1 /*
2 * Copyright (c) 1998, 2015, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation. Oracle designates this
8 * particular file as subject to the "Classpath" exception as provided
9 * by Oracle in the LICENSE file that accompanied this code.
10 *
11 * This code is distributed in the hope that it will be useful, but WITHOUT
12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
14 * version 2 for more details (a copy is included in the LICENSE file that
15 * accompanied this code).
16 *
17 * You should have received a copy of the GNU General Public License version
18 * 2 along with this work; if not, write to the Free Software Foundation,
19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
20 *
21 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
22 * or visit www.oracle.com if you need additional information or have any
23 * questions.
24 */
25
26 package com.sun.xml.internal.dtdparser;
27
28 import java.io.InputStream;
29 import java.text.FieldPosition;
30 import java.text.MessageFormat;
31 import java.util.Hashtable;
32 import java.util.Locale;
33 import java.util.MissingResourceException;
34 import java.util.ResourceBundle;
35
36
37 /**
38 * This class provides support for multi-language string lookup, as needed
39 * to localize messages from applications supporting multiple languages
40 * at the same time. One class of such applications is network services,
41 * such as HTTP servers, which talk to clients who may not be from the
42 * same locale as the server. This class supports a form of negotiation
43 * for the language used in presenting a message from some package, where
44 * both user (client) preferences and application (server) support are
45 * accounted for when choosing locales and formatting messages.
46 *
47 * <p>Each package should have a singleton package-private message catalog
48 * class. This ensures that the correct class loader will always be used to
49 * access message resources, and minimizes use of memory:</p><pre>
50 * package <em>some.package</em>;
51 *
52 * // "foo" might be public
53 * class foo {
54 * ...
55 * // package private
56 * static final Catalog messages = new Catalog ();
57 * static final class Catalog extends MessageCatalog {
58 * Catalog () { super (Catalog.class); }
59 * }
60 * ...
61 * }
62 * </pre>
63 *
64 * <p> Messages for a known client could be generated using code
65 * something like this:</p><pre>
66 * String clientLanguages [];
67 * Locale clientLocale;
68 * String clientMessage;
69 *
70 * // client languages will probably be provided by client,
71 * // e.g. by an HTTP/1.1 "Accept-Language" header.
72 * clientLanguages = new String [] { "en-ca", "fr-ca", "ja", "zh" };
73 * clientLocale = foo.messages.chooseLocale (clientLanguages);
74 * clientMessage = foo.messages.getMessage (clientLocale,
75 * "fileCount",
76 * new Object [] { new Integer (numberOfFiles) }
77 * );
78 * </pre>
79 *
80 * <p> At this time, this class does not include functionality permitting
81 * messages to be passed around and localized after-the-fact. The consequence
82 * of this is that the locale for messages must be passed down through layers
83 * which have no normal reason to support such passdown, or else the system
84 * default locale must be used instead of the one the client needs.</p>
85 *
86 * <p> The following guidelines should be used when constructiong
87 * multi-language applications:</p>
88 * <ol>
89 * <li> Always use <a href=#chooseLocale>chooseLocale</a> to select the
90 * locale you pass to your <code>getMessage</code> call. This lets your
91 * applications use IETF standard locale names, and avoids needless
92 * use of system defaults.</li>
93 *
94 * <li> The localized messages for a given package should always go in
95 * a separate <em>resources</em> sub-package. There are security
96 * implications; see below.</li>
97 *
98 * <li> Make sure that a language name is included in each bundle name,
99 * so that the developer's locale will not be inadvertently used. That
100 * is, don't create defaults like <em>resources/Messages.properties</em>
101 * or <em>resources/Messages.class</em>, since ResourceBundle will choose
102 * such defaults rather than giving software a chance to choose a more
103 * appropriate language for its messages. Your message bundles should
104 * have names like <em>Messages_en.properties</em> (for the "en", or
105 * English, language) or <em>Messages_ja.class</em> ("ja" indicates the
106 * Japanese language).</li>
107 *
108 * <li> Only use property files for messages in languages which can
109 * be limited to the ISO Latin/1 (8859-1) characters supported by the
110 * property file format. (This is mostly Western European languages.)
111 * Otherwise, subclass ResourceBundle to provide your messages; it is
112 * simplest to subclass <code>java.util.ListResourceBundle</code>.</li>
113 *
114 * <li> Never use another package's message catalog or resource bundles.
115 * It should not be possible for a change internal to one package (such
116 * as eliminating or improving messages) to break another package.</li>
117 *</ol>
118 *
119 * <p>The "resources" sub-package can be treated separately from the
120 * package with which it is associated. That main package may be sealed
121 * and possibly signed, preventing other software from adding classes to
122 * the package which would be able to access methods and data which are
123 * not designed to be publicly accessible. On the other hand, resources
124 * such as localized messages are often provided after initial product
125 * shipment, without a full release cycle for the product. Such files
126 * (text and class files) need to be added to some package. Since they
127 * should not be added to the main package, the "resources" subpackage is
128 * used without risking the security or integrity of that main package
129 * as distributed in its JAR file.</p>
130 *
131 * @author David Brownell
132 * @version 1.1, 00/08/05
133 * @see java.util.Locale
134 * @see java.util.ListResourceBundle
135 * @see java.text.MessageFormat
136 */
137 // leave this as "abstract" -- each package needs its own subclass,
138 // else it's not always going to be using the right class loader.
139 abstract public class MessageCatalog {
140 private String bundleName;
141
142 /**
143 * Create a message catalog for use by classes in the same package
144 * as the specified class. This uses <em>Messages</em> resource
145 * bundles in the <em>resources</em> sub-package of class passed as
146 * a parameter.
147 *
148 * @param packageMember Class whose package has localized messages
149 */
150 protected MessageCatalog(Class packageMember) {
151 this(packageMember, "Messages");
152 }
153
154 /**
155 * Create a message catalog for use by classes in the same package
156 * as the specified class. This uses the specified resource
157 * bundle name in the <em>resources</em> sub-package of class passed
158 * as a parameter; for example, <em>resources.Messages</em>.
159 *
160 * @param packageMember Class whose package has localized messages
161 * @param bundle Name of a group of resource bundles
162 */
163 private MessageCatalog(Class packageMember, String bundle) {
164 int index;
165
166 bundleName = packageMember.getName();
167 index = bundleName.lastIndexOf('.');
168 if (index == -1) // "ClassName"
169 bundleName = "";
170 else // "some.package.ClassName"
171 bundleName = bundleName.substring(0, index) + ".";
172 bundleName = bundleName + "resources." + bundle;
173 }
174
175
176 /**
177 * Get a message localized to the specified locale, using the message ID
178 * and package name if no message is available. The locale is normally
179 * that of the client of a service, chosen with knowledge that both the
180 * client and this server support that locale. There are two error
181 * cases: first, when the specified locale is unsupported or null, the
182 * default locale is used if possible; second, when no bundle supports
183 * that locale, the message ID and package name are used.
184 *
185 * @param locale The locale of the message to use. If this is null,
186 * the default locale will be used.
187 * @param messageId The ID of the message to use.
188 * @return The message, localized as described above.
189 */
190 public String getMessage(Locale locale,
191 String messageId) {
192 ResourceBundle bundle;
193
194 // cope with unsupported locale...
195 if (locale == null)
196 locale = Locale.getDefault();
197
198 try {
199 bundle = ResourceBundle.getBundle(bundleName, locale);
200 } catch (MissingResourceException e) {
201 bundle = ResourceBundle.getBundle(bundleName, Locale.ENGLISH);
202 }
203 return bundle.getString(messageId);
204 }
205
206
207 /**
208 * Format a message localized to the specified locale, using the message
209 * ID with its package name if none is available. The locale is normally
210 * the client of a service, chosen with knowledge that both the client
211 * server support that locale. There are two error cases: first, if the
212 * specified locale is unsupported or null, the default locale is used if
213 * possible; second, when no bundle supports that locale, the message ID
214 * and package name are used.
215 *
216 * @param locale The locale of the message to use. If this is null,
217 * the default locale will be used.
218 * @param messageId The ID of the message format to use.
219 * @param parameters Used when formatting the message. Objects in
220 * this list are turned to strings if they are not Strings, Numbers,
221 * or Dates (that is, if MessageFormat would treat them as errors).
222 * @return The message, localized as described above.
223 * @see java.text.MessageFormat
224 */
225 public String getMessage(Locale locale,
226 String messageId,
227 Object parameters []) {
228 if (parameters == null)
229 return getMessage(locale, messageId);
230
231 // since most messages won't be tested (sigh), be friendly to
232 // the inevitable developer errors of passing random data types
233 // to the message formatting code.
234 for (int i = 0; i < parameters.length; i++) {
235 if (!(parameters[i] instanceof String)
236 && !(parameters[i] instanceof Number)
237 && !(parameters[i] instanceof java.util.Date)) {
238 if (parameters[i] == null)
239 parameters[i] = "(null)";
240 else
241 parameters[i] = parameters[i].toString();
242 }
243 }
244
245 // similarly, cope with unsupported locale...
246 if (locale == null)
247 locale = Locale.getDefault();
248
249 // get the appropriately localized MessageFormat object
250 ResourceBundle bundle;
251 MessageFormat format;
252
253 try {
254 bundle = ResourceBundle.getBundle(bundleName, locale);
255 } catch (MissingResourceException e) {
256 bundle = ResourceBundle.getBundle(bundleName, Locale.ENGLISH);
257 /*String retval;
258
259 retval = packagePrefix (messageId);
260 for (int i = 0; i < parameters.length; i++) {
261 retval += ' ';
262 retval += parameters [i];
263 }
264 return retval;*/
265 }
266 format = new MessageFormat(bundle.getString(messageId));
267 format.setLocale(locale);
268
269 // return the formatted message
270 StringBuffer result = new StringBuffer();
271
272 result = format.format(parameters, result, new FieldPosition(0));
273 return result.toString();
274 }
275
276
277 /**
278 * Chooses a client locale to use, using the first language specified in
279 * the list that is supported by this catalog. If none of the specified
280 * languages is supported, a null value is returned. Such a list of
281 * languages might be provided in an HTTP/1.1 "Accept-Language" header
282 * field, or through some other content negotiation mechanism.
283 *
284 * <p> The language specifiers recognized are RFC 1766 style ("fr" for
285 * all French, "fr-ca" for Canadian French), although only the strict
286 * ISO subset (two letter language and country specifiers) is currently
287 * supported. Java-style locale strings ("fr_CA") are also supported.</p>
288 *
289 * @param languages Array of language specifiers, ordered with the most
290 * preferable one at the front. For example, "en-ca" then "fr-ca",
291 * followed by "zh_CN".
292 * @return The most preferable supported locale, or null.
293 * @see java.util.Locale
294 */
295 public Locale chooseLocale(String languages []) {
296 if ((languages = canonicalize(languages)) != null) {
297 for (int i = 0; i < languages.length; i++)
298 if (isLocaleSupported(languages[i]))
299 return getLocale(languages[i]);
300 }
301 return null;
302 }
303
304
305 //
306 // Canonicalizes the RFC 1766 style language strings ("en-in") to
307 // match standard Java usage ("en_IN"), removing strings that don't
308 // use two character ISO language and country codes. Avoids all
309 // memory allocations possible, so that if the strings passed in are
310 // just lowercase ISO codes (a common case) the input is returned.
311 //
312 private String[] canonicalize(String languages []) {
313 boolean didClone = false;
314 int trimCount = 0;
315
316 if (languages == null)
317 return languages;
318
319 for (int i = 0; i < languages.length; i++) {
320 String lang = languages[i];
321 int len = lang.length();
322
323 // no RFC1766 extensions allowed; "zh" and "zh-tw" (etc) are OK
324 // as are regular locale names with no variant ("de_CH").
325 if (!(len == 2 || len == 5)) {
326 if (!didClone) {
327 languages = (String[]) languages.clone();
328 didClone = true;
329 }
330 languages[i] = null;
331 trimCount++;
332 continue;
333 }
334
335 // language code ... if already lowercase, we change nothing
336 if (len == 2) {
337 lang = lang.toLowerCase(Locale.ENGLISH);
338 if (lang != languages[i]) {
339 if (!didClone) {
340 languages = (String[]) languages.clone();
341 didClone = true;
342 }
343 languages[i] = lang;
344 }
345 continue;
346 }
347
348 // language_country ... fixup case, force "_"
349 char buf [] = new char[5];
350
351 buf[0] = Character.toLowerCase(lang.charAt(0));
352 buf[1] = Character.toLowerCase(lang.charAt(1));
353 buf[2] = '_';
354 buf[3] = Character.toUpperCase(lang.charAt(3));
355 buf[4] = Character.toUpperCase(lang.charAt(4));
356 if (!didClone) {
357 languages = (String[]) languages.clone();
358 didClone = true;
359 }
360 languages[i] = new String(buf);
361 }
362
363 // purge any shadows of deleted RFC1766 extended language codes
364 if (trimCount != 0) {
365 String temp [] = new String[languages.length - trimCount];
366 int i;
367
368 for (i = 0, trimCount = 0; i < temp.length; i++) {
369 while (languages[i + trimCount] == null)
370 trimCount++;
371 temp[i] = languages[i + trimCount];
372 }
373 languages = temp;
374 }
375 return languages;
376 }
377
378
379 //
380 // Returns a locale object supporting the specified locale, using
381 // a small cache to speed up some common languages and reduce the
382 // needless allocation of memory.
383 //
384 private Locale getLocale(String localeName) {
385 String language, country;
386 int index;
387
388 index = localeName.indexOf('_');
389 if (index == -1) {
390 //
391 // Special case the builtin JDK languages
392 //
393 if (localeName.equals("de"))
394 return Locale.GERMAN;
395 if (localeName.equals("en"))
396 return Locale.ENGLISH;
397 if (localeName.equals("fr"))
398 return Locale.FRENCH;
399 if (localeName.equals("it"))
400 return Locale.ITALIAN;
401 if (localeName.equals("ja"))
402 return Locale.JAPANESE;
403 if (localeName.equals("ko"))
404 return Locale.KOREAN;
405 if (localeName.equals("zh"))
406 return Locale.CHINESE;
407
408 language = localeName;
409 country = "";
410 } else {
411 if (localeName.equals("zh_CN"))
412 return Locale.SIMPLIFIED_CHINESE;
413 if (localeName.equals("zh_TW"))
414 return Locale.TRADITIONAL_CHINESE;
415
416 //
417 // JDK also has constants for countries: en_GB, en_US, en_CA,
418 // fr_FR, fr_CA, de_DE, ja_JP, ko_KR. We don't use those.
419 //
420 language = localeName.substring(0, index);
421 country = localeName.substring(index + 1);
422 }
423
424 return new Locale(language, country);
425 }
426
427
428 //
429 // cache for isLanguageSupported(), below ... key is a language
430 // or locale name, value is a Boolean
431 //
432 private Hashtable cache = new Hashtable(5);
433
434
435 /**
436 * Returns true iff the specified locale has explicit language support.
437 * For example, the traditional Chinese locale "zh_TW" has such support
438 * if there are message bundles suffixed with either "zh_TW" or "zh".
439 *
440 * <P> This method is used to bypass part of the search path mechanism
441 * of the <code>ResourceBundle</code> class, specifically the parts which
442 * force use of default locales and bundles. Such bypassing is required
443 * in order to enable use of a client's preferred languages. Following
444 * the above example, if a client prefers "zh_TW" but can also accept
445 * "ja", this method would be used to detect that there are no "zh_TW"
446 * resource bundles and hence that "ja" messages should be used. This
447 * bypasses the ResourceBundle mechanism which will return messages in
448 * some other locale (picking some hard-to-anticipate default) instead
449 * of reporting an error and letting the client choose another locale.
450 *
451 * @param localeName A standard Java locale name, using two character
452 * language codes optionally suffixed by country codes.
453 * @return True iff the language of that locale is supported.
454 * @see java.util.Locale
455 */
456 public boolean isLocaleSupported(String localeName) {
457 //
458 // Use previous results if possible. We expect that the codebase
459 // is immutable, so we never worry about changing the cache.
460 //
461 Boolean value = (Boolean) cache.get(localeName);
462
463 if (value != null)
464 return value.booleanValue();
465
466 //
467 // Try "language_country_variant", then "language_country",
468 // then finally "language" ... assuming the longest locale name
469 // is passed. If not, we'll try fewer options.
470 //
471 ClassLoader loader = null;
472
473 for (; ;) {
474 String name = bundleName + "_" + localeName;
475
476 // look up classes ...
477 try {
478 Class.forName(name);
479 cache.put(localeName, Boolean.TRUE);
480 return true;
481 } catch (Exception e) {
482 }
483
484 // ... then property files (only for ISO Latin/1 messages)
485 InputStream in;
486
487 if (loader == null)
488 loader = getClass().getClassLoader();
489
490 name = name.replace('.', '/');
491 name = name + ".properties";
492 if (loader == null)
493 in = ClassLoader.getSystemResourceAsStream(name);
494 else
495 in = loader.getResourceAsStream(name);
496 if (in != null) {
497 cache.put(localeName, Boolean.TRUE);
498 return true;
499 }
500
501 int index = localeName.indexOf('_');
502
503 if (index > 0)
504 localeName = localeName.substring(0, index);
505 else
506 break;
507 }
508
509 //
510 // If we got this far, we failed. Remember for later.
511 //
512 cache.put(localeName, Boolean.FALSE);
513 return false;
514 }
515 }