(&r);
346 ures_getByKey(fResource, key, &r, &status);
347 ResourceBundle res(&r, status);
348 if (U_SUCCESS(status)) {
349 ures_close(&r);
350 }
351 return res;
352}
353
354ResourceBundle ResourceBundle::getWithFallback(const char* key, UErrorCode& status){
355 UResourceBundle r;
356 ures_initStackObject(&r);
357 ures_getByKeyWithFallback(fResource, key, &r, &status);
358 ResourceBundle res(&r, status);
359 if(U_SUCCESS(status)){
360 ures_close(&r);
361 }
362 return res;
363}
364UnicodeString ResourceBundle::getStringEx(const char* key, UErrorCode& status) const {
365 int32_t len = 0;
366 const char16_t* r = ures_getStringByKey(fResource, key, &len, &status);
367 return UnicodeString(true, r, len);
368}
369
370const char*
371ResourceBundle::getVersionNumber() const
372{
373 return ures_getVersionNumberInternal(fResource);
374}
375
376void ResourceBundle::getVersion(UVersionInfo versionInfo) const {
377 ures_getVersion(fResource, versionInfo);
378}
379
380const Locale &ResourceBundle::getLocale() const {
381 static UMutex gLocaleLock;
382 Mutex lock(&gLocaleLock);
383 if (fLocale != nullptr) {
384 return *fLocale;
385 }
386 UErrorCode status = U_ZERO_ERROR;
387 const char *localeName = ures_getLocaleInternal(fResource, &status);
388 ResourceBundle *ncThis = const_cast<ResourceBundle *>(this);
389 ncThis->fLocale = new Locale(localeName);
390 return ncThis->fLocale != nullptr ? *ncThis->fLocale : Locale::getDefault();
391}
392
393const Locale ResourceBundle::getLocale(ULocDataLocaleType type, UErrorCode &status) const
394{
395 return ures_getLocaleByType(fResource, type, &status);
396}
397
398U_NAMESPACE_END
399//eof
400