1 /*
2 **********************************************************************
3 * Copyright (C) 1997-2010, International Business Machines
4 * Corporation and others. All Rights Reserved.
5 **********************************************************************
6 *
7 * File resbund.cpp
8 *
9 * Modification History:
10 *
11 * Date Name Description
12 * 02/05/97 aliu Fixed bug in chopLocale. Added scanForLocaleInFile
13 * based on code taken from scanForLocale. Added
14 * constructor which attempts to read resource bundle
15 * from a specific file, without searching other files.
16 * 02/11/97 aliu Added UErrorCode return values to constructors. Fixed
17 * infinite loops in scanForFile and scanForLocale.
18 * Modified getRawResourceData to not delete storage in
19 * localeData and resourceData which it doesn't own.
20 * Added Mac compatibility #ifdefs for tellp() and
21 * ios::nocreate.
22 * 03/04/97 aliu Modified to use ExpandingDataSink objects instead of
23 * the highly inefficient ostrstream objects.
24 * 03/13/97 aliu Rewrote to load in entire resource bundle and store
25 * it as a Hashtable of ResourceBundleData objects.
26 * Added state table to govern parsing of files.
27 * Modified to load locale index out of new file distinct
28 * from default.txt.
29 * 03/25/97 aliu Modified to support 2-d arrays, needed for timezone data.
30 * Added support for custom file suffixes. Again, needed
31 * to support timezone data. Improved error handling to
32 * detect duplicate tags and subtags.
33 * 04/07/97 aliu Fixed bug in getHashtableForLocale(). Fixed handling
34 * of failing UErrorCode values on entry to API methods.
35 * Fixed bugs in getArrayItem() for negative indices.
36 * 04/29/97 aliu Update to use new Hashtable deletion protocol.
37 * 05/06/97 aliu Flattened kTransitionTable for HP compiler.
38 * Fixed usage of CharString.
39 * 06/11/99 stephen Removed parsing of .txt files.
40 * Reworked to use new binary format.
41 * Cleaned up.
42 * 06/14/99 stephen Removed methods taking a filename suffix.
43 * 06/22/99 stephen Added missing T_FileStream_close in parse()
44 * 11/09/99 weiv Added getLocale(), rewritten constructForLocale()
45 * March 2000 weiv complete overhaul.
46 ******************************************************************************
47 */
48
49 #include "unicode/utypes.h"
50 #include "unicode/resbund.h"
51 #include "umutex.h"
52
53 #include "uresimp.h"
54
55 U_NAMESPACE_BEGIN
56
57 /*-----------------------------------------------------------------------------
58 * Implementation Notes
59 *
60 * Resource bundles are read in once, and thereafter cached.
61 * ResourceBundle statically keeps track of which files have been
62 * read, so we are guaranteed that each file is read at most once.
63 * Resource bundles can be loaded from different data directories and
64 * will be treated as distinct, even if they are for the same locale.
65 *
66 * Resource bundles are lightweight objects, which have pointers to
67 * one or more shared Hashtable objects containing all the data.
68 * Copying would be cheap, but there is no copy constructor, since
69 * there wasn't one in the original API.
70 *
71 * The ResourceBundle parsing mechanism is implemented as a transition
72 * network, for easy maintenance and modification. The network is
73 * implemented as a matrix (instead of in code) to make this even
74 * easier. The matrix contains Transition objects. Each Transition
75 * object describes a destination node and an action to take before
76 * moving to the destination node. The source node is encoded by the
77 * index of the object in the array that contains it. The pieces
78 * needed to understand the transition network are the enums for node
79 * IDs and actions, the parse() method, which walks through the
80 * network and implements the actions, and the network itself. The
81 * network guarantees certain conditions, for example, that a new
82 * resource will not be closed until one has been opened first; or
83 * that data will not be stored into a TaggedList until a TaggedList
84 * has been created. Nonetheless, the code in parse() does some
85 * consistency checks as it runs the network, and fails with an
86 * U_INTERNAL_PROGRAM_ERROR if one of these checks fails. If the input
87 * data has a bad format, an U_INVALID_FORMAT_ERROR is returned. If you
88 * see an U_INTERNAL_PROGRAM_ERROR the transition matrix has a bug in
89 * it.
90 *
91 * Old functionality of multiple locales in a single file is still
92 * supported. For this reason, LOCALE names override FILE names. If
93 * data for en_US is located in the en.txt file, once it is loaded,
94 * the code will not care where it came from (other than remembering
95 * which directory it came from). However, if there is an en_US
96 * resource in en_US.txt, that will take precedence. There is no
97 * limit to the number or type of resources that can be stored in a
98 * file, however, files are only searched in a specific way. If
99 * en_US_CA is requested, then first en_US_CA.txt is searched, then
100 * en_US.txt, then en.txt, then default.txt. So it only makes sense
101 * to put certain locales in certain files. In this example, it would
102 * be logical to put en_US_CA, en_US, and en into the en.txt file,
103 * since they would be found there if asked for. The extreme example
104 * is to place all locale resources into default.txt, which should
105 * also work.
106 *
107 * Inheritance is implemented. For example, xx_YY_zz inherits as
108 * follows: xx_YY_zz, xx_YY, xx, default. Inheritance is implemented
109 * as an array of hashtables. There will be from 1 to 4 hashtables in
110 * the array.
111 *
112 * Fallback files are implemented. The fallback pattern is Language
113 * Country Variant (LCV) -> LC -> L. Fallback is first done for the
114 * requested locale. Then it is done for the default locale, as
115 * returned by Locale::getDefault(). Then the special file
116 * default.txt is searched for the default locale. The overall FILE
117 * fallback path is LCV -> LC -> L -> dLCV -> dLC -> dL -> default.
118 *
119 * Note that although file name searching includes the default locale,
120 * once a ResourceBundle object is constructed, the inheritance path
121 * no longer includes the default locale. The path is LCV -> LC -> L
122 * -> default.
123 *
124 * File parsing is lazy. Nothing is parsed unless it is called for by
125 * someone. So when a ResourceBundle for xx_YY_zz is constructed,
126 * only that locale is parsed (along with anything else in the same
127 * file). Later, if the FooBar tag is asked for, and if it isn't
128 * found in xx_YY_zz, then xx_YY.txt will be parsed and checked, and
129 * so forth, until the chain is exhausted or the tag is found.
130 *
131 * Thread-safety is implemented around caches, both the cache that
132 * stores all the resouce data, and the cache that stores flags
133 * indicating whether or not a file has been visited. These caches
134 * delete their storage at static cleanup time, when the process
135 * quits.
136 *
137 * ResourceBundle supports TableCollation as a special case. This
138 * involves having special ResourceBundle objects which DO own their
139 * data, since we don't want large collation rule strings in the
140 * ResourceBundle cache (these are already cached in the
141 * TableCollation cache). TableCollation files (.ctx files) have the
142 * same format as normal resource data files, with a different
143 * interpretation, from the standpoint of ResourceBundle. .ctx files
144 * are loaded into otherwise ordinary ResourceBundle objects. They
145 * don't inherit (that's implemented by TableCollation) and they own
146 * their data (as mentioned above). However, they still support
147 * possible multiple locales in a single .ctx file. (This is in
148 * practice a bad idea, since you only want the one locale you're
149 * looking for, and only one tag will be present
150 * ("CollationElements"), so you don't need an inheritance chain of
151 * multiple locales.) Up to 4 locale resources will be loaded from a
152 * .ctx file; everything after the first 4 is ignored (parsed and
153 * deleted). (Normal .txt files have no limit.) Instead of being
154 * loaded into the cache, and then looked up as needed, the locale
155 * resources are read straight into the ResourceBundle object.
156 *
157 * The Index, which used to reside in default.txt, has been moved to a
158 * new file, index.txt. This file contains a slightly modified format
159 * with the addition of the "InstalledLocales" tag; it looks like:
160 *
161 * Index {
162 * InstalledLocales {
163 * ar
164 * ..
165 * zh_TW
166 * }
167 * }
168 */
169 //-----------------------------------------------------------------------------
170
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ResourceBundle)171 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ResourceBundle)
172
173 ResourceBundle::ResourceBundle(UErrorCode &err)
174 :UObject(), fLocale(NULL)
175 {
176 fResource = ures_open(0, Locale::getDefault().getName(), &err);
177 }
178
ResourceBundle(const ResourceBundle & other)179 ResourceBundle::ResourceBundle(const ResourceBundle &other)
180 :UObject(other), fLocale(NULL)
181 {
182 UErrorCode status = U_ZERO_ERROR;
183
184 if (other.fResource) {
185 fResource = ures_copyResb(0, other.fResource, &status);
186 } else {
187 /* Copying a bad resource bundle */
188 fResource = NULL;
189 }
190 }
191
ResourceBundle(UResourceBundle * res,UErrorCode & err)192 ResourceBundle::ResourceBundle(UResourceBundle *res, UErrorCode& err)
193 :UObject(), fLocale(NULL)
194 {
195 if (res) {
196 fResource = ures_copyResb(0, res, &err);
197 } else {
198 /* Copying a bad resource bundle */
199 fResource = NULL;
200 }
201 }
202
ResourceBundle(const char * path,const Locale & locale,UErrorCode & err)203 ResourceBundle::ResourceBundle(const char* path, const Locale& locale, UErrorCode& err)
204 :UObject(), fLocale(NULL)
205 {
206 fResource = ures_open(path, locale.getName(), &err);
207 }
208
209
operator =(const ResourceBundle & other)210 ResourceBundle& ResourceBundle::operator=(const ResourceBundle& other)
211 {
212 if(this == &other) {
213 return *this;
214 }
215 if(fResource != 0) {
216 ures_close(fResource);
217 fResource = NULL;
218 }
219 UErrorCode status = U_ZERO_ERROR;
220 if (other.fResource) {
221 fResource = ures_copyResb(0, other.fResource, &status);
222 } else {
223 /* Copying a bad resource bundle */
224 fResource = NULL;
225 }
226 return *this;
227 }
228
~ResourceBundle()229 ResourceBundle::~ResourceBundle()
230 {
231 if(fResource != 0) {
232 ures_close(fResource);
233 }
234 if(fLocale != NULL) {
235 delete(fLocale);
236 }
237 }
238
239 ResourceBundle *
clone() const240 ResourceBundle::clone() const {
241 return new ResourceBundle(*this);
242 }
243
getString(UErrorCode & status) const244 UnicodeString ResourceBundle::getString(UErrorCode& status) const {
245 int32_t len = 0;
246 const UChar *r = ures_getString(fResource, &len, &status);
247 return UnicodeString(TRUE, r, len);
248 }
249
getBinary(int32_t & len,UErrorCode & status) const250 const uint8_t *ResourceBundle::getBinary(int32_t& len, UErrorCode& status) const {
251 return ures_getBinary(fResource, &len, &status);
252 }
253
getIntVector(int32_t & len,UErrorCode & status) const254 const int32_t *ResourceBundle::getIntVector(int32_t& len, UErrorCode& status) const {
255 return ures_getIntVector(fResource, &len, &status);
256 }
257
getUInt(UErrorCode & status) const258 uint32_t ResourceBundle::getUInt(UErrorCode& status) const {
259 return ures_getUInt(fResource, &status);
260 }
261
getInt(UErrorCode & status) const262 int32_t ResourceBundle::getInt(UErrorCode& status) const {
263 return ures_getInt(fResource, &status);
264 }
265
getName(void) const266 const char *ResourceBundle::getName(void) const {
267 return ures_getName(fResource);
268 }
269
getKey(void) const270 const char *ResourceBundle::getKey(void) const {
271 return ures_getKey(fResource);
272 }
273
getType(void) const274 UResType ResourceBundle::getType(void) const {
275 return ures_getType(fResource);
276 }
277
getSize(void) const278 int32_t ResourceBundle::getSize(void) const {
279 return ures_getSize(fResource);
280 }
281
hasNext(void) const282 UBool ResourceBundle::hasNext(void) const {
283 return ures_hasNext(fResource);
284 }
285
resetIterator(void)286 void ResourceBundle::resetIterator(void) {
287 ures_resetIterator(fResource);
288 }
289
getNext(UErrorCode & status)290 ResourceBundle ResourceBundle::getNext(UErrorCode& status) {
291 UResourceBundle r;
292
293 ures_initStackObject(&r);
294 ures_getNextResource(fResource, &r, &status);
295 ResourceBundle res(&r, status);
296 if (U_SUCCESS(status)) {
297 ures_close(&r);
298 }
299 return res;
300 }
301
getNextString(UErrorCode & status)302 UnicodeString ResourceBundle::getNextString(UErrorCode& status) {
303 int32_t len = 0;
304 const UChar* r = ures_getNextString(fResource, &len, 0, &status);
305 return UnicodeString(TRUE, r, len);
306 }
307
getNextString(const char ** key,UErrorCode & status)308 UnicodeString ResourceBundle::getNextString(const char ** key, UErrorCode& status) {
309 int32_t len = 0;
310 const UChar* r = ures_getNextString(fResource, &len, key, &status);
311 return UnicodeString(TRUE, r, len);
312 }
313
get(int32_t indexR,UErrorCode & status) const314 ResourceBundle ResourceBundle::get(int32_t indexR, UErrorCode& status) const {
315 UResourceBundle r;
316
317 ures_initStackObject(&r);
318 ures_getByIndex(fResource, indexR, &r, &status);
319 ResourceBundle res(&r, status);
320 if (U_SUCCESS(status)) {
321 ures_close(&r);
322 }
323 return res;
324 }
325
getStringEx(int32_t indexS,UErrorCode & status) const326 UnicodeString ResourceBundle::getStringEx(int32_t indexS, UErrorCode& status) const {
327 int32_t len = 0;
328 const UChar* r = ures_getStringByIndex(fResource, indexS, &len, &status);
329 return UnicodeString(TRUE, r, len);
330 }
331
get(const char * key,UErrorCode & status) const332 ResourceBundle ResourceBundle::get(const char* key, UErrorCode& status) const {
333 UResourceBundle r;
334
335 ures_initStackObject(&r);
336 ures_getByKey(fResource, key, &r, &status);
337 ResourceBundle res(&r, status);
338 if (U_SUCCESS(status)) {
339 ures_close(&r);
340 }
341 return res;
342 }
343
getWithFallback(const char * key,UErrorCode & status)344 ResourceBundle ResourceBundle::getWithFallback(const char* key, UErrorCode& status){
345 UResourceBundle r;
346 ures_initStackObject(&r);
347 ures_getByKeyWithFallback(fResource, key, &r, &status);
348 ResourceBundle res(&r, status);
349 if(U_SUCCESS(status)){
350 ures_close(&r);
351 }
352 return res;
353 }
getStringEx(const char * key,UErrorCode & status) const354 UnicodeString ResourceBundle::getStringEx(const char* key, UErrorCode& status) const {
355 int32_t len = 0;
356 const UChar* r = ures_getStringByKey(fResource, key, &len, &status);
357 return UnicodeString(TRUE, r, len);
358 }
359
360 const char*
getVersionNumber() const361 ResourceBundle::getVersionNumber() const
362 {
363 return ures_getVersionNumberInternal(fResource);
364 }
365
getVersion(UVersionInfo versionInfo) const366 void ResourceBundle::getVersion(UVersionInfo versionInfo) const {
367 ures_getVersion(fResource, versionInfo);
368 }
369
getLocale(void) const370 const Locale &ResourceBundle::getLocale(void) const
371 {
372 UBool needInit;
373 UMTX_CHECK(NULL, (fLocale == NULL), needInit);
374 if(needInit) {
375 UErrorCode status = U_ZERO_ERROR;
376 const char *localeName = ures_getLocaleInternal(fResource, &status);
377 Locale *tLocale = new Locale(localeName);
378 // Null pointer check
379 if (tLocale == NULL) {
380 return Locale::getDefault(); // Return default locale if one could not be created.
381 }
382 umtx_lock(NULL);
383 ResourceBundle *me = (ResourceBundle *)this; // semantically const
384 if (me->fLocale == NULL) {
385 me->fLocale = tLocale;
386 tLocale = NULL;
387 }
388 umtx_unlock(NULL);
389 delete tLocale;
390 }
391 return *fLocale;
392 }
393
getLocale(ULocDataLocaleType type,UErrorCode & status) const394 const Locale ResourceBundle::getLocale(ULocDataLocaleType type, UErrorCode &status) const
395 {
396 return ures_getLocaleByType(fResource, type, &status);
397 }
398
399 //eof
400 U_NAMESPACE_END
401