1. /*

  2.  * @(#)CollationRules.java	1.21 01/11/29

  3.  *

  4.  * Copyright 2002 Sun Microsystems, Inc. All rights reserved.

  5.  * SUN PROPRIETARY/CONFIDENTIAL. Use is subject to license terms.

  6.  */

  7. 

  8. /*

  9.  * @(#)CollationRules.java	1.19 98/07/23

  10.  *

  11.  * (C) Copyright Taligent, Inc. 1996,1997 - All Rights Reserved

  12.  * (C) Copyright IBM Corp. 1996, 1997 - All Rights Reserved

  13.  *

  14.  * Portions copyright (c) 1996-1998 Sun Microsystems, Inc. All Rights Reserved.

  15.  *

  16.  *   The original version of this source code and documentation is copyrighted

  17.  * and owned by Taligent, Inc., a wholly-owned subsidiary of IBM. These

  18.  * materials are provided under terms of a License Agreement between Taligent

  19.  * and Sun. This technology is protected by multiple US and International

  20.  * patents. This notice and attribution to Taligent may not be removed.

  21.  *   Taligent is a registered trademark of Taligent, Inc.

  22.  *

  23.  * Permission to use, copy, modify, and distribute this software

  24.  * and its documentation for NON-COMMERCIAL purposes and without

  25.  * fee is hereby granted provided that this copyright notice

  26.  * appears in all copies. Please refer to the file "copyright.html"

  27.  * for further important copyright and licensing information.

  28.  *

  29.  * SUN MAKES NO REPRESENTATIONS OR WARRANTIES ABOUT THE SUITABILITY OF

  30.  * THE SOFTWARE, EITHER EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED

  31.  * TO THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A

  32.  * PARTICULAR PURPOSE, OR NON-INFRINGEMENT. SUN SHALL NOT BE LIABLE FOR

  33.  * ANY DAMAGES SUFFERED BY LICENSEE AS A RESULT OF USING, MODIFYING OR

  34.  * DISTRIBUTING THIS SOFTWARE OR ITS DERIVATIVES.

  35.  *

  36.  */

  37. 

  38. package java.text;

  39. /**

  40.  * CollationRules contains the default en_US collation rules as a base

  41.  * for building other collation tables.

  42.  * <p>Note that decompositions are done before these rules are used,

  43.  * so they do not have to contain accented characters, such as A-grave.

  44.  * @see                RuleBasedCollator

  45.  * @see                LocaleElements

  46.  * @version            1.19 07/23/98

  47.  * @author             Helena Shih, Mark Davis

  48.  */

  49. final class CollationRules {

  50.     final static String DEFAULTRULES = new String(

  51.         "" // no FRENCH accent order by default, add in French Delta

  52.         // IGNORABLES (up to first < character)

  53.         // COMPLETELY IGNORE format characters

  54.         + "='\u200B'=\u200C=\u200D=\u200E=\u200F"

  55.         // Control Characters

  56.         + "=\u0000 =\u0001 =\u0002 =\u0003 =\u0004" //null, .. eot

  57.         + "=\u0005 =\u0006 =\u0007 =\u0008 ='\u0009'" //enq, ...

  58.         + "='\u000b' =\u000e" //vt,, so

  59.         + "=\u000f ='\u0010' =\u0011 =\u0012 =\u0013" //si, dle, dc1, dc2, dc3

  60.         + "=\u0014 =\u0015 =\u0016 =\u0017 =\u0018" //dc4, nak, syn, etb, can

  61.         + "=\u0019 =\u001a =\u001b =\u001c =\u001d" //em, sub, esc, fs, gs

  62.         + "=\u001e =\u001f =\u007f"                   //rs, us, del

  63.         //....then the C1 Latin 1 reserved control codes

  64.         + "=\u0080 =\u0081 =\u0082 =\u0083 =\u0084 =\u0085"

  65.         + "=\u0086 =\u0087 =\u0088 =\u0089 =\u008a =\u008b"

  66.         + "=\u008c =\u008d =\u008e =\u008f =\u0090 =\u0091"

  67.         + "=\u0092 =\u0093 =\u0094 =\u0095 =\u0096 =\u0097"

  68.         + "=\u0098 =\u0099 =\u009a =\u009b =\u009c =\u009d"

  69.         + "=\u009e =\u009f"

  70.         // IGNORE except for secondary, tertiary difference

  71.         // Spaces

  72.         + ";'\u0020';'\u00A0'"                  // spaces

  73.         + ";'\u2000';'\u2001';'\u2002';'\u2003';'\u2004'"  // spaces

  74.         + ";'\u2005';'\u2006';'\u2007';'\u2008';'\u2009'"  // spaces

  75.         + ";'\u200A';'\u3000';'\uFEFF'"                // spaces

  76.         + ";'\r' ;'\t' ;'\n';'\f';'\u000b'"  // whitespace

  77. 

  78.         // Non-spacing accents

  79. 

  80.         + ";\u0301"          // non-spacing acute accent

  81.         + ";\u0300"          // non-spacing grave accent

  82.         + ";\u0306"          // non-spacing breve accent

  83.         + ";\u0302"          // non-spacing circumflex accent

  84.         + ";\u030c"          // non-spacing caron/hacek accent

  85.         + ";\u030a"          // non-spacing ring above accent

  86.         + ";\u030d"          // non-spacing vertical line above

  87.         + ";\u0308"          // non-spacing diaeresis accent

  88.         + ";\u030b"          // non-spacing double acute accent

  89.         + ";\u0303"          // non-spacing tilde accent

  90.         + ";\u0307"          // non-spacing dot above/overdot accent

  91.         + ";\u0304"          // non-spacing macron accent

  92.         + ";\u0337"          // non-spacing short slash overlay (overstruck diacritic)

  93.         + ";\u0327"          // non-spacing cedilla accent

  94.         + ";\u0328"          // non-spacing ogonek accent

  95.         + ";\u0323"          // non-spacing dot-below/underdot accent

  96.         + ";\u0332"          // non-spacing underscore/underline accent

  97.         // with the rest of the general diacritical marks in binary order

  98.         + ";\u0305"          // non-spacing overscore/overline

  99.         + ";\u0309"          // non-spacing hook above

  100.         + ";\u030e"          // non-spacing double vertical line above

  101.         + ";\u030f"          // non-spacing double grave

  102.         + ";\u0310"          // non-spacing chandrabindu

  103.         + ";\u0311"          // non-spacing inverted breve

  104.         + ";\u0312"          // non-spacing turned comma above/cedilla above

  105.         + ";\u0313"          // non-spacing comma above

  106.         + ";\u0314"          // non-spacing reversed comma above

  107.         + ";\u0315"          // non-spacing comma above right

  108.         + ";\u0316"          // non-spacing grave below

  109.         + ";\u0317"          // non-spacing acute below

  110.         + ";\u0318"          // non-spacing left tack below

  111.         + ";\u0319"          // non-spacing tack below

  112.         + ";\u031a"          // non-spacing left angle above

  113.         + ";\u031b"          // non-spacing horn

  114.         + ";\u031c"          // non-spacing left half ring below

  115.         + ";\u031d"          // non-spacing up tack below

  116.         + ";\u031e"          // non-spacing down tack below

  117.         + ";\u031f"          // non-spacing plus sign below

  118.         + ";\u0320"          // non-spacing minus sign below

  119.         + ";\u0321"          // non-spacing palatalized hook below

  120.         + ";\u0322"          // non-spacing retroflex hook below

  121.         + ";\u0324"          // non-spacing double dot below

  122.         + ";\u0325"          // non-spacing ring below

  123.         + ";\u0326"          // non-spacing comma below

  124.         + ";\u0329"          // non-spacing vertical line below

  125.         + ";\u032a"          // non-spacing bridge below

  126.         + ";\u032b"          // non-spacing inverted double arch below

  127.         + ";\u032c"          // non-spacing hacek below

  128.         + ";\u032d"          // non-spacing circumflex below

  129.         + ";\u032e"          // non-spacing breve below

  130.         + ";\u032f"          // non-spacing inverted breve below

  131.         + ";\u0330"          // non-spacing tilde below

  132.         + ";\u0331"          // non-spacing macron below

  133.         + ";\u0333"          // non-spacing double underscore

  134.         + ";\u0334"          // non-spacing tilde overlay

  135.         + ";\u0335"          // non-spacing short bar overlay

  136.         + ";\u0336"          // non-spacing long bar overlay

  137.         + ";\u0338"          // non-spacing long slash overlay

  138.         + ";\u0339"          // non-spacing right half ring below

  139.         + ";\u033a"          // non-spacing inverted bridge below

  140.         + ";\u033b"          // non-spacing square below

  141.         + ";\u033c"          // non-spacing seagull below

  142.         + ";\u033d"          // non-spacing x above

  143.         + ";\u033e"          // non-spacing vertical tilde

  144.         + ";\u033f"          // non-spacing double overscore

  145.         + ";\u0340"          // non-spacing grave tone mark

  146.         + ";\u0341"          // non-spacing acute tone mark

  147.         + ";\u0342;\u0343;\u0344;\u0345;\u0360;\u0361"    // newer

  148.         + ";\u0483;\u0484;\u0485;\u0486"    // Cyrillic accents

  149. 

  150.         + ";\u20D0;\u20D1;\u20D2"           // symbol accents

  151.         + ";\u20D3;\u20D4;\u20D5"           // symbol accents

  152.         + ";\u20D6;\u20D7;\u20D8"           // symbol accents

  153.         + ";\u20D9;\u20DA;\u20DB"           // symbol accents

  154.         + ";\u20DC;\u20DD;\u20DE"           // symbol accents

  155.         + ";\u20DF;\u20E0;\u20E1"           // symbol accents

  156. 

  157.         + ",'\u002D';\u00AD"                // dashes

  158.         + ";\u2010;\u2011;\u2012"           // dashes

  159.         + ";\u2013;\u2014;\u2015"           // dashes

  160.         + ";\u2212"                         // dashes

  161. 

  162.         // other punctuation

  163. 

  164.         + "<'\u005f'"        // underline/underscore (spacing)

  165.         + "<\u00af"          // overline or macron (spacing)

  166.         + "<'\u002c'"        // comma (spacing)

  167.         + "<'\u003b'"        // semicolon

  168.         + "<'\u003a'"        // colon

  169.         + "<'\u0021'"        // exclamation point

  170.         + "<\u00a1"          // inverted exclamation point

  171.         + "<'\u003f'"        // question mark

  172.         + "<\u00bf"          // inverted question mark

  173.         + "<'\u002f'"        // slash

  174.         + "<'\u002e'"        // period/full stop

  175.         + "<\u00b4"          // acute accent (spacing)

  176.         + "<'\u0060'"        // grave accent (spacing)

  177.         + "<'\u005e'"        // circumflex accent (spacing)

  178.         + "<\u00a8"          // diaresis/umlaut accent (spacing)

  179.         + "<'\u007e'"        // tilde accent (spacing)

  180.         + "<\u00b7"          // middle dot (spacing)

  181.         + "<\u00b8"          // cedilla accent (spacing)

  182.         + "<'\u0027'"        // apostrophe

  183.         + "<'\"'"            // quotation marks

  184.         + "<\u00ab"          // left angle quotes

  185.         + "<\u00bb"          // right angle quotes

  186.         + "<'\u0028'"        // left parenthesis

  187.         + "<'\u0029'"        // right parenthesis

  188.         + "<'\u005b'"        // left bracket

  189.         + "<'\u005d'"        // right bracket

  190.         + "<'\u007b'"        // left brace

  191.         + "<'\u007d'"        // right brace

  192.         + "<\u00a7"          // section symbol

  193.         + "<\u00b6"          // paragraph symbol

  194.         + "<\u00a9"          // copyright symbol

  195.         + "<\u00ae"          // registered trademark symbol

  196.         + "<'\u0040'"          // at sign

  197.         + "<\u00a4"          // international currency symbol

  198.         + "<\u0e3f"          // baht sign

  199.         + "<\u00a2"          // cent sign

  200.         + "<\u20a1"          // colon sign

  201.         + "<\u20a2"          // cruzeiro sign

  202.         + "<'\u0024'"        // dollar sign

  203.         + "<\u20ab"          // dong sign

  204.         + "<\u20ac"          // euro sign

  205.         + "<\u20a3"          // franc sign

  206.         + "<\u20a4"          // lira sign

  207.         + "<\u20a5"          // mill sign

  208.         + "<\u20a6"          // naira sign

  209.         + "<\u20a7"          // peseta sign        

  210.         + "<\u00a3"          // pound-sterling sign

  211.         + "<\u20a8"          // rupee sign

  212.         + "<\u20aa"          // new shekel sign

  213.         + "<\u20a9"          // won sign

  214.         + "<\u00a5"          // yen sign

  215.         + "<'\u002a'"        // asterisk

  216.         + "<'\\'"            // backslash

  217.         + "<'\u0026'"        // ampersand

  218.         + "<'\u0023'"        // number sign

  219.         + "<'\u0025'"        // percent sign

  220.         + "<'\u002b'"        // plus sign

  221.         + "<\u00b1"          // plus-or-minus sign

  222.         + "<\u00f7"          // divide sign

  223.         + "<\u00d7"          // multiply sign

  224.         + "<'\u003c'"        // less-than sign

  225.         + "<'\u003d'"        // equal sign

  226.         + "<'\u003e'"        // greater-than sign

  227.         + "<\u00ac"          // end of line symbol/logical NOT symbol

  228.         + "<'\u007c'"          // vertical line/logical OR symbol

  229.         + "<\u00a6"          // broken vertical line

  230.         + "<\u00b0"          // degree symbol

  231.         + "<\u00b5"          // micro symbol

  232. 

  233.         // NUMERICS

  234. 

  235.         + "<0<1<2<3<4<5<6<7<8<9"

  236.         + "<\u00bc<\u00bd<\u00be"   // 1/4,1/2,3/4 fractions

  237. 

  238.         // NON-IGNORABLES

  239.         + "<a,A"

  240.         + "<b,B"

  241.         + "<c,C"

  242.         + "<d,D"

  243.         + "<\u00F0,\u00D0"                  // eth

  244.         + "<e,E"

  245.         + "<f,F"

  246.         + "<g,G"

  247.         + "<h,H"

  248.         + "<i,I"

  249.         + "<j,J"

  250.         + "<k,K"

  251.         + "<l,L"

  252.         + "<m,M"

  253.         + "<n,N"

  254.         + "<o,O"

  255.         + "<p,P"

  256.         + "<q,Q"

  257.         + "<r,R"

  258.         + "<s, S & SS,\u00DF"             // s-zet

  259.         + "<t,T"

  260.         + "& TH, \u00DE &TH, \u00FE "     // thorn

  261.         + "<u,U"

  262.         + "<v,V"

  263.         + "<w,W"

  264.         + "<x,X"

  265.         + "<y,Y"

  266.         + "<z,Z"

  267.         + "&AE,\u00C6"                    // ae & AE ligature

  268.         + "&AE,\u00E6"

  269.         + "&OE,\u0152"                    // oe & OE ligature

  270.         + "&OE,\u0153"

  271.     );

  272. }