3196N/A * Copyright (c) 1997, 2010, Oracle and/or its affiliates. All rights reserved. 0N/A * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 0N/A * This code is free software; you can redistribute it and/or modify it 0N/A * under the terms of the GNU General Public License version 2 only, as 2362N/A * published by the Free Software Foundation. Oracle designates this 0N/A * particular file as subject to the "Classpath" exception as provided 2362N/A * by Oracle in the LICENSE file that accompanied this code. 0N/A * This code is distributed in the hope that it will be useful, but WITHOUT 0N/A * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 0N/A * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 0N/A * version 2 for more details (a copy is included in the LICENSE file that 0N/A * accompanied this code). 0N/A * You should have received a copy of the GNU General Public License version 0N/A * 2 along with this work; if not, write to the Free Software Foundation, 0N/A * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. 2362N/A * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA 2362N/A * or visit www.oracle.com if you need additional information or have any 0N/A * (C) Copyright Taligent, Inc. 1996, 1997 - All Rights Reserved 0N/A * (C) Copyright IBM Corp. 1996-1998 - All Rights Reserved 0N/A * The original version of this source code and documentation is copyrighted 0N/A * and owned by Taligent, Inc., a wholly-owned subsidiary of IBM. These 0N/A * materials are provided under terms of a License Agreement between Taligent 0N/A * and Sun. This technology is protected by multiple US and International 0N/A * patents. This notice and attribution to Taligent may not be removed. 0N/A * Taligent is a registered trademark of Taligent, Inc. 0N/A * The <code>RuleBasedCollator</code> class is a concrete subclass of 0N/A * <code>Collator</code> that provides a simple, data-driven, table 0N/A * collator. With this class you can create a customized table-based 0N/A * <code>Collator</code>. <code>RuleBasedCollator</code> maps 0N/A * characters to sort keys. 0N/A * <code>RuleBasedCollator</code> has the following restrictions 0N/A * for efficiency (other subclasses may be used for more complex languages) : 0N/A * <li>If a special collation rule controlled by a <modifier> is 0N/A specified it applies to the whole collator object. 0N/A * <li>All non-mentioned characters are at the end of the 0N/A * The collation table is composed of a list of collation rules, where each 0N/A * rule is of one of three forms: 0N/A * <relation> <text-argument> 0N/A * <reset> <text-argument> 0N/A * The definitions of the rule elements is as follows: 0N/A * <LI><strong>Text-Argument</strong>: A text-argument is any sequence of 0N/A * characters, excluding special characters (that is, common 0N/A * whitespace characters [0009-000D, 0020] and rule syntax characters 0N/A * [0021-002F, 003A-0040, 005B-0060, 007B-007E]). If those 0N/A * characters are desired, you can put them in single quotes 0N/A * (e.g. ampersand => '&'). Note that unquoted white space characters 0N/A * are ignored; e.g. <code>b c</code> is treated as <code>bc</code>. 0N/A * <LI><strong>Modifier</strong>: There are currently two modifiers that 0N/A * turn on special collation rules. 0N/A * <LI>'@' : Turns on backwards sorting of accents (secondary 0N/A * differences), as in French. 0N/A * <LI>'!' : Turns on Thai/Lao vowel-consonant swapping. If this 0N/A * rule is in force when a Thai vowel of the range 0N/A * \U0E40-\U0E44 precedes a Thai consonant of the range 0N/A * \U0E01-\U0E2E OR a Lao vowel of the range \U0EC0-\U0EC4 0N/A * precedes a Lao consonant of the range \U0E81-\U0EAE then 0N/A * the vowel is placed after the consonant for collation 0N/A * <p>'@' : Indicates that accents are sorted backwards, as in French. 0N/A * <LI><strong>Relation</strong>: The relations are the following: 0N/A * <LI>'<' : Greater, as a letter difference (primary) 0N/A * <LI>';' : Greater, as an accent difference (secondary) 0N/A * <LI>',' : Greater, as a case difference (tertiary) 0N/A * <LI><strong>Reset</strong>: There is a single reset 0N/A * which is used primarily for contractions and expansions, but which 0N/A * can also be used to add a modification at the end of a set of rules. 0N/A * <p>'&' : Indicates that the next rule follows the position to where 0N/A * the reset text-argument would be sorted. 0N/A * This sounds more complicated than it is in practice. For example, the 0N/A * following are equivalent ways of expressing the same thing: 0N/A * a < b & b < c 0N/A * a < c & a < b 0N/A * Notice that the order is important, as the subsequent item goes immediately 0N/A * after the text-argument. The following are not equivalent: 0N/A * a < b & a < c 0N/A * a < c & a < b 0N/A * Either the text-argument must already be present in the sequence, or some 0N/A * initial substring of the text-argument must be present. (e.g. "a < b & ae < 0N/A * e" is valid since "a" is present in the sequence before "ae" is reset). In 0N/A * this latter case, "ae" is not entered and treated as a single character; 0N/A * instead, "e" is sorted as if it were expanded to two characters: "a" 0N/A * followed by an "e". This difference appears in natural languages: in 0N/A * traditional Spanish "ch" is treated as though it contracts to a single 0N/A * character (expressed as "c < ch < d"), while in traditional German 0N/A * a-umlaut is treated as though it expanded to two characters 0N/A * (expressed as "a,A < b,B ... &ae;\u00e3&AE;\u00c3"). 0N/A * [\u00e3 and \u00c3 are, of course, the escape sequences for a-umlaut.] 0N/A * <strong>Ignorable Characters</strong> 0N/A * For ignorable characters, the first rule must start with a relation (the 0N/A * examples we have used above are really fragments; "a < b" really should be 0N/A * "< a < b"). If, however, the first relation is not "<", then all the all 0N/A * text-arguments up to the first "<" are ignorable. For example, ", - < a < b" 0N/A * makes "-" an ignorable character, as we saw earlier in the word 0N/A * "black-birds". In the samples for different languages, you see that most 0N/A * accents are ignorable. 0N/A * <p><strong>Normalization and Accents</strong> 0N/A * <code>RuleBasedCollator</code> automatically processes its rule table to 0N/A * include both pre-composed and combining-character versions of 0N/A * accented characters. Even if the provided rule string contains only 0N/A * base characters and separate combining accent characters, the pre-composed 0N/A * accented characters matching all canonical combinations of characters from 0N/A * the rule string will be entered in the table. 0N/A * This allows you to use a RuleBasedCollator to compare accented strings 0N/A * even when the collator is set to NO_DECOMPOSITION. There are two caveats, 0N/A * however. First, if the strings to be collated contain combining 0N/A * sequences that may not be in canonical order, you should set the collator to 0N/A * CANONICAL_DECOMPOSITION or FULL_DECOMPOSITION to enable sorting of 0N/A * combining sequences. Second, if the strings contain characters with 0N/A * compatibility decompositions (such as full-width and half-width forms), 0N/A * you must use FULL_DECOMPOSITION, since the rule tables only include 0N/A * canonical mappings. 0N/A * <p><strong>Errors</strong> 0N/A * The following are errors: 0N/A * <LI>A text-argument contains unquoted punctuation symbols 0N/A * (e.g. "a < b-c < d"). 0N/A * <LI>A relation or reset character not followed by a text-argument 0N/A * (e.g. "a < ,b"). 0N/A * <LI>A reset where the text-argument (or an initial substring of the 0N/A * text-argument) is not already in the sequence. 0N/A * (e.g. "a < b & e < f") 0N/A * If you produce one of these errors, a <code>RuleBasedCollator</code> throws 0N/A * a <code>ParseException</code>. 0N/A * <p><strong>Examples</strong> 0N/A * <p>Simple: "< a < b < c < d" 3196N/A * <p>Norwegian: "< a, A < b, B < c, C < d, D < e, E < f, F 3196N/A * < g, G < h, H < i, I < j, J < k, K < l, L 3196N/A * < m, M < n, N < o, O < p, P < q, Q < r, R 3196N/A * < s, S < t, T < u, U < v, V < w, W < x, X 3196N/A * < \u00E6, \u00C6 3196N/A * < \u00F8, \u00D8 3196N/A * < \u00E5 = a\u030A, \u00C5 = A\u030A; 0N/A * To create a <code>RuleBasedCollator</code> object with specialized 0N/A * rules tailored to your needs, you construct the <code>RuleBasedCollator</code> 0N/A * with the rules contained in a <code>String</code> object. For example: 0N/A * String simple = "< a< b< c< d"; 0N/A * RuleBasedCollator mySimple = new RuleBasedCollator(simple); 3196N/A * String Norwegian = "< a, A < b, B < c, C < d, D < e, E < f, F < g, G < h, H < i, I" + 3196N/A * "< j, J < k, K < l, L < m, M < n, N < o, O < p, P < q, Q < r, R" + 3196N/A * "< s, S < t, T < u, U < v, V < w, W < x, X < y, Y < z, Z" + 3196N/A * "< \u00E6, \u00C6" + // Latin letter ae & AE 3196N/A * "< \u00F8, \u00D8" + // Latin letter o & O with stroke 3196N/A * "< \u00E5 = a\u030A," + // Latin letter a with ring above 3196N/A * " \u00C5 = A\u030A;" + // Latin letter A with ring above 0N/A * RuleBasedCollator myNorwegian = new RuleBasedCollator(Norwegian); 0N/A * A new collation rules string can be created by concatenating rules 0N/A * strings. For example, the rules returned by {@link #getRules()} could 0N/A * be concatenated to combine multiple <code>RuleBasedCollator</code>s. 0N/A * The following example demonstrates how to change the order of 0N/A * non-spacing accents, 0N/A * String oldRules = "=\u0301;\u0300;\u0302;\u0308" // main accents 0N/A * + ";\u0327;\u0303;\u0304;\u0305" // main accents 0N/A * + ";\u0306;\u0307;\u0309;\u030A" // main accents 0N/A * + ";\u030B;\u030C;\u030D;\u030E" // main accents 0N/A * + ";\u030F;\u0310;\u0311;\u0312" // main accents 0N/A * + "< a , A ; ae, AE ; \u00e6 , \u00c6" 0N/A * + "< b , B < c, C < e, E & C < d, D"; 0N/A * // change the order of accent characters 0N/A * String addOn = "& \u0300 ; \u0308 ; \u0302"; 0N/A * RuleBasedCollator myCollator = new RuleBasedCollator(oldRules + addOn); 0N/A * @see CollationElementIterator 0N/A * @author Helena Shih, Laura Werner, Richard Gillam 0N/A // IMPLEMENTATION NOTES: The implementation of the collation algorithm is 0N/A // divided across three classes: RuleBasedCollator, RBCollationTables, and 0N/A // CollationElementIterator. RuleBasedCollator contains the collator's 0N/A // transient state and includes the code that uses the other classes to 0N/A // implement comparison and sort-key building. RuleBasedCollator also 0N/A // contains the logic to handle French secondary accent sorting. 0N/A // A RuleBasedCollator has two CollationElementIterators. State doesn't 0N/A // need to be preserved in these objects between calls to compare() or 0N/A // getCollationKey(), but the objects persist anyway to avoid wasting extra 0N/A // creation time. compare() and getCollationKey() are synchronized to ensure 0N/A // thread safety with this scheme. The CollationElementIterator is responsible 0N/A // for generating collation elements from strings and returning one element at 0N/A // a time (sometimes there's a one-to-many or many-to-one mapping between 0N/A // characters and collation elements-- this class handles that). 0N/A // CollationElementIterator depends on RBCollationTables, which contains the 0N/A // collator's static state. RBCollationTables contains the actual data 0N/A // tables specifying the collation order of characters for a particular locale 0N/A // or use. It also contains the base logic that CollationElementIterator 0N/A // uses to map from characters to collation elements. A single RBCollationTables 0N/A // object is shared among all RuleBasedCollators for the same locale, and 0N/A // thus by all the CollationElementIterators they create. 0N/A * RuleBasedCollator constructor. This takes the table rules and builds 0N/A * a collation table out of them. Please see RuleBasedCollator class 0N/A * description for more details on the collation rule syntax. 0N/A * @see java.util.Locale 0N/A * @param rules the collation rules to build the collation table from. 0N/A * @exception ParseException A format exception 0N/A * will be thrown if the build process of the rules fails. For 0N/A * example, build rule "a < ? < d" will cause the constructor to 0N/A * throw the ParseException because the '?' is not quoted. 0N/A * RuleBasedCollator constructor. This takes the table rules and builds 0N/A * a collation table out of them. Please see RuleBasedCollator class 0N/A * description for more details on the collation rule syntax. 0N/A * @see java.util.Locale 0N/A * @param rules the collation rules to build the collation table from. 0N/A * @param decomp the decomposition strength used to build the 0N/A * collation table and to perform comparisons. 0N/A * @exception ParseException A format exception 0N/A * will be thrown if the build process of the rules fails. For 0N/A * example, build rule "a < ? < d" will cause the constructor to 0N/A * throw the ParseException because the '?' is not quoted. 0N/A * "Copy constructor." Used in clone() for performance. 0N/A * Gets the table-based rules for the collation object. 0N/A * @return returns the collation rules that the table collation object 0N/A * Return a CollationElementIterator for the given String. 0N/A * @see java.text.CollationElementIterator 0N/A * Return a CollationElementIterator for the given String. 0N/A * @see java.text.CollationElementIterator 0N/A * Compares the character data stored in two different strings based on the 0N/A * collation rules. Returns information about whether a string is less 0N/A * than, greater than or equal to another string in a language. 0N/A * This can be overriden in a subclass. 2868N/A * @exception NullPointerException if <code>source</code> or <code>target</code> is null. 0N/A // The basic algorithm here is that we use CollationElementIterators 0N/A // to step through both the source and target strings. We compare each 0N/A // collation element in the source string against the corresponding one 0N/A // in the target, checking for differences. 0N/A // If a difference is found, we set <result> to LESS or GREATER to 0N/A // indicate whether the source string is less or greater than the target. 0N/A // However, it's not that simple. If we find a tertiary difference 0N/A // (e.g. 'A' vs. 'a') near the beginning of a string, it can be 0N/A // overridden by a primary difference (e.g. "A" vs. "B") later in 0N/A // the string. For example, "AA" < "aB", even though 'A' > 'a'. 0N/A // To keep track of this, we use strengthResult to keep track of the 0N/A // strength of the most significant difference that has been found 0N/A // so far. When we find a difference whose strength is greater than 0N/A // strengthResult, it overrides the last difference (if any) that 0N/A // Get the next collation element in each of the strings, unless 0N/A // we've been requested to skip it. 0N/A // If we've hit the end of one of the strings, jump out of the loop 0N/A // If there's no difference at this position, we can skip it 0N/A // in french, a secondary difference more to the right is stronger, 0N/A // so accents have to be checked with each base element 0N/A // but tertiary differences are less important than the first 0N/A // secondary difference, so checking tertiary remains disabled 0N/A // Compare primary differences first. 0N/A // The entire source element is ignorable. 0N/A // Skip to the next source element, but don't fetch another target element. 0N/A // The source and target elements aren't ignorable, but it's still possible 0N/A // for the primary component of one of the elements to be ignorable.... 0N/A if (
pSOrder ==
0)
// primary order in source is ignorable 0N/A // The source's primary is ignorable, but the target's isn't. We treat ignorables 0N/A // as a secondary difference, so remember that we found one. 0N/A // Skip to the next source element, but don't fetch another target element. 0N/A // record differences - see the comment above. 0N/A // Skip to the next source element, but don't fetch another target element. 0N/A // Neither of the orders is ignorable, and we already know that the primary 0N/A // orders are different because of the (pSOrder != pTOrder) test above. 0N/A // Record the difference and stop the comparison. 0N/A }
else {
// else of if ( pSOrder != pTOrder ) 0N/A // primary order is the same, but complete order is different. So there 0N/A // are no base elements at this point, only ignorables (Since the strings are 0N/A // a secondary or tertiary difference may still matter 0N/A // there is a secondary difference 0N/A // (strength is SECONDARY) 0N/A // (even in french, only the first secondary difference within 0N/A // a base character matters) 0N/A // a tertiary difference may still matter 0N/A // there is a tertiary difference 0N/A // (strength is TERTIARY) 0N/A }
// if (checkSecTer) 0N/A }
// if ( pSOrder != pTOrder ) 0N/A // (tOrder must be CollationElementIterator::NULLORDER, 0N/A // since this point is only reached when sOrder or tOrder is NULLORDER.) 0N/A // The source string has more elements, but the target string hasn't. 0N/A // We found an additional non-ignorable base character in the source string. 0N/A // This is a primary difference, so the source is greater 0N/A // Additional secondary elements mean the source string is greater 0N/A // The target string has more elements, but the source string hasn't. 0N/A // We found an additional non-ignorable base character in the target string. 0N/A // This is a primary difference, so the source is less 0N/A // Additional secondary elements in the target mean the source string is less 0N/A // For IDENTICAL comparisons, we use a bitwise character comparison 0N/A // as a tiebreaker if all else is equal 0N/A * Transforms the string into a series of characters that can be compared 0N/A * with CollationKey.compareTo. This overrides java.text.Collator.getCollationKey. 0N/A * It can be overriden in a subclass. 0N/A // The basic algorithm here is to find all of the collation elements for each 0N/A // character in the source string, convert them to a char representation, 0N/A // and put them into the collation key. But it's trickier than that. 0N/A // Each collation element in a string has three components: primary (A vs B), 0N/A // secondary (A vs A-acute), and tertiary (A' vs a); and a primary difference 0N/A // at the end of a string takes precedence over a secondary or tertiary 0N/A // difference earlier in the string. 0N/A // To account for this, we put all of the primary orders at the beginning of the 0N/A // string, followed by the secondary and tertiary orders, separated by nulls. 0N/A // Here's a hypothetical example, with the collation element represented as 0N/A // a three-digit number, one digit for primary, one for secondary, etc. 0N/A // String: A a B \u00e9 <--(e-acute) 0N/A // Collation Elements: 101 100 201 510 0N/A // Collation Key: 1125<null>0001<null>1010 0N/A // To make things even trickier, secondary differences (accent marks) are compared 0N/A // starting at the *end* of the string in languages with French secondary ordering. 0N/A // But when comparing the accent marks on a single base character, they are compared 0N/A // from the beginning. To handle this, we reverse all of the accents that belong 0N/A // to each base character, then we reverse the entire string of secondary orderings 0N/A // at the end. Taking the same example above, a French collator might return 0N/A // Collation Key: 1125<null>1000<null>1010 0N/A // walk through each character 0N/A // to a given base character 0N/A // We're doing reversed secondary ordering and we've hit a base 0N/A // (non-ignorable) character. Reverse any secondary orderings 0N/A // that applied to the last base character. (see block comment above.) 0N/A // Remember where we are in the secondary orderings - this is how far 0N/A // back to go if we need to reverse them later. 0N/A // If we've accumlated any secondary characters after the last base character, 0N/A // And now reverse the entire secResult to get French secondary ordering. 0N/A * Standard override; no change in semantics. 0N/A // if we know we're not actually a subclass of RuleBasedCollator 0N/A // (this class really should have been made final), bypass 0N/A // Object.clone() and use our "copy constructor". This is faster. 0N/A * Compares the equality of two collation objects. 0N/A * @param obj the table-based collation object to be compared with this. 0N/A * @return true if the current table-based collation object is the same 0N/A * as the table-based collation object obj; false otherwise. 0N/A if (!
super.
equals(
obj))
return false;
// super does class check 0N/A // all other non-transient information is also contained in rules. 0N/A * Generates the hash code for the table-based collation object 0N/A * Allows CollationElementIterator access to the tables object 0N/A // ============================================================== 0N/A // ============================================================== 0N/A final static int CHARINDEX =
0x70000000;
// need look up in .commit() 0N/A // Internal objects that are cached across calls so that they don't have to