Source Code Cross Referenced for RuleBasedCollator.java in  » 6.0-JDK-Modules » j2me » java » text » Java Source Code / Java DocumentationJava Source Code and Java Documentation

001:        /*
002:         *
003:         * @(#)RuleBasedCollator.java	1.36 06/10/03
004:         *
005:         * Portions Copyright  2000-2006 Sun Microsystems, Inc. All Rights
006:         * Reserved.  Use is subject to license terms.
007:         * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER
008:         * 
009:         * This program is free software; you can redistribute it and/or
010:         * modify it under the terms of the GNU General Public License version
011:         * 2 only, as published by the Free Software Foundation.
012:         * 
013:         * This program is distributed in the hope that it will be useful, but
014:         * WITHOUT ANY WARRANTY; without even the implied warranty of
015:         * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
016:         * General Public License version 2 for more details (a copy is
017:         * included at /legal/license.txt).
018:         * 
019:         * You should have received a copy of the GNU General Public License
020:         * version 2 along with this work; if not, write to the Free Software
021:         * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
022:         * 02110-1301 USA
023:         * 
024:         * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa
025:         * Clara, CA 95054 or visit www.sun.com if you need additional
026:         * information or have any questions.
027:         */
028:
029:        /*
030:         * (C) Copyright Taligent, Inc. 1996, 1997 - All Rights Reserved
031:         * (C) Copyright IBM Corp. 1996-1998 - All Rights Reserved
032:         *
033:         *   The original version of this source code and documentation is copyrighted
034:         * and owned by Taligent, Inc., a wholly-owned subsidiary of IBM. These
035:         * materials are provided under terms of a License Agreement between Taligent
036:         * and Sun. This technology is protected by multiple US and International
037:         * patents. This notice and attribution to Taligent may not be removed.
038:         *   Taligent is a registered trademark of Taligent, Inc.
039:         *
040:         */
041:
042:        package java.text;
043:
044:        import java.util.Vector;
045:        import java.util.Locale;
046:        import sun.text.Normalizer;
047:        import sun.text.NormalizerUtilities;
048:
049:        /**
050:         * The <code>RuleBasedCollator</code> class is a concrete subclass of
051:         * <code>Collator</code> that provides a simple, data-driven, table
052:         * collator.  With this class you can create a customized table-based
053:         * <code>Collator</code>.  <code>RuleBasedCollator</code> maps
054:         * characters to sort keys.
055:         *
056:         * <p>
057:         * <code>RuleBasedCollator</code> has the following restrictions
058:         * for efficiency (other subclasses may be used for more complex languages) :
059:         * <ol>
060:         * <li>If a special collation rule controlled by a &lt;modifier&gt; is
061:         specified it applies to the whole collator object.
062:         * <li>All non-mentioned characters are at the end of the
063:         *     collation order.
064:         * </ol>
065:         *
066:         * <p>
067:         * The collation table is composed of a list of collation rules, where each
068:         * rule is of one of three forms:
069:         * <pre>
070:         *    &lt;modifier&gt;
071:         *    &lt;relation&gt; &lt;text-argument&gt;
072:         *    &lt;reset&gt; &lt;text-argument&gt;
073:         * </pre>
074:         * The definitions of the rule elements is as follows:
075:         * <UL Type=disc>
076:         *    <LI><strong>Text-Argument</strong>: A text-argument is any sequence of
077:         *        characters, excluding special characters (that is, common
078:         *        whitespace characters [0009-000D, 0020] and rule syntax characters
079:         *        [0021-002F, 003A-0040, 005B-0060, 007B-007E]). If those
080:         *        characters are desired, you can put them in single quotes
081:         *        (e.g. ampersand => '&'). Note that unquoted white space characters
082:         *        are ignored; e.g. <code>b c</code> is treated as <code>bc</code>.
083:         *    <LI><strong>Modifier</strong>: There are currently two modifiers that 
084:         *        turn on special collation rules.
085:         *        <UL Type=square>
086:         *            <LI>'@' : Turns on backwards sorting of accents (secondary
087:         *                      differences), as in French.
088:         *            <LI>'!' : Turns on Thai/Lao vowel-consonant swapping.  If this
089:         *                      rule is in force when a Thai vowel of the range
090:         *                      &#92;U0E40-&#92;U0E44 precedes a Thai consonant of the range
091:         *                      &#92;U0E01-&#92;U0E2E OR a Lao vowel of the range &#92;U0EC0-&#92;U0EC4
092:         *                      precedes a Lao consonant of the range &#92;U0E81-&#92;U0EAE then
093:         *                      the vowel is placed after the consonant for collation
094:         *                      purposes.
095:         *        </UL>
096:         *        <p>'@' : Indicates that accents are sorted backwards, as in French.
097:         *    <LI><strong>Relation</strong>: The relations are the following:
098:         *        <UL Type=square>
099:         *            <LI>'&lt;' : Greater, as a letter difference (primary)
100:         *            <LI>';' : Greater, as an accent difference (secondary)
101:         *            <LI>',' : Greater, as a case difference (tertiary)
102:         *            <LI>'=' : Equal
103:         *        </UL>
104:         *    <LI><strong>Reset</strong>: There is a single reset
105:         *        which is used primarily for contractions and expansions, but which
106:         *        can also be used to add a modification at the end of a set of rules.
107:         *        <p>'&' : Indicates that the next rule follows the position to where
108:         *            the reset text-argument would be sorted.
109:         * </UL>
110:         *
111:         * <p>
112:         * This sounds more complicated than it is in practice. For example, the
113:         * following are equivalent ways of expressing the same thing:
114:         * <blockquote>
115:         * <pre>
116:         * a &lt; b &lt; c
117:         * a &lt; b &amp; b &lt; c
118:         * a &lt; c &amp; a &lt; b
119:         * </pre>
120:         * </blockquote>
121:         * Notice that the order is important, as the subsequent item goes immediately
122:         * after the text-argument. The following are not equivalent:
123:         * <blockquote>
124:         * <pre>
125:         * a &lt; b &amp; a &lt; c
126:         * a &lt; c &amp; a &lt; b
127:         * </pre>
128:         * </blockquote>
129:         * Either the text-argument must already be present in the sequence, or some
130:         * initial substring of the text-argument must be present. (e.g. "a &lt; b &amp; ae &lt; 
131:         * e" is valid since "a" is present in the sequence before "ae" is reset). In
132:         * this latter case, "ae" is not entered and treated as a single character;
133:         * instead, "e" is sorted as if it were expanded to two characters: "a"
134:         * followed by an "e". This difference appears in natural languages: in
135:         * traditional Spanish "ch" is treated as though it contracts to a single
136:         * character (expressed as "c &lt; ch &lt; d"), while in traditional German
137:         * a-umlaut is treated as though it expanded to two characters
138:         * (expressed as "a,A &lt; b,B ... &amp;ae;&#92;u00e3&amp;AE;&#92;u00c3").
139:         * [&#92;u00e3 and &#92;u00c3 are, of course, the escape sequences for a-umlaut.]
140:         * <p>
141:         * <strong>Ignorable Characters</strong>
142:         * <p>
143:         * For ignorable characters, the first rule must start with a relation (the
144:         * examples we have used above are really fragments; "a &lt; b" really should be
145:         * "&lt; a &lt; b"). If, however, the first relation is not "&lt;", then all the all
146:         * text-arguments up to the first "&lt;" are ignorable. For example, ", - &lt; a &lt; b"
147:         * makes "-" an ignorable character, as we saw earlier in the word
148:         * "black-birds". In the samples for different languages, you see that most
149:         * accents are ignorable.
150:         *
151:         * <p><strong>Normalization and Accents</strong>
152:         * <p>
153:         * <code>RuleBasedCollator</code> automatically processes its rule table to
154:         * include both pre-composed and combining-character versions of
155:         * accented characters.  Even if the provided rule string contains only
156:         * base characters and separate combining accent characters, the pre-composed
157:         * accented characters matching all canonical combinations of characters from
158:         * the rule string will be entered in the table.
159:         * <p>
160:         * This allows you to use a RuleBasedCollator to compare accented strings
161:         * even when the collator is set to NO_DECOMPOSITION.  There are two caveats,
162:         * however.  First, if the strings to be collated contain combining
163:         * sequences that may not be in canonical order, you should set the collator to
164:         * CANONICAL_DECOMPOSITION or FULL_DECOMPOSITION to enable sorting of
165:         * combining sequences.  Second, if the strings contain characters with
166:         * compatibility decompositions (such as full-width and half-width forms),
167:         * you must use FULL_DECOMPOSITION, since the rule tables only include
168:         * canonical mappings.
169:         *
170:         * <p><strong>Errors</strong>
171:         * <p>
172:         * The following are errors:
173:         * <UL Type=disc>
174:         *     <LI>A text-argument contains unquoted punctuation symbols
175:         *        (e.g. "a &lt; b-c &lt; d").
176:         *     <LI>A relation or reset character not followed by a text-argument
177:         *        (e.g. "a &lt; ,b").
178:         *     <LI>A reset where the text-argument (or an initial substring of the
179:         *         text-argument) is not already in the sequence.
180:         *         (e.g. "a &lt; b &amp; e &lt; f")
181:         * </UL>
182:         * If you produce one of these errors, a <code>RuleBasedCollator</code> throws
183:         * a <code>ParseException</code>.
184:         * 
185:         * <p><strong>Examples</strong>
186:         * <p>Simple:     "&lt; a &lt; b &lt; c &lt; d"
187:         * <p>Norwegian:  "&lt; a,A&lt; b,B&lt; c,C&lt; d,D&lt; e,E&lt; f,F&lt; g,G&lt; h,H&lt; i,I&lt; j,J
188:         *                 &lt; k,K&lt; l,L&lt; m,M&lt; n,N&lt; o,O&lt; p,P&lt; q,Q&lt; r,R&lt; s,S&lt; t,T
189:         *                 &lt; u,U&lt; v,V&lt; w,W&lt; x,X&lt; y,Y&lt; z,Z
190:         *                 &lt; &#92;u00E5=a&#92;u030A,&#92;u00C5=A&#92;u030A
191:         *                 ;aa,AA&lt; &#92;u00E6,&#92;u00C6&lt; &#92;u00F8,&#92;u00D8"
192:         *
193:         * <p>
194:         * Normally, to create a rule-based Collator object, you will use
195:         * <code>Collator</code>'s factory method <code>getInstance</code>.
196:         * However, to create a rule-based Collator object with specialized
197:         * rules tailored to your needs, you construct the <code>RuleBasedCollator</code>
198:         * with the rules contained in a <code>String</code> object. For example:
199:         * <blockquote>
200:         * <pre>
201:         * String Simple = "&lt; a&lt; b&lt; c&lt; d";
202:         * RuleBasedCollator mySimple = new RuleBasedCollator(Simple);
203:         * </pre>
204:         * </blockquote>
205:         * Or:
206:         * <blockquote>
207:         * <pre>
208:         * String Norwegian = "&lt; a,A&lt; b,B&lt; c,C&lt; d,D&lt; e,E&lt; f,F&lt; g,G&lt; h,H&lt; i,I&lt; j,J" +
209:         *                 "&lt; k,K&lt; l,L&lt; m,M&lt; n,N&lt; o,O&lt; p,P&lt; q,Q&lt; r,R&lt; s,S&lt; t,T" +
210:         *                 "&lt; u,U&lt; v,V&lt; w,W&lt; x,X&lt; y,Y&lt; z,Z" +
211:         *                 "&lt; &#92;u00E5=a&#92;u030A,&#92;u00C5=A&#92;u030A" +
212:         *                 ";aa,AA&lt; &#92;u00E6,&#92;u00C6&lt; &#92;u00F8,&#92;u00D8";
213:         * RuleBasedCollator myNorwegian = new RuleBasedCollator(Norwegian);
214:         * </pre>
215:         * </blockquote>
216:         *
217:         * <p>
218:         * Combining <code>Collator</code>s is as simple as concatenating strings.
219:         * Here's an example that combines two <code>Collator</code>s from two
220:         * different locales:
221:         * <blockquote>
222:         * <pre>
223:         * // Create an en_US Collator object
224:         * RuleBasedCollator en_USCollator = (RuleBasedCollator)
225:         *     Collator.getInstance(new Locale("en", "US", ""));
226:         * // Create a da_DK Collator object
227:         * RuleBasedCollator da_DKCollator = (RuleBasedCollator)
228:         *     Collator.getInstance(new Locale("da", "DK", ""));
229:         * // Combine the two
230:         * // First, get the collation rules from en_USCollator
231:         * String en_USRules = en_USCollator.getRules();
232:         * // Second, get the collation rules from da_DKCollator
233:         * String da_DKRules = da_DKCollator.getRules();
234:         * RuleBasedCollator newCollator =
235:         *     new RuleBasedCollator(en_USRules + da_DKRules);
236:         * // newCollator has the combined rules
237:         * </pre>
238:         * </blockquote>
239:         *
240:         * <p>
241:         * Another more interesting example would be to make changes on an existing
242:         * table to create a new <code>Collator</code> object.  For example, add
243:         * "&amp;C&lt; ch, cH, Ch, CH" to the <code>en_USCollator</code> object to create
244:         * your own:
245:         * <blockquote>
246:         * <pre>
247:         * // Create a new Collator object with additional rules
248:         * String addRules = "&amp;C&lt; ch, cH, Ch, CH";
249:         * RuleBasedCollator myCollator =
250:         *     new RuleBasedCollator(en_USCollator + addRules);
251:         * // myCollator contains the new rules
252:         * </pre>
253:         * </blockquote>
254:         *
255:         * <p>
256:         * The following example demonstrates how to change the order of
257:         * non-spacing accents,
258:         * <blockquote>
259:         * <pre>
260:         * // old rule
261:         * String oldRules = "=&#92;u0301;&#92;u0300;&#92;u0302;&#92;u0308"    // main accents
262:         *                 + ";&#92;u0327;&#92;u0303;&#92;u0304;&#92;u0305"    // main accents
263:         *                 + ";&#92;u0306;&#92;u0307;&#92;u0309;&#92;u030A"    // main accents
264:         *                 + ";&#92;u030B;&#92;u030C;&#92;u030D;&#92;u030E"    // main accents
265:         *                 + ";&#92;u030F;&#92;u0310;&#92;u0311;&#92;u0312"    // main accents
266:         *                 + "&lt; a , A ; ae, AE ; &#92;u00e6 , &#92;u00c6"
267:         *                 + "&lt; b , B &lt; c, C &lt; e, E & C &lt; d, D";
268:         * // change the order of accent characters
269:         * String addOn = "& &#92;u0300 ; &#92;u0308 ; &#92;u0302";
270:         * RuleBasedCollator myCollator = new RuleBasedCollator(oldRules + addOn);
271:         * </pre>
272:         * </blockquote>
273:         *
274:         * <p>
275:         * The last example shows how to put new primary ordering in before the
276:         * default setting. For example, in Japanese <code>Collator</code>, you
277:         * can either sort English characters before or after Japanese characters,
278:         * <blockquote>
279:         * <pre>
280:         * // get en_US Collator rules
281:         * RuleBasedCollator en_USCollator = (RuleBasedCollator)Collator.getInstance(Locale.US);
282:         * // add a few Japanese character to sort before English characters
283:         * // suppose the last character before the first base letter 'a' in
284:         * // the English collation rule is &#92;u2212
285:         * String jaString = "& &#92;u2212 &lt; &#92;u3041, &#92;u3042 &lt; &#92;u3043, &#92;u3044";
286:         * RuleBasedCollator myJapaneseCollator = new
287:         *     RuleBasedCollator(en_USCollator.getRules() + jaString);
288:         * </pre>
289:         * </blockquote>
290:         *
291:         * @see        Collator
292:         * @see        CollationElementIterator
293:         * @version    1.25 07/24/98
294:         * @author     Helena Shih, Laura Werner, Richard Gillam
295:         */
296:        public class RuleBasedCollator extends Collator {
297:            // IMPLEMENTATION NOTES:  The implementation of the collation algorithm is
298:            // divided across three classes: RuleBasedCollator, RBCollationTables, and
299:            // CollationElementIterator.  RuleBasedCollator contains the collator's
300:            // transient state and includes the code that uses the other classes to
301:            // implement comparison and sort-key building.  RuleBasedCollator also
302:            // contains the logic to handle French secondary accent sorting.
303:            // A RuleBasedCollator has two CollationElementIterators.  State doesn't
304:            // need to be preserved in these objects between calls to compare() or
305:            // getCollationKey(), but the objects persist anyway to avoid wasting extra
306:            // creation time.  compare() and getCollationKey() are synchronized to ensure
307:            // thread safety with this scheme.  The CollationElementIterator is responsible
308:            // for generating collation elements from strings and returning one element at
309:            // a time (sometimes there's a one-to-many or many-to-one mapping between
310:            // characters and collation elements-- this class handles that).
311:            // CollationElementIterator depends on RBCollationTables, which contains the
312:            // collator's static state.  RBCollationTables contains the actual data
313:            // tables specifying the collation order of characters for a particular locale
314:            // or use.  It also contains the base logic that CollationElementIterator
315:            // uses to map from characters to collation elements.  A single RBCollationTables
316:            // object is shared among all RuleBasedCollators for the same locale, and
317:            // thus by all the CollationElementIterators they create.
318:
319:            /**
320:             * RuleBasedCollator constructor.  This takes the table rules and builds
321:             * a collation table out of them.  Please see RuleBasedCollator class
322:             * description for more details on the collation rule syntax.
323:             * @see java.util.Locale
324:             * @param rules the collation rules to build the collation table from.
325:             * @exception ParseException A format exception
326:             * will be thrown if the build process of the rules fails. For
327:             * example, build rule "a < ? < d" will cause the constructor to
328:             * throw the ParseException because the '?' is not quoted.
329:             */
330:            public RuleBasedCollator(String rules) throws ParseException {
331:                this (rules, Collator.CANONICAL_DECOMPOSITION);
332:            }
333:
334:            /**
335:             * RuleBasedCollator constructor.  This takes the table rules and builds
336:             * a collation table out of them.  Please see RuleBasedCollator class
337:             * description for more details on the collation rule syntax.
338:             * @see java.util.Locale
339:             * @param rules the collation rules to build the collation table from.
340:             * @param decomp the decomposition strength used to build the
341:             * collation table and to perform comparisons.
342:             * @exception ParseException A format exception
343:             * will be thrown if the build process of the rules fails. For
344:             * example, build rule "a < ? < d" will cause the constructor to
345:             * throw the ParseException because the '?' is not quoted.
346:             */
347:            RuleBasedCollator(String rules, int decomp) throws ParseException {
348:                setStrength(Collator.TERTIARY);
349:                setDecomposition(decomp);
350:                tables = new RBCollationTables(rules, decomp);
351:            }
352:
353:            /**
354:             * "Copy constructor."  Used in clone() for performance.
355:             */
356:            private RuleBasedCollator(RuleBasedCollator that) {
357:                setStrength(that.getStrength());
358:                setDecomposition(that.getDecomposition());
359:                tables = that.tables;
360:            }
361:
362:            /**
363:             * Gets the table-based rules for the collation object.
364:             * @return returns the collation rules that the table collation object
365:             * was created from.
366:             */
367:            public String getRules() {
368:                return tables.getRules();
369:            }
370:
371:            /**
372:             * Return a CollationElementIterator for the given String.
373:             * @see java.text.CollationElementIterator
374:             */
375:            public CollationElementIterator getCollationElementIterator(
376:                    String source) {
377:                return new CollationElementIterator(source, this );
378:            }
379:
380:            /**
381:             * Return a CollationElementIterator for the given String.
382:             * @see java.text.CollationElementIterator
383:             * @since 1.2
384:             */
385:            public CollationElementIterator getCollationElementIterator(
386:                    CharacterIterator source) {
387:                return new CollationElementIterator(source, this );
388:            }
389:
390:            /**
391:             * Compares the character data stored in two different strings based on the
392:             * collation rules.  Returns information about whether a string is less
393:             * than, greater than or equal to another string in a language.
394:             * This can be overriden in a subclass.
395:             */
396:            public synchronized int compare(String source, String target) {
397:                // The basic algorithm here is that we use CollationElementIterators
398:                // to step through both the source and target strings.  We compare each
399:                // collation element in the source string against the corresponding one
400:                // in the target, checking for differences.
401:                //
402:                // If a difference is found, we set <result> to LESS or GREATER to
403:                // indicate whether the source string is less or greater than the target.
404:                //
405:                // However, it's not that simple.  If we find a tertiary difference
406:                // (e.g. 'A' vs. 'a') near the beginning of a string, it can be
407:                // overridden by a primary difference (e.g. "A" vs. "B") later in
408:                // the string.  For example, "AA" < "aB", even though 'A' > 'a'.
409:                //
410:                // To keep track of this, we use strengthResult to keep track of the
411:                // strength of the most significant difference that has been found
412:                // so far.  When we find a difference whose strength is greater than
413:                // strengthResult, it overrides the last difference (if any) that
414:                // was found.
415:
416:                int result = Collator.EQUAL;
417:
418:                if (sourceCursor == null) {
419:                    sourceCursor = getCollationElementIterator(source);
420:                } else {
421:                    sourceCursor.setText(source);
422:                }
423:                if (targetCursor == null) {
424:                    targetCursor = getCollationElementIterator(target);
425:                } else {
426:                    targetCursor.setText(target);
427:                }
428:
429:                int sOrder = 0, tOrder = 0;
430:
431:                boolean initialCheckSecTer = getStrength() >= Collator.SECONDARY;
432:                boolean checkSecTer = initialCheckSecTer;
433:                boolean checkTertiary = getStrength() >= Collator.TERTIARY;
434:
435:                boolean gets = true, gett = true;
436:
437:                while (true) {
438:                    // Get the next collation element in each of the strings, unless
439:                    // we've been requested to skip it.
440:                    if (gets)
441:                        sOrder = sourceCursor.next();
442:                    else
443:                        gets = true;
444:                    if (gett)
445:                        tOrder = targetCursor.next();
446:                    else
447:                        gett = true;
448:
449:                    // If we've hit the end of one of the strings, jump out of the loop
450:                    if ((sOrder == CollationElementIterator.NULLORDER)
451:                            || (tOrder == CollationElementIterator.NULLORDER))
452:                        break;
453:
454:                    int pSOrder = CollationElementIterator.primaryOrder(sOrder);
455:                    int pTOrder = CollationElementIterator.primaryOrder(tOrder);
456:
457:                    // If there's no difference at this position, we can skip it
458:                    if (sOrder == tOrder) {
459:                        if (tables.isFrenchSec() && pSOrder != 0) {
460:                            if (!checkSecTer) {
461:                                // in french, a secondary difference more to the right is stronger,
462:                                // so accents have to be checked with each base element
463:                                checkSecTer = initialCheckSecTer;
464:                                // but tertiary differences are less important than the first
465:                                // secondary difference, so checking tertiary remains disabled
466:                                checkTertiary = false;
467:                            }
468:                        }
469:                        continue;
470:                    }
471:
472:                    // Compare primary differences first.
473:                    if (pSOrder != pTOrder) {
474:                        if (sOrder == 0) {
475:                            // The entire source element is ignorable.
476:                            // Skip to the next source element, but don't fetch another target element.
477:                            gett = false;
478:                            continue;
479:                        }
480:                        if (tOrder == 0) {
481:                            gets = false;
482:                            continue;
483:                        }
484:
485:                        // The source and target elements aren't ignorable, but it's still possible
486:                        // for the primary component of one of the elements to be ignorable....
487:
488:                        if (pSOrder == 0) // primary order in source is ignorable
489:                        {
490:                            // The source's primary is ignorable, but the target's isn't.  We treat ignorables
491:                            // as a secondary difference, so remember that we found one.
492:                            if (checkSecTer) {
493:                                result = Collator.GREATER; // (strength is SECONDARY)
494:                                checkSecTer = false;
495:                            }
496:                            // Skip to the next source element, but don't fetch another target element.
497:                            gett = false;
498:                        } else if (pTOrder == 0) {
499:                            // record differences - see the comment above.
500:                            if (checkSecTer) {
501:                                result = Collator.LESS; // (strength is SECONDARY)
502:                                checkSecTer = false;
503:                            }
504:                            // Skip to the next source element, but don't fetch another target element.
505:                            gets = false;
506:                        } else {
507:                            // Neither of the orders is ignorable, and we already know that the primary
508:                            // orders are different because of the (pSOrder != pTOrder) test above.
509:                            // Record the difference and stop the comparison.
510:                            if (pSOrder < pTOrder) {
511:                                return Collator.LESS; // (strength is PRIMARY)
512:                            } else {
513:                                return Collator.GREATER; // (strength is PRIMARY)
514:                            }
515:                        }
516:                    } else { // else of if ( pSOrder != pTOrder )
517:                        // primary order is the same, but complete order is different. So there
518:                        // are no base elements at this point, only ignorables (Since the strings are
519:                        // normalized)
520:
521:                        if (checkSecTer) {
522:                            // a secondary or tertiary difference may still matter
523:                            short secSOrder = CollationElementIterator
524:                                    .secondaryOrder(sOrder);
525:                            short secTOrder = CollationElementIterator
526:                                    .secondaryOrder(tOrder);
527:                            if (secSOrder != secTOrder) {
528:                                // there is a secondary difference
529:                                result = (secSOrder < secTOrder) ? Collator.LESS
530:                                        : Collator.GREATER;
531:                                // (strength is SECONDARY)
532:                                checkSecTer = false;
533:                                // (even in french, only the first secondary difference within
534:                                //  a base character matters)
535:                            } else {
536:                                if (checkTertiary) {
537:                                    // a tertiary difference may still matter
538:                                    short terSOrder = CollationElementIterator
539:                                            .tertiaryOrder(sOrder);
540:                                    short terTOrder = CollationElementIterator
541:                                            .tertiaryOrder(tOrder);
542:                                    if (terSOrder != terTOrder) {
543:                                        // there is a tertiary difference
544:                                        result = (terSOrder < terTOrder) ? Collator.LESS
545:                                                : Collator.GREATER;
546:                                        // (strength is TERTIARY)
547:                                        checkTertiary = false;
548:                                    }
549:                                }
550:                            }
551:                        } // if (checkSecTer)
552:
553:                    } // if ( pSOrder != pTOrder )
554:                } // while()
555:
556:                if (sOrder != CollationElementIterator.NULLORDER) {
557:                    // (tOrder must be CollationElementIterator::NULLORDER,
558:                    //  since this point is only reached when sOrder or tOrder is NULLORDER.)
559:                    // The source string has more elements, but the target string hasn't.
560:                    do {
561:                        if (CollationElementIterator.primaryOrder(sOrder) != 0) {
562:                            // We found an additional non-ignorable base character in the source string.
563:                            // This is a primary difference, so the source is greater
564:                            return Collator.GREATER; // (strength is PRIMARY)
565:                        } else if (CollationElementIterator
566:                                .secondaryOrder(sOrder) != 0) {
567:                            // Additional secondary elements mean the source string is greater
568:                            if (checkSecTer) {
569:                                result = Collator.GREATER; // (strength is SECONDARY)
570:                                checkSecTer = false;
571:                            }
572:                        }
573:                    } while ((sOrder = sourceCursor.next()) != CollationElementIterator.NULLORDER);
574:                } else if (tOrder != CollationElementIterator.NULLORDER) {
575:                    // The target string has more elements, but the source string hasn't.
576:                    do {
577:                        if (CollationElementIterator.primaryOrder(tOrder) != 0)
578:                            // We found an additional non-ignorable base character in the target string.
579:                            // This is a primary difference, so the source is less
580:                            return Collator.LESS; // (strength is PRIMARY)
581:                        else if (CollationElementIterator
582:                                .secondaryOrder(tOrder) != 0) {
583:                            // Additional secondary elements in the target mean the source string is less
584:                            if (checkSecTer) {
585:                                result = Collator.LESS; // (strength is SECONDARY)
586:                                checkSecTer = false;
587:                            }
588:                        }
589:                    } while ((tOrder = targetCursor.next()) != CollationElementIterator.NULLORDER);
590:                }
591:
592:                // For IDENTICAL comparisons, we use a bitwise character comparison
593:                // as a tiebreaker if all else is equal
594:                if (result == 0 && getStrength() == IDENTICAL) {
595:                    Normalizer.Mode mode = NormalizerUtilities
596:                            .toNormalizerMode(getDecomposition());
597:                    String sourceDecomposition = Normalizer.normalize(source,
598:                            mode, 0);
599:                    String targetDecomposition = Normalizer.normalize(target,
600:                            mode, 0);
601:                    result = sourceDecomposition.compareTo(targetDecomposition);
602:                }
603:                return result;
604:            }
605:
606:            /**
607:             * Transforms the string into a series of characters that can be compared
608:             * with CollationKey.compareTo. This overrides java.text.Collator.getCollationKey.
609:             * It can be overriden in a subclass.
610:             */
611:            public synchronized CollationKey getCollationKey(String source) {
612:                //
613:                // The basic algorithm here is to find all of the collation elements for each
614:                // character in the source string, convert them to a char representation,
615:                // and put them into the collation key.  
616:                // Each collation element in a string has three components: primary (A vs B),
617:                // secondary (A vs A-acute), and tertiary (A' vs a); and a primary difference
618:                // at the end of a string takes precedence over a secondary or tertiary
619:                // difference earlier in the string.
620:                //
621:                // To account for this, we put all of the primary orders at the beginning of the
622:                // string, followed by the secondary and tertiary orders, separated by nulls.
623:                //
624:                // Here's a hypothetical example, with the collation element represented as
625:                // a three-digit number, one digit for primary, one for secondary, etc.
626:                //
627:                // String:              A     a     B   \u00e9 <--(e-acute)
628:                // Collation Elements: 101   100   201  510
629:                //
630:                // Collation Key:      1125<null>0001<null>1010
631:                //
632:                // Secondary differences (accent marks) are compared
633:                // starting at the *end* of the string in languages with French secondary ordering.
634:                // But when comparing the accent marks on a single base character, they are compared
635:                // from the beginning.  To handle this, we reverse all of the accents that belong
636:                // to each base character, then we reverse the entire string of secondary orderings
637:                // at the end.  Taking the same example above, a French collator might return
638:                // this instead:
639:                //
640:                // Collation Key:      1125<null>1000<null>1010
641:                //
642:                if (source == null)
643:                    return null;
644:
645:                if (primResult == null) {
646:                    primResult = new StringBuffer();
647:                    secResult = new StringBuffer();
648:                    terResult = new StringBuffer();
649:                } else {
650:                    primResult.setLength(0);
651:                    secResult.setLength(0);
652:                    terResult.setLength(0);
653:                }
654:                int order = 0;
655:                boolean compareSec = (getStrength() >= Collator.SECONDARY);
656:                boolean compareTer = (getStrength() >= Collator.TERTIARY);
657:                int secOrder = CollationElementIterator.NULLORDER;
658:                int terOrder = CollationElementIterator.NULLORDER;
659:                int preSecIgnore = 0;
660:
661:                if (sourceCursor == null) {
662:                    sourceCursor = getCollationElementIterator(source);
663:                } else {
664:                    sourceCursor.setText(source);
665:                }
666:
667:                // walk through each character
668:                while ((order = sourceCursor.next()) != CollationElementIterator.NULLORDER) {
669:                    secOrder = CollationElementIterator.secondaryOrder(order);
670:                    terOrder = CollationElementIterator.tertiaryOrder(order);
671:                    if (!CollationElementIterator.isIgnorable(order)) {
672:                        primResult.append((char) (CollationElementIterator
673:                                .primaryOrder(order) + COLLATIONKEYOFFSET));
674:
675:                        if (compareSec) {
676:                            //
677:                            // accumulate all of the ignorable/secondary characters attached
678:                            // to a given base character
679:                            //
680:                            if (tables.isFrenchSec()
681:                                    && preSecIgnore < secResult.length()) {
682:                                //
683:                                // We're doing reversed secondary ordering and we've hit a base
684:                                // (non-ignorable) character.  Reverse any secondary orderings
685:                                // that applied to the last base character.  (see block comment above.)
686:                                //
687:                                RBCollationTables.reverse(secResult,
688:                                        preSecIgnore, secResult.length());
689:                            }
690:                            // Remember where we are in the secondary orderings - this is how far
691:                            // back to go if we need to reverse them later.
692:                            secResult
693:                                    .append((char) (secOrder + COLLATIONKEYOFFSET));
694:                            preSecIgnore = secResult.length();
695:                        }
696:                        if (compareTer) {
697:                            terResult
698:                                    .append((char) (terOrder + COLLATIONKEYOFFSET));
699:                        }
700:                    } else {
701:                        if (compareSec && secOrder != 0)
702:                            secResult
703:                                    .append((char) (secOrder
704:                                            + tables.getMaxSecOrder() + COLLATIONKEYOFFSET));
705:                        if (compareTer && terOrder != 0)
706:                            terResult
707:                                    .append((char) (terOrder
708:                                            + tables.getMaxTerOrder() + COLLATIONKEYOFFSET));
709:                    }
710:                }
711:                if (tables.isFrenchSec()) {
712:                    if (preSecIgnore < secResult.length()) {
713:                        // If we've accumlated any secondary characters after the last base character,
714:                        // reverse them.
715:                        RBCollationTables.reverse(secResult, preSecIgnore,
716:                                secResult.length());
717:                    }
718:                    // And now reverse the entire secResult to get French secondary ordering.
719:                    RBCollationTables.reverse(secResult, 0, secResult.length());
720:                }
721:                primResult.append((char) 0);
722:                secResult.append((char) 0);
723:                secResult.append(terResult.toString());
724:                primResult.append(secResult.toString());
725:
726:                if (getStrength() == IDENTICAL) {
727:                    primResult.append((char) 0);
728:                    Normalizer.Mode mode = NormalizerUtilities
729:                            .toNormalizerMode(getDecomposition());
730:                    primResult.append(Normalizer.normalize(source, mode, 0));
731:                }
732:                return new CollationKey(source, primResult.toString());
733:            }
734:
735:            /**
736:             * Standard override; no change in semantics.
737:             */
738:            public Object clone() {
739:                // if we know we're not actually a subclass of RuleBasedCollator
740:                // (this class really should have been made final), bypass
741:                // Object.clone() and use our "copy constructor".  This is faster.
742:                if (getClass() == RuleBasedCollator.class) {
743:                    return new RuleBasedCollator(this );
744:                } else {
745:                    RuleBasedCollator result = (RuleBasedCollator) super 
746:                            .clone();
747:                    result.primResult = null;
748:                    result.secResult = null;
749:                    result.terResult = null;
750:                    result.sourceCursor = null;
751:                    result.targetCursor = null;
752:                    return result;
753:                }
754:            }
755:
756:            /**
757:             * Compares the equality of two collation objects.
758:             * @param obj the table-based collation object to be compared with this.
759:             * @return true if the current table-based collation object is the same
760:             * as the table-based collation object obj; false otherwise.
761:             */
762:            public boolean equals(Object obj) {
763:                if (obj == null)
764:                    return false;
765:                if (!super .equals(obj))
766:                    return false; // super does class check
767:                RuleBasedCollator other = (RuleBasedCollator) obj;
768:                // all other non-transient information is also contained in rules.
769:                return (getRules().equals(other.getRules()));
770:            }
771:
772:            /**
773:             * Generates the hash code for the table-based collation object
774:             */
775:            public int hashCode() {
776:                return getRules().hashCode();
777:            }
778:
779:            /**
780:             * Allows CollationElementIterator access to the tables object
781:             */
782:            RBCollationTables getTables() {
783:                return tables;
784:            }
785:
786:            // ==============================================================
787:            // private
788:            // ==============================================================
789:
790:            final static int CHARINDEX = 0x70000000; // need look up in .commit()
791:            final static int EXPANDCHARINDEX = 0x7E000000; // Expand index follows
792:            final static int CONTRACTCHARINDEX = 0x7F000000; // contract indexes follow
793:            final static int UNMAPPED = 0xFFFFFFFF;
794:
795:            private final static int COLLATIONKEYOFFSET = 1;
796:
797:            private RBCollationTables tables = null;
798:
799:            // Internal objects that are cached across calls so that they don't have to
800:            // be created/destroyed on every call to compare() and getCollationKey()
801:            private StringBuffer primResult = null;
802:            private StringBuffer secResult = null;
803:            private StringBuffer terResult = null;
804:            private CollationElementIterator sourceCursor = null;
805:            private CollationElementIterator targetCursor = null;
806:        }