diff --git a/icu4j/src/com/ibm/icu/dev/test/normalizer/BasicTest.java b/icu4j/src/com/ibm/icu/dev/test/normalizer/BasicTest.java index 070fe55598..df4575aefa 100755 --- a/icu4j/src/com/ibm/icu/dev/test/normalizer/BasicTest.java +++ b/icu4j/src/com/ibm/icu/dev/test/normalizer/BasicTest.java @@ -5,8 +5,8 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/dev/test/normalizer/BasicTest.java,v $ - * $Date: 2002/10/31 01:02:42 $ - * $Revision: 1.19 $ + * $Date: 2002/11/15 21:04:20 $ + * $Revision: 1.20 $ * ***************************************************************************************** */ @@ -2004,17 +2004,17 @@ public class BasicTest extends TestFmwk { }; TestStruct[] tests= new TestStruct[]{ - new TestStruct( 0x037A, new String( "\u0020\u03B9") ), - new TestStruct( 0x03D2, new String( "\u03C5") ), - new TestStruct( 0x20A8, new String( "\u0072\u0073") ) , - new TestStruct( 0x210B, new String( "\u0068") ), - new TestStruct( 0x210C, new String( "\u0068") ), - new TestStruct( 0x2121, new String( "\u0074\u0065\u006C") ), - new TestStruct( 0x2122, new String( "\u0074\u006D") ), - new TestStruct( 0x2128, new String( "\u007A") ), - new TestStruct( 0x1D5DB, new String( "\u0068") ), - new TestStruct( 0x1D5ED, new String( "\u007A") ), - new TestStruct( 0x0061, new String() ) + new TestStruct( 0x037A, "\u0020\u03B9" ), + new TestStruct( 0x03D2, "\u03C5" ), + new TestStruct( 0x20A8, "\u0072\u0073" ) , + new TestStruct( 0x210B, "\u0068" ), + new TestStruct( 0x210C, "\u0068" ), + new TestStruct( 0x2121, "\u0074\u0065\u006C" ), + new TestStruct( 0x2122, "\u0074\u006D" ), + new TestStruct( 0x2128, "\u007A" ), + new TestStruct( 0x1D5DB,"\u0068" ), + new TestStruct( 0x1D5ED,"\u007A" ), + new TestStruct( 0x0061, "" ) }; @@ -2028,8 +2028,9 @@ public class BasicTest extends TestFmwk { /* error handling */ int length=Normalizer.getFC_NFKC_Closure(0x5c, null); - - + if(length!=0){ + errln("getFC_NFKC_Closure did not perform error handling correctly"); + } } public void TestBugJ2324(){ /* String[] input = new String[]{ @@ -2052,4 +2053,519 @@ public class BasicTest extends TestFmwk { } } + /* + UNSAFE_STARTS[D] = new UnicodeSet("[\u0F73\u0F75\u0F81]", false); + UNSAFE_STARTS[C] = new UnicodeSet("[\u09BE\u09D7\u0B3E\u0B56-\u0B57\u0BBE\u0BD7\u0CC2\u0CD5-\u0CD6" + + "\u0D3E\u0D57\u0DCF\u0DDF\u0F73\u0F75\u0F81\u102E\u1161-\u1175\u11A8-\u11C2]", false); + UNSAFE_STARTS[KD] = new UnicodeSet("[\u0F73\u0F75\u0F81\uFF9E-\uFF9F]", false); + UNSAFE_STARTS[KC] = new UnicodeSet("[\u09BE\u09D7\u0B3E\u0B56-\u0B57\u0BBE\u0BD7\u0CC2\u0CD5-\u0CD6" + + "\u0D3E\u0D57\u0DCF\u0DDF\u0F73\u0F75\u0F81\u102E\u1161-\u1175\u11A8-\u11C2\u3133\u3135-\u3136" + + "\u313A-\u313F\u314F-\u3163\uFF9E-\uFF9F\uFFA3\uFFA5-\uFFA6\uFFAA-\uFFAF\uFFC2-\uFFC7\uFFCA-\uFFCF" + + "\uFFD2-\uFFD7\uFFDA-\uFFDC]", false); + */ + static final int D = 0, C = 1, KD= 2, KC = 3; + private static UnicodeSet[] initSkippables(UnicodeSet[] skippables){ + if( skippables.length < 4 ){ + return null; + } + skippables[D] = new UnicodeSet( + "[^\\u00C0-\\u00C5\\u00C7-\\u00CF\\u00D1-\\u00D6\\u00D9-\\u00DD" + + "\\u00E0-\\u00E5\\u00E7-\\u00EF\\u00F1-\\u00F6\\u00F9-\\u00FD" + + "\\u00FF-\\u010F\\u0112-\\u0125\\u0128-\\u0130\\u0134-\\u0137" + + "\\u0139-\\u013E\\u0143-\\u0148\\u014C-\\u0151\\u0154-\\u0165" + + "\\u0168-\\u017E\\u01A0-\\u01A1\\u01AF-\\u01B0\\u01CD-\\u01DC" + + "\\u01DE-\\u01E3\\u01E6-\\u01F0\\u01F4-\\u01F5\\u01F8-\\u021B" + + "\\u021E-\\u021F\\u0226-\\u0233\\u0300-\\u034E\\u0360-\\u036F" + + "\\u0374\\u037E\\u0385-\\u038A\\u038C\\u038E-\\u0390\\u03AA-" + + "\\u03B0\\u03CA-\\u03CE\\u03D3-\\u03D4\\u0400-\\u0401\\u0403" + + "\\u0407\\u040C-\\u040E\\u0419\\u0439\\u0450-\\u0451\\u0453" + + "\\u0457\\u045C-\\u045E\\u0476-\\u0477\\u0483-\\u0486\\u04C1-" + + "\\u04C2\\u04D0-\\u04D3\\u04D6-\\u04D7\\u04DA-\\u04DF\\u04E2-" + + "\\u04E7\\u04EA-\\u04F5\\u04F8-\\u04F9\\u0591-\\u05A1\\u05A3-" + + "\\u05B9\\u05BB-\\u05BD\\u05BF\\u05C1-\\u05C2\\u05C4\\u0622-" + + "\\u0626\\u064B-\\u0655\\u0670\\u06C0\\u06C2\\u06D3\\u06D6-" + + "\\u06DC\\u06DF-\\u06E4\\u06E7-\\u06E8\\u06EA-\\u06ED\\u0711" + + "\\u0730-\\u074A\\u0929\\u0931\\u0934\\u093C\\u094D\\u0951-" + + "\\u0954\\u0958-\\u095F\\u09BC\\u09CB-\\u09CD\\u09DC-\\u09DD" + + "\\u09DF\\u0A33\\u0A36\\u0A3C\\u0A4D\\u0A59-\\u0A5B\\u0A5E\\u0ABC" + + "\\u0ACD\\u0B3C\\u0B48\\u0B4B-\\u0B4D\\u0B5C-\\u0B5D\\u0B94" + + "\\u0BCA-\\u0BCD\\u0C48\\u0C4D\\u0C55-\\u0C56\\u0CC0\\u0CC7-" + + "\\u0CC8\\u0CCA-\\u0CCB\\u0CCD\\u0D4A-\\u0D4D\\u0DCA\\u0DDA" + + "\\u0DDC-\\u0DDE\\u0E38-\\u0E3A\\u0E48-\\u0E4B\\u0EB8-\\u0EB9" + + "\\u0EC8-\\u0ECB\\u0F18-\\u0F19\\u0F35\\u0F37\\u0F39\\u0F43" + + "\\u0F4D\\u0F52\\u0F57\\u0F5C\\u0F69\\u0F71-\\u0F76\\u0F78\\u0F7A" + + "-\\u0F7D\\u0F80-\\u0F84\\u0F86-\\u0F87\\u0F93\\u0F9D\\u0FA2" + + "\\u0FA7\\u0FAC\\u0FB9\\u0FC6\\u1026\\u1037\\u1039\\u1714\\u1734" + + "\\u17D2\\u18A9\\u1E00-\\u1E99\\u1E9B\\u1EA0-\\u1EF9\\u1F00-" + + "\\u1F15\\u1F18-\\u1F1D\\u1F20-\\u1F45\\u1F48-\\u1F4D\\u1F50-" + + "\\u1F57\\u1F59\\u1F5B\\u1F5D\\u1F5F-\\u1F7D\\u1F80-\\u1FB4" + + "\\u1FB6-\\u1FBC\\u1FBE\\u1FC1-\\u1FC4\\u1FC6-\\u1FD3\\u1FD6-" + + "\\u1FDB\\u1FDD-\\u1FEF\\u1FF2-\\u1FF4\\u1FF6-\\u1FFD\\u2000-" + + "\\u2001\\u20D0-\\u20DC\\u20E1\\u20E5-\\u20EA\\u2126\\u212A-" + + "\\u212B\\u219A-\\u219B\\u21AE\\u21CD-\\u21CF\\u2204\\u2209" + + "\\u220C\\u2224\\u2226\\u2241\\u2244\\u2247\\u2249\\u2260\\u2262" + + "\\u226D-\\u2271\\u2274-\\u2275\\u2278-\\u2279\\u2280-\\u2281" + + "\\u2284-\\u2285\\u2288-\\u2289\\u22AC-\\u22AF\\u22E0-\\u22E3" + + "\\u22EA-\\u22ED\\u2329-\\u232A\\u2ADC\\u302A-\\u302F\\u304C" + + "\\u304E\\u3050\\u3052\\u3054\\u3056\\u3058\\u305A\\u305C\\u305E" + + "\\u3060\\u3062\\u3065\\u3067\\u3069\\u3070-\\u3071\\u3073-" + + "\\u3074\\u3076-\\u3077\\u3079-\\u307A\\u307C-\\u307D\\u3094" + + "\\u3099-\\u309A\\u309E\\u30AC\\u30AE\\u30B0\\u30B2\\u30B4\\u30B6" + + "\\u30B8\\u30BA\\u30BC\\u30BE\\u30C0\\u30C2\\u30C5\\u30C7\\u30C9" + + "\\u30D0-\\u30D1\\u30D3-\\u30D4\\u30D6-\\u30D7\\u30D9-\\u30DA" + + "\\u30DC-\\u30DD\\u30F4\\u30F7-\\u30FA\\u30FE\\uAC00-\\uD7A3" + + "\\uF900-\\uFA0D\\uFA10\\uFA12\\uFA15-\\uFA1E\\uFA20\\uFA22" + + "\\uFA25-\\uFA26\\uFA2A-\\uFA2D\\uFA30-\\uFA6A\\uFB1D-\\uFB1F" + + "\\uFB2A-\\uFB36\\uFB38-\\uFB3C\\uFB3E\\uFB40-\\uFB41\\uFB43-" + + "\\uFB44\\uFB46-\\uFB4E\\uFE20-\\uFE23\\U0001D15E-\\U0001D169" + + "\\U0001D16D-\\U0001D172\\U0001D17B-\\U0001D182\\U0001D185-" + + "\\U0001D18B\\U0001D1AA-\\U0001D1AD\\U0001D1BB-\\U0001D1C0\\U0002" + + "F800-\\U0002FA1D]", false); + /*Unicode: + "[^À-ÅÇ-ÏÑ-ÖÙ-Ýà-åç-ïñ-öù-ýÿ-??-??-??-??-??-??-??-??-ž?-??-??-??-" + + "??-??-??-??-??-??-??-????-???-??-??-??-??-????-????-????-??-??-?" + + "?-??-??-??-??-??-??-??-??-??-???-???-??-??????-??-??-??-???-????" + + "???-??-???-??-???????-???????-??-???-????-???-??-???-????-??-??-" + + "??-??-??-???????????-???-??-??-????????????????-???-??-??-??-??-" + + "??-?????-??-??-???-??-??-??-??-??-?\?-\??-???-???-??-???-???????" + + "??????-??-??-??-??-??-??-??-??-??-???-?????????????????-??-??-??" + + "-??-???-??????????????????-??-??-??-??-???-???-??-????-????-??-?" + + "?-??-??-??-???-??-??-??-??-??-??-??-??-??-??-?]"*/ + + skippables[C] = new UnicodeSet( + "[^<->A-PR-Za-pr-z\\u00A8\\u00C0-\\u00CF\\u00D1-\\u00D6\\u00D8-" + + "\\u00DD\\u00E0-\\u00EF\\u00F1-\\u00F6\\u00F8-\\u00FD\\u00FF-" + + "\\u0103\\u0106-\\u010F\\u0112-\\u0117\\u011A-\\u0121\\u0124-" + + "\\u0125\\u0128-\\u012D\\u0130\\u0139-\\u013A\\u013D-\\u013E" + + "\\u0143-\\u0144\\u0147-\\u0148\\u014C-\\u0151\\u0154-\\u0155" + + "\\u0158-\\u015D\\u0160-\\u0161\\u0164-\\u0165\\u0168-\\u0171" + + "\\u0174-\\u017F\\u01A0-\\u01A1\\u01AF-\\u01B0\\u01B7\\u01CD-" + + "\\u01DC\\u01DE-\\u01E1\\u01E6-\\u01EB\\u01F4-\\u01F5\\u01F8-" + + "\\u01FB\\u0200-\\u021B\\u021E-\\u021F\\u0226-\\u0233\\u0292" + + "\\u0300-\\u034E\\u0360-\\u036F\\u0374\\u037E\\u0387\\u0391" + + "\\u0395\\u0397\\u0399\\u039F\\u03A1\\u03A5\\u03A9\\u03AC\\u03AE" + + "\\u03B1\\u03B5\\u03B7\\u03B9\\u03BF\\u03C1\\u03C5\\u03C9-\\u03CB" + + "\\u03CE\\u03D2\\u0406\\u0410\\u0413\\u0415-\\u0418\\u041A\\u041E" + + "\\u0423\\u0427\\u042B\\u042D\\u0430\\u0433\\u0435-\\u0438\\u043A" + + "\\u043E\\u0443\\u0447\\u044B\\u044D\\u0456\\u0474-\\u0475\\u0483" + + "-\\u0486\\u04D8-\\u04D9\\u04E8-\\u04E9\\u0591-\\u05A1\\u05A3-" + + "\\u05B9\\u05BB-\\u05BD\\u05BF\\u05C1-\\u05C2\\u05C4\\u0622-" + + "\\u0623\\u0627\\u0648\\u064A-\\u0655\\u0670\\u06C1\\u06D2\\u06D5" + + "-\\u06DC\\u06DF-\\u06E4\\u06E7-\\u06E8\\u06EA-\\u06ED\\u0711" + + "\\u0730-\\u074A\\u0928\\u0930\\u0933\\u093C\\u094D\\u0951-" + + "\\u0954\\u0958-\\u095F\\u09BC\\u09BE\\u09C7\\u09CD\\u09D7\\u09DC" + + "-\\u09DD\\u09DF\\u0A33\\u0A36\\u0A3C\\u0A4D\\u0A59-\\u0A5B" + + "\\u0A5E\\u0ABC\\u0ACD\\u0B3C\\u0B3E\\u0B47\\u0B4D\\u0B56-\\u0B57" + + "\\u0B5C-\\u0B5D\\u0B92\\u0BBE\\u0BC6-\\u0BC7\\u0BCD\\u0BD7" + + "\\u0C46\\u0C4D\\u0C55-\\u0C56\\u0CBF\\u0CC2\\u0CC6\\u0CCA\\u0CCD" + + "\\u0CD5-\\u0CD6\\u0D3E\\u0D46-\\u0D47\\u0D4D\\u0D57\\u0DCA" + + "\\u0DCF\\u0DD9\\u0DDC\\u0DDF\\u0E38-\\u0E3A\\u0E48-\\u0E4B" + + "\\u0EB8-\\u0EB9\\u0EC8-\\u0ECB\\u0F18-\\u0F19\\u0F35\\u0F37" + + "\\u0F39\\u0F43\\u0F4D\\u0F52\\u0F57\\u0F5C\\u0F69\\u0F71-\\u0F76" + + "\\u0F78\\u0F7A-\\u0F7D\\u0F80-\\u0F84\\u0F86-\\u0F87\\u0F93" + + "\\u0F9D\\u0FA2\\u0FA7\\u0FAC\\u0FB9\\u0FC6\\u1025\\u102E\\u1037" + + "\\u1039\\u1100-\\u1112\\u1161-\\u1175\\u11A8-\\u11C2\\u1714" + + "\\u1734\\u17D2\\u18A9\\u1E00-\\u1E03\\u1E0A-\\u1E0F\\u1E12-" + + "\\u1E1B\\u1E20-\\u1E27\\u1E2A-\\u1E41\\u1E44-\\u1E53\\u1E58-" + + "\\u1E7D\\u1E80-\\u1E87\\u1E8E-\\u1E91\\u1E96-\\u1E99\\u1EA0-" + + "\\u1EF3\\u1EF6-\\u1EF9\\u1F00-\\u1F11\\u1F18-\\u1F19\\u1F20-" + + "\\u1F31\\u1F38-\\u1F39\\u1F40-\\u1F41\\u1F48-\\u1F49\\u1F50-" + + "\\u1F51\\u1F59\\u1F60-\\u1F71\\u1F73-\\u1F75\\u1F77\\u1F79" + + "\\u1F7B-\\u1F7D\\u1F80-\\u1F81\\u1F88-\\u1F89\\u1F90-\\u1F91" + + "\\u1F98-\\u1F99\\u1FA0-\\u1FA1\\u1FA8-\\u1FA9\\u1FB3\\u1FB6" + + "\\u1FBB-\\u1FBC\\u1FBE-\\u1FBF\\u1FC3\\u1FC6\\u1FC9\\u1FCB-" + + "\\u1FCC\\u1FD3\\u1FDB\\u1FE3\\u1FEB\\u1FEE-\\u1FEF\\u1FF3\\u1FF6" + + "\\u1FF9\\u1FFB-\\u1FFE\\u2000-\\u2001\\u20D0-\\u20DC\\u20E1" + + "\\u20E5-\\u20EA\\u2126\\u212A-\\u212B\\u2190\\u2192\\u2194" + + "\\u21D0\\u21D2\\u21D4\\u2203\\u2208\\u220B\\u2223\\u2225\\u223C" + + "\\u2243\\u2245\\u2248\\u224D\\u2261\\u2264-\\u2265\\u2272-" + + "\\u2273\\u2276-\\u2277\\u227A-\\u227D\\u2282-\\u2283\\u2286-" + + "\\u2287\\u2291-\\u2292\\u22A2\\u22A8-\\u22A9\\u22AB\\u22B2-" + + "\\u22B5\\u2329-\\u232A\\u2ADC\\u302A-\\u302F\\u3046\\u304B" + + "\\u304D\\u304F\\u3051\\u3053\\u3055\\u3057\\u3059\\u305B\\u305D" + + "\\u305F\\u3061\\u3064\\u3066\\u3068\\u306F\\u3072\\u3075\\u3078" + + "\\u307B\\u3099-\\u309A\\u309D\\u30A6\\u30AB\\u30AD\\u30AF\\u30B1" + + "\\u30B3\\u30B5\\u30B7\\u30B9\\u30BB\\u30BD\\u30BF\\u30C1\\u30C4" + + "\\u30C6\\u30C8\\u30CF\\u30D2\\u30D5\\u30D8\\u30DB\\u30EF-\\u30F2" + + "\\u30FD\\uAC00\\uAC1C\\uAC38\\uAC54\\uAC70\\uAC8C\\uACA8\\uACC4" + + "\\uACE0\\uACFC\\uAD18\\uAD34\\uAD50\\uAD6C\\uAD88\\uADA4\\uADC0" + + "\\uADDC\\uADF8\\uAE14\\uAE30\\uAE4C\\uAE68\\uAE84\\uAEA0\\uAEBC" + + "\\uAED8\\uAEF4\\uAF10\\uAF2C\\uAF48\\uAF64\\uAF80\\uAF9C\\uAFB8" + + "\\uAFD4\\uAFF0\\uB00C\\uB028\\uB044\\uB060\\uB07C\\uB098\\uB0B4" + + "\\uB0D0\\uB0EC\\uB108\\uB124\\uB140\\uB15C\\uB178\\uB194\\uB1B0" + + "\\uB1CC\\uB1E8\\uB204\\uB220\\uB23C\\uB258\\uB274\\uB290\\uB2AC" + + "\\uB2C8\\uB2E4\\uB300\\uB31C\\uB338\\uB354\\uB370\\uB38C\\uB3A8" + + "\\uB3C4\\uB3E0\\uB3FC\\uB418\\uB434\\uB450\\uB46C\\uB488\\uB4A4" + + "\\uB4C0\\uB4DC\\uB4F8\\uB514\\uB530\\uB54C\\uB568\\uB584\\uB5A0" + + "\\uB5BC\\uB5D8\\uB5F4\\uB610\\uB62C\\uB648\\uB664\\uB680\\uB69C" + + "\\uB6B8\\uB6D4\\uB6F0\\uB70C\\uB728\\uB744\\uB760\\uB77C\\uB798" + + "\\uB7B4\\uB7D0\\uB7EC\\uB808\\uB824\\uB840\\uB85C\\uB878\\uB894" + + "\\uB8B0\\uB8CC\\uB8E8\\uB904\\uB920\\uB93C\\uB958\\uB974\\uB990" + + "\\uB9AC\\uB9C8\\uB9E4\\uBA00\\uBA1C\\uBA38\\uBA54\\uBA70\\uBA8C" + + "\\uBAA8\\uBAC4\\uBAE0\\uBAFC\\uBB18\\uBB34\\uBB50\\uBB6C\\uBB88" + + "\\uBBA4\\uBBC0\\uBBDC\\uBBF8\\uBC14\\uBC30\\uBC4C\\uBC68\\uBC84" + + "\\uBCA0\\uBCBC\\uBCD8\\uBCF4\\uBD10\\uBD2C\\uBD48\\uBD64\\uBD80" + + "\\uBD9C\\uBDB8\\uBDD4\\uBDF0\\uBE0C\\uBE28\\uBE44\\uBE60\\uBE7C" + + "\\uBE98\\uBEB4\\uBED0\\uBEEC\\uBF08\\uBF24\\uBF40\\uBF5C\\uBF78" + + "\\uBF94\\uBFB0\\uBFCC\\uBFE8\\uC004\\uC020\\uC03C\\uC058\\uC074" + + "\\uC090\\uC0AC\\uC0C8\\uC0E4\\uC100\\uC11C\\uC138\\uC154\\uC170" + + "\\uC18C\\uC1A8\\uC1C4\\uC1E0\\uC1FC\\uC218\\uC234\\uC250\\uC26C" + + "\\uC288\\uC2A4\\uC2C0\\uC2DC\\uC2F8\\uC314\\uC330\\uC34C\\uC368" + + "\\uC384\\uC3A0\\uC3BC\\uC3D8\\uC3F4\\uC410\\uC42C\\uC448\\uC464" + + "\\uC480\\uC49C\\uC4B8\\uC4D4\\uC4F0\\uC50C\\uC528\\uC544\\uC560" + + "\\uC57C\\uC598\\uC5B4\\uC5D0\\uC5EC\\uC608\\uC624\\uC640\\uC65C" + + "\\uC678\\uC694\\uC6B0\\uC6CC\\uC6E8\\uC704\\uC720\\uC73C\\uC758" + + "\\uC774\\uC790\\uC7AC\\uC7C8\\uC7E4\\uC800\\uC81C\\uC838\\uC854" + + "\\uC870\\uC88C\\uC8A8\\uC8C4\\uC8E0\\uC8FC\\uC918\\uC934\\uC950" + + "\\uC96C\\uC988\\uC9A4\\uC9C0\\uC9DC\\uC9F8\\uCA14\\uCA30\\uCA4C" + + "\\uCA68\\uCA84\\uCAA0\\uCABC\\uCAD8\\uCAF4\\uCB10\\uCB2C\\uCB48" + + "\\uCB64\\uCB80\\uCB9C\\uCBB8\\uCBD4\\uCBF0\\uCC0C\\uCC28\\uCC44" + + "\\uCC60\\uCC7C\\uCC98\\uCCB4\\uCCD0\\uCCEC\\uCD08\\uCD24\\uCD40" + + "\\uCD5C\\uCD78\\uCD94\\uCDB0\\uCDCC\\uCDE8\\uCE04\\uCE20\\uCE3C" + + "\\uCE58\\uCE74\\uCE90\\uCEAC\\uCEC8\\uCEE4\\uCF00\\uCF1C\\uCF38" + + "\\uCF54\\uCF70\\uCF8C\\uCFA8\\uCFC4\\uCFE0\\uCFFC\\uD018\\uD034" + + "\\uD050\\uD06C\\uD088\\uD0A4\\uD0C0\\uD0DC\\uD0F8\\uD114\\uD130" + + "\\uD14C\\uD168\\uD184\\uD1A0\\uD1BC\\uD1D8\\uD1F4\\uD210\\uD22C" + + "\\uD248\\uD264\\uD280\\uD29C\\uD2B8\\uD2D4\\uD2F0\\uD30C\\uD328" + + "\\uD344\\uD360\\uD37C\\uD398\\uD3B4\\uD3D0\\uD3EC\\uD408\\uD424" + + "\\uD440\\uD45C\\uD478\\uD494\\uD4B0\\uD4CC\\uD4E8\\uD504\\uD520" + + "\\uD53C\\uD558\\uD574\\uD590\\uD5AC\\uD5C8\\uD5E4\\uD600\\uD61C" + + "\\uD638\\uD654\\uD670\\uD68C\\uD6A8\\uD6C4\\uD6E0\\uD6FC\\uD718" + + "\\uD734\\uD750\\uD76C\\uD788\\uF900-\\uFA0D\\uFA10\\uFA12\\uFA15" + + "-\\uFA1E\\uFA20\\uFA22\\uFA25-\\uFA26\\uFA2A-\\uFA2D\\uFA30-" + + "\\uFA6A\\uFB1D-\\uFB1F\\uFB2A-\\uFB36\\uFB38-\\uFB3C\\uFB3E" + + "\\uFB40-\\uFB41\\uFB43-\\uFB44\\uFB46-\\uFB4E\\uFE20-\\uFE23" + + "\\U0001D15E-\\U0001D169\\U0001D16D-\\U0001D172\\U0001D17B-" + + "\\U0001D182\\U0001D185-\\U0001D18B\\U0001D1AA-\\U0001D1AD\\U0001" + + "D1BB-\\U0001D1C0\\U0002F800-\\U0002FA1D]", false); + /*Unicode: + "[^<->A-PR-Za-pr-z¨À-ÏÑ-ÖØ-Ýà-ïñ-öø-ýÿ-??-??-??-??-??-???-??-??-?" + + "?-??-??-??-?Š-š?-??-??-??-??-???-??-??-??-??-??-??-??-???-??-???" + + "???????????????????-???????-??????????-?????????-??-??-??-??-??-" + + "??-???-???-????-?????-??-??-??-???-???????-??-???????-???????-??" + + "???????-??-????-??????-???????-???-?????????-??-??-??-??-???????" + + "????-???-??-??-?????????????-??-??-??????-??-??-??-??-??-??-??-?" + + "?-??-??-??-??-??-??-??-??-??-??-???-??-????-??-??-??-??-??-??-??" + + "??-??-?????-??????-?????-?\?-\??-???-???-???????????????????-??-" + + "??-??-??-??-??-???-???-??-???-???????????????????????-??????????" + + "??????????????-?????????????????????????????????????????????????" + + "????????????????????????????????????????????????????????????????" + + "????????????????????????????????????????????????????????????????" + + "????????????????????????????????????????????????????????????????" + + "????????????????????????????????????????????????????????????????" + + "????????????????????????????????????????????????????????????????" + + "?????????????????????????????????-????-????-??-??-??-??-??-???-?" + + "?-??-??-??-??-??-??-??-??-??-?]"*/ + skippables[KD] = new UnicodeSet( + "[^\\u00A0\\u00A8\\u00AA\\u00AF\\u00B2-\\u00B5\\u00B8-\\u00BA" + + "\\u00BC-\\u00BE\\u00C0-\\u00C5\\u00C7-\\u00CF\\u00D1-\\u00D6" + + "\\u00D9-\\u00DD\\u00E0-\\u00E5\\u00E7-\\u00EF\\u00F1-\\u00F6" + + "\\u00F9-\\u00FD\\u00FF-\\u010F\\u0112-\\u0125\\u0128-\\u0130" + + "\\u0132-\\u0137\\u0139-\\u0140\\u0143-\\u0149\\u014C-\\u0151" + + "\\u0154-\\u0165\\u0168-\\u017F\\u01A0-\\u01A1\\u01AF-\\u01B0" + + "\\u01C4-\\u01DC\\u01DE-\\u01E3\\u01E6-\\u01F5\\u01F8-\\u021B" + + "\\u021E-\\u021F\\u0226-\\u0233\\u02B0-\\u02B8\\u02D8-\\u02DD" + + "\\u02E0-\\u02E4\\u0300-\\u034E\\u0360-\\u036F\\u0374\\u037A" + + "\\u037E\\u0384-\\u038A\\u038C\\u038E-\\u0390\\u03AA-\\u03B0" + + "\\u03CA-\\u03CE\\u03D0-\\u03D6\\u03F0-\\u03F2\\u03F4-\\u03F5" + + "\\u0400-\\u0401\\u0403\\u0407\\u040C-\\u040E\\u0419\\u0439" + + "\\u0450-\\u0451\\u0453\\u0457\\u045C-\\u045E\\u0476-\\u0477" + + "\\u0483-\\u0486\\u04C1-\\u04C2\\u04D0-\\u04D3\\u04D6-\\u04D7" + + "\\u04DA-\\u04DF\\u04E2-\\u04E7\\u04EA-\\u04F5\\u04F8-\\u04F9" + + "\\u0587\\u0591-\\u05A1\\u05A3-\\u05B9\\u05BB-\\u05BD\\u05BF" + + "\\u05C1-\\u05C2\\u05C4\\u0622-\\u0626\\u064B-\\u0655\\u0670" + + "\\u0675-\\u0678\\u06C0\\u06C2\\u06D3\\u06D6-\\u06DC\\u06DF-" + + "\\u06E4\\u06E7-\\u06E8\\u06EA-\\u06ED\\u0711\\u0730-\\u074A" + + "\\u0929\\u0931\\u0934\\u093C\\u094D\\u0951-\\u0954\\u0958-" + + "\\u095F\\u09BC\\u09CB-\\u09CD\\u09DC-\\u09DD\\u09DF\\u0A33" + + "\\u0A36\\u0A3C\\u0A4D\\u0A59-\\u0A5B\\u0A5E\\u0ABC\\u0ACD\\u0B3C" + + "\\u0B48\\u0B4B-\\u0B4D\\u0B5C-\\u0B5D\\u0B94\\u0BCA-\\u0BCD" + + "\\u0C48\\u0C4D\\u0C55-\\u0C56\\u0CC0\\u0CC7-\\u0CC8\\u0CCA-" + + "\\u0CCB\\u0CCD\\u0D4A-\\u0D4D\\u0DCA\\u0DDA\\u0DDC-\\u0DDE" + + "\\u0E33\\u0E38-\\u0E3A\\u0E48-\\u0E4B\\u0EB3\\u0EB8-\\u0EB9" + + "\\u0EC8-\\u0ECB\\u0EDC-\\u0EDD\\u0F0C\\u0F18-\\u0F19\\u0F35" + + "\\u0F37\\u0F39\\u0F43\\u0F4D\\u0F52\\u0F57\\u0F5C\\u0F69\\u0F71-" + + "\\u0F7D\\u0F80-\\u0F84\\u0F86-\\u0F87\\u0F93\\u0F9D\\u0FA2" + + "\\u0FA7\\u0FAC\\u0FB9\\u0FC6\\u1026\\u1037\\u1039\\u1714\\u1734" + + "\\u17D2\\u18A9\\u1E00-\\u1E9B\\u1EA0-\\u1EF9\\u1F00-\\u1F15" + + "\\u1F18-\\u1F1D\\u1F20-\\u1F45\\u1F48-\\u1F4D\\u1F50-\\u1F57" + + "\\u1F59\\u1F5B\\u1F5D\\u1F5F-\\u1F7D\\u1F80-\\u1FB4\\u1FB6-" + + "\\u1FC4\\u1FC6-\\u1FD3\\u1FD6-\\u1FDB\\u1FDD-\\u1FEF\\u1FF2-" + + "\\u1FF4\\u1FF6-\\u1FFE\\u2000-\\u200A\\u2011\\u2017\\u2024-" + + "\\u2026\\u202F\\u2033-\\u2034\\u2036-\\u2037\\u203C\\u203E" + + "\\u2047-\\u2049\\u2057\\u205F\\u2070-\\u2071\\u2074-\\u208E" + + "\\u20A8\\u20D0-\\u20DC\\u20E1\\u20E5-\\u20EA\\u2100-\\u2103" + + "\\u2105-\\u2107\\u2109-\\u2113\\u2115-\\u2116\\u2119-\\u211D" + + "\\u2120-\\u2122\\u2124\\u2126\\u2128\\u212A-\\u212D\\u212F-" + + "\\u2131\\u2133-\\u2139\\u213D-\\u2140\\u2145-\\u2149\\u2153-" + + "\\u217F\\u219A-\\u219B\\u21AE\\u21CD-\\u21CF\\u2204\\u2209" + + "\\u220C\\u2224\\u2226\\u222C-\\u222D\\u222F-\\u2230\\u2241" + + "\\u2244\\u2247\\u2249\\u2260\\u2262\\u226D-\\u2271\\u2274-" + + "\\u2275\\u2278-\\u2279\\u2280-\\u2281\\u2284-\\u2285\\u2288-" + + "\\u2289\\u22AC-\\u22AF\\u22E0-\\u22E3\\u22EA-\\u22ED\\u2329-" + + "\\u232A\\u2460-\\u24EA\\u2A0C\\u2A74-\\u2A76\\u2ADC\\u2E9F" + + "\\u2EF3\\u2F00-\\u2FD5\\u3000\\u302A-\\u302F\\u3036\\u3038-" + + "\\u303A\\u304C\\u304E\\u3050\\u3052\\u3054\\u3056\\u3058\\u305A" + + "\\u305C\\u305E\\u3060\\u3062\\u3065\\u3067\\u3069\\u3070-\\u3071" + + "\\u3073-\\u3074\\u3076-\\u3077\\u3079-\\u307A\\u307C-\\u307D" + + "\\u3094\\u3099-\\u309C\\u309E-\\u309F\\u30AC\\u30AE\\u30B0" + + "\\u30B2\\u30B4\\u30B6\\u30B8\\u30BA\\u30BC\\u30BE\\u30C0\\u30C2" + + "\\u30C5\\u30C7\\u30C9\\u30D0-\\u30D1\\u30D3-\\u30D4\\u30D6-" + + "\\u30D7\\u30D9-\\u30DA\\u30DC-\\u30DD\\u30F4\\u30F7-\\u30FA" + + "\\u30FE-\\u30FF\\u3131-\\u318E\\u3192-\\u319F\\u3200-\\u321C" + + "\\u3220-\\u3243\\u3251-\\u327B\\u3280-\\u32CB\\u32D0-\\u32FE" + + "\\u3300-\\u3376\\u337B-\\u33DD\\u33E0-\\u33FE\\uAC00-\\uD7A3" + + "\\uF900-\\uFA0D\\uFA10\\uFA12\\uFA15-\\uFA1E\\uFA20\\uFA22" + + "\\uFA25-\\uFA26\\uFA2A-\\uFA2D\\uFA30-\\uFA6A\\uFB00-\\uFB06" + + "\\uFB13-\\uFB17\\uFB1D-\\uFB36\\uFB38-\\uFB3C\\uFB3E\\uFB40-" + + "\\uFB41\\uFB43-\\uFB44\\uFB46-\\uFBB1\\uFBD3-\\uFD3D\\uFD50-" + + "\\uFD8F\\uFD92-\\uFDC7\\uFDF0-\\uFDFC\\uFE20-\\uFE23\\uFE30-" + + "\\uFE44\\uFE49-\\uFE52\\uFE54-\\uFE66\\uFE68-\\uFE6B\\uFE70-" + + "\\uFE72\\uFE74\\uFE76-\\uFEFC\\uFF01-\\uFFBE\\uFFC2-\\uFFC7" + + "\\uFFCA-\\uFFCF\\uFFD2-\\uFFD7\\uFFDA-\\uFFDC\\uFFE0-\\uFFE6" + + "\\uFFE8-\\uFFEE\\U0001D15E-\\U0001D169\\U0001D16D-\\U0001D172" + + "\\U0001D17B-\\U0001D182\\U0001D185-\\U0001D18B\\U0001D1AA-" + + "\\U0001D1AD\\U0001D1BB-\\U0001D1C0\\U0001D400-\\U0001D454\\U0001" + + "D456-\\U0001D49C\\U0001D49E-\\U0001D49F\\U0001D4A2\\U0001D4A5-" + + "\\U0001D4A6\\U0001D4A9-\\U0001D4AC\\U0001D4AE-\\U0001D4B9\\U0001" + + "D4BB\\U0001D4BD-\\U0001D4C0\\U0001D4C2-\\U0001D4C3\\U0001D4C5-" + + "\\U0001D505\\U0001D507-\\U0001D50A\\U0001D50D-\\U0001D514\\U0001" + + "D516-\\U0001D51C\\U0001D51E-\\U0001D539\\U0001D53B-\\U0001D53E" + + "\\U0001D540-\\U0001D544\\U0001D546\\U0001D54A-\\U0001D550\\U0001" + + "D552-\\U0001D6A3\\U0001D6A8-\\U0001D7C9\\U0001D7CE-\\U0001D7FF" + + "\\U0002F800-\\U0002FA1D]", false); + /*Unicode: + "[^ ¨ª¯²-µ¸-º¼-¾À-ÅÇ-ÏÑ-ÖÙ-Ýà-åç-ïñ-öù-ýÿ-??-??-??-??-??-??-??-??" + + "-??-??-??-??-??-??-??-??-??-??-??-??-??-?????-???-??-??-??-??-??" + + "-??-????-????-????-??-??-??-??-??-??-??-??-??-???-??-??-???-???-" + + "??-???-?????-??-??-??-???-???????-??-???-??-???????-???????-??-?" + + "??-????-???-??-???-????-???-??-???-??-??-???-???????????-??-??-?" + + "???????????????-??-??-??-??-??-??-?????-??-??-??-??-??-??-??-?" + + "\?-\????-…??-??-????-??\??-??-???-???-??-??-??-??-??-??-™????-??" + + "-??-??-??-??-??-???-???????-??-????????-??-??-??-??-??-??-??-??-" + + "??-??-???-?????-?\??-???-?????????????????-??-??-??-??-???-??-??" + + "???????????????-??-??-??-??-???-??-??-??-??-??-??-??-??-??-??-??" + + "-??-??-????-????-??-??-??-??-??-??-???-??-??-??-??-??-??-??-??-?" + + "?-??-??-??-???-??-??-??-??-??-??-??-??-??-??-??-??-??" + + "-??-??-??-???-??-??-???-??-??-??-??-??" + + "-??-??-??-???-??-??-??-??-?]"*/ + + skippables[KC] = new UnicodeSet( + "[^<->A-PR-Za-pr-z\\u00A0\\u00A8\\u00AA\\u00AF\\u00B2-\\u00B5" + + "\\u00B8-\\u00BA\\u00BC-\\u00BE\\u00C0-\\u00CF\\u00D1-\\u00D6" + + "\\u00D8-\\u00DD\\u00E0-\\u00EF\\u00F1-\\u00F6\\u00F8-\\u00FD" + + "\\u00FF-\\u0103\\u0106-\\u010F\\u0112-\\u0117\\u011A-\\u0121" + + "\\u0124-\\u0125\\u0128-\\u012D\\u0130\\u0132-\\u0133\\u0139-" + + "\\u013A\\u013D-\\u0140\\u0143-\\u0144\\u0147-\\u0149\\u014C-" + + "\\u0151\\u0154-\\u0155\\u0158-\\u015D\\u0160-\\u0161\\u0164-" + + "\\u0165\\u0168-\\u0171\\u0174-\\u017F\\u01A0-\\u01A1\\u01AF-" + + "\\u01B0\\u01B7\\u01C4-\\u01DC\\u01DE-\\u01E1\\u01E6-\\u01EB" + + "\\u01F1-\\u01F5\\u01F8-\\u01FB\\u0200-\\u021B\\u021E-\\u021F" + + "\\u0226-\\u0233\\u0292\\u02B0-\\u02B8\\u02D8-\\u02DD\\u02E0-" + + "\\u02E4\\u0300-\\u034E\\u0360-\\u036F\\u0374\\u037A\\u037E" + + "\\u0384-\\u0385\\u0387\\u0391\\u0395\\u0397\\u0399\\u039F\\u03A1" + + "\\u03A5\\u03A9\\u03AC\\u03AE\\u03B1\\u03B5\\u03B7\\u03B9\\u03BF" + + "\\u03C1\\u03C5\\u03C9-\\u03CB\\u03CE\\u03D0-\\u03D6\\u03F0-" + + "\\u03F2\\u03F4-\\u03F5\\u0406\\u0410\\u0413\\u0415-\\u0418" + + "\\u041A\\u041E\\u0423\\u0427\\u042B\\u042D\\u0430\\u0433\\u0435-" + + "\\u0438\\u043A\\u043E\\u0443\\u0447\\u044B\\u044D\\u0456\\u0474-" + + "\\u0475\\u0483-\\u0486\\u04D8-\\u04D9\\u04E8-\\u04E9\\u0587" + + "\\u0591-\\u05A1\\u05A3-\\u05B9\\u05BB-\\u05BD\\u05BF\\u05C1-" + + "\\u05C2\\u05C4\\u0622-\\u0623\\u0627\\u0648\\u064A-\\u0655" + + "\\u0670\\u0675-\\u0678\\u06C1\\u06D2\\u06D5-\\u06DC\\u06DF-" + + "\\u06E4\\u06E7-\\u06E8\\u06EA-\\u06ED\\u0711\\u0730-\\u074A" + + "\\u0928\\u0930\\u0933\\u093C\\u094D\\u0951-\\u0954\\u0958-" + + "\\u095F\\u09BC\\u09BE\\u09C7\\u09CD\\u09D7\\u09DC-\\u09DD\\u09DF" + + "\\u0A33\\u0A36\\u0A3C\\u0A4D\\u0A59-\\u0A5B\\u0A5E\\u0ABC\\u0ACD" + + "\\u0B3C\\u0B3E\\u0B47\\u0B4D\\u0B56-\\u0B57\\u0B5C-\\u0B5D" + + "\\u0B92\\u0BBE\\u0BC6-\\u0BC7\\u0BCD\\u0BD7\\u0C46\\u0C4D\\u0C55" + + "-\\u0C56\\u0CBF\\u0CC2\\u0CC6\\u0CCA\\u0CCD\\u0CD5-\\u0CD6" + + "\\u0D3E\\u0D46-\\u0D47\\u0D4D\\u0D57\\u0DCA\\u0DCF\\u0DD9\\u0DDC" + + "\\u0DDF\\u0E33\\u0E38-\\u0E3A\\u0E48-\\u0E4B\\u0EB3\\u0EB8-" + + "\\u0EB9\\u0EC8-\\u0ECB\\u0EDC-\\u0EDD\\u0F0C\\u0F18-\\u0F19" + + "\\u0F35\\u0F37\\u0F39\\u0F43\\u0F4D\\u0F52\\u0F57\\u0F5C\\u0F69" + + "\\u0F71-\\u0F7D\\u0F80-\\u0F84\\u0F86-\\u0F87\\u0F93\\u0F9D" + + "\\u0FA2\\u0FA7\\u0FAC\\u0FB9\\u0FC6\\u1025\\u102E\\u1037\\u1039" + + "\\u1100-\\u1112\\u1161-\\u1175\\u11A8-\\u11C2\\u1714\\u1734" + + "\\u17D2\\u18A9\\u1E00-\\u1E03\\u1E0A-\\u1E0F\\u1E12-\\u1E1B" + + "\\u1E20-\\u1E27\\u1E2A-\\u1E41\\u1E44-\\u1E53\\u1E58-\\u1E7D" + + "\\u1E80-\\u1E87\\u1E8E-\\u1E91\\u1E96-\\u1E9B\\u1EA0-\\u1EF3" + + "\\u1EF6-\\u1EF9\\u1F00-\\u1F11\\u1F18-\\u1F19\\u1F20-\\u1F31" + + "\\u1F38-\\u1F39\\u1F40-\\u1F41\\u1F48-\\u1F49\\u1F50-\\u1F51" + + "\\u1F59\\u1F60-\\u1F71\\u1F73-\\u1F75\\u1F77\\u1F79\\u1F7B-" + + "\\u1F7D\\u1F80-\\u1F81\\u1F88-\\u1F89\\u1F90-\\u1F91\\u1F98-" + + "\\u1F99\\u1FA0-\\u1FA1\\u1FA8-\\u1FA9\\u1FB3\\u1FB6\\u1FBB-" + + "\\u1FC1\\u1FC3\\u1FC6\\u1FC9\\u1FCB-\\u1FCF\\u1FD3\\u1FDB\\u1FDD" + + "-\\u1FDF\\u1FE3\\u1FEB\\u1FED-\\u1FEF\\u1FF3\\u1FF6\\u1FF9" + + "\\u1FFB-\\u1FFE\\u2000-\\u200A\\u2011\\u2017\\u2024-\\u2026" + + "\\u202F\\u2033-\\u2034\\u2036-\\u2037\\u203C\\u203E\\u2047-" + + "\\u2049\\u2057\\u205F\\u2070-\\u2071\\u2074-\\u208E\\u20A8" + + "\\u20D0-\\u20DC\\u20E1\\u20E5-\\u20EA\\u2100-\\u2103\\u2105-" + + "\\u2107\\u2109-\\u2113\\u2115-\\u2116\\u2119-\\u211D\\u2120-" + + "\\u2122\\u2124\\u2126\\u2128\\u212A-\\u212D\\u212F-\\u2131" + + "\\u2133-\\u2139\\u213D-\\u2140\\u2145-\\u2149\\u2153-\\u217F" + + "\\u2190\\u2192\\u2194\\u21D0\\u21D2\\u21D4\\u2203\\u2208\\u220B" + + "\\u2223\\u2225\\u222C-\\u222D\\u222F-\\u2230\\u223C\\u2243" + + "\\u2245\\u2248\\u224D\\u2261\\u2264-\\u2265\\u2272-\\u2273" + + "\\u2276-\\u2277\\u227A-\\u227D\\u2282-\\u2283\\u2286-\\u2287" + + "\\u2291-\\u2292\\u22A2\\u22A8-\\u22A9\\u22AB\\u22B2-\\u22B5" + + "\\u2329-\\u232A\\u2460-\\u24EA\\u2A0C\\u2A74-\\u2A76\\u2ADC" + + "\\u2E9F\\u2EF3\\u2F00-\\u2FD5\\u3000\\u302A-\\u302F\\u3036" + + "\\u3038-\\u303A\\u3046\\u304B\\u304D\\u304F\\u3051\\u3053\\u3055" + + "\\u3057\\u3059\\u305B\\u305D\\u305F\\u3061\\u3064\\u3066\\u3068" + + "\\u306F\\u3072\\u3075\\u3078\\u307B\\u3099-\\u309D\\u309F\\u30A6" + + "\\u30AB\\u30AD\\u30AF\\u30B1\\u30B3\\u30B5\\u30B7\\u30B9\\u30BB" + + "\\u30BD\\u30BF\\u30C1\\u30C4\\u30C6\\u30C8\\u30CF\\u30D2\\u30D5" + + "\\u30D8\\u30DB\\u30EF-\\u30F2\\u30FD\\u30FF\\u3131-\\u318E" + + "\\u3192-\\u319F\\u3200-\\u321C\\u3220-\\u3243\\u3251-\\u327B" + + "\\u3280-\\u32CB\\u32D0-\\u32FE\\u3300-\\u3376\\u337B-\\u33DD" + + "\\u33E0-\\u33FE\\uAC00\\uAC1C\\uAC38\\uAC54\\uAC70\\uAC8C\\uACA8" + + "\\uACC4\\uACE0\\uACFC\\uAD18\\uAD34\\uAD50\\uAD6C\\uAD88\\uADA4" + + "\\uADC0\\uADDC\\uADF8\\uAE14\\uAE30\\uAE4C\\uAE68\\uAE84\\uAEA0" + + "\\uAEBC\\uAED8\\uAEF4\\uAF10\\uAF2C\\uAF48\\uAF64\\uAF80\\uAF9C" + + "\\uAFB8\\uAFD4\\uAFF0\\uB00C\\uB028\\uB044\\uB060\\uB07C\\uB098" + + "\\uB0B4\\uB0D0\\uB0EC\\uB108\\uB124\\uB140\\uB15C\\uB178\\uB194" + + "\\uB1B0\\uB1CC\\uB1E8\\uB204\\uB220\\uB23C\\uB258\\uB274\\uB290" + + "\\uB2AC\\uB2C8\\uB2E4\\uB300\\uB31C\\uB338\\uB354\\uB370\\uB38C" + + "\\uB3A8\\uB3C4\\uB3E0\\uB3FC\\uB418\\uB434\\uB450\\uB46C\\uB488" + + "\\uB4A4\\uB4C0\\uB4DC\\uB4F8\\uB514\\uB530\\uB54C\\uB568\\uB584" + + "\\uB5A0\\uB5BC\\uB5D8\\uB5F4\\uB610\\uB62C\\uB648\\uB664\\uB680" + + "\\uB69C\\uB6B8\\uB6D4\\uB6F0\\uB70C\\uB728\\uB744\\uB760\\uB77C" + + "\\uB798\\uB7B4\\uB7D0\\uB7EC\\uB808\\uB824\\uB840\\uB85C\\uB878" + + "\\uB894\\uB8B0\\uB8CC\\uB8E8\\uB904\\uB920\\uB93C\\uB958\\uB974" + + "\\uB990\\uB9AC\\uB9C8\\uB9E4\\uBA00\\uBA1C\\uBA38\\uBA54\\uBA70" + + "\\uBA8C\\uBAA8\\uBAC4\\uBAE0\\uBAFC\\uBB18\\uBB34\\uBB50\\uBB6C" + + "\\uBB88\\uBBA4\\uBBC0\\uBBDC\\uBBF8\\uBC14\\uBC30\\uBC4C\\uBC68" + + "\\uBC84\\uBCA0\\uBCBC\\uBCD8\\uBCF4\\uBD10\\uBD2C\\uBD48\\uBD64" + + "\\uBD80\\uBD9C\\uBDB8\\uBDD4\\uBDF0\\uBE0C\\uBE28\\uBE44\\uBE60" + + "\\uBE7C\\uBE98\\uBEB4\\uBED0\\uBEEC\\uBF08\\uBF24\\uBF40\\uBF5C" + + "\\uBF78\\uBF94\\uBFB0\\uBFCC\\uBFE8\\uC004\\uC020\\uC03C\\uC058" + + "\\uC074\\uC090\\uC0AC\\uC0C8\\uC0E4\\uC100\\uC11C\\uC138\\uC154" + + "\\uC170\\uC18C\\uC1A8\\uC1C4\\uC1E0\\uC1FC\\uC218\\uC234\\uC250" + + "\\uC26C\\uC288\\uC2A4\\uC2C0\\uC2DC\\uC2F8\\uC314\\uC330\\uC34C" + + "\\uC368\\uC384\\uC3A0\\uC3BC\\uC3D8\\uC3F4\\uC410\\uC42C\\uC448" + + "\\uC464\\uC480\\uC49C\\uC4B8\\uC4D4\\uC4F0\\uC50C\\uC528\\uC544" + + "\\uC560\\uC57C\\uC598\\uC5B4\\uC5D0\\uC5EC\\uC608\\uC624\\uC640" + + "\\uC65C\\uC678\\uC694\\uC6B0\\uC6CC\\uC6E8\\uC704\\uC720\\uC73C" + + "\\uC758\\uC774\\uC790\\uC7AC\\uC7C8\\uC7E4\\uC800\\uC81C\\uC838" + + "\\uC854\\uC870\\uC88C\\uC8A8\\uC8C4\\uC8E0\\uC8FC\\uC918\\uC934" + + "\\uC950\\uC96C\\uC988\\uC9A4\\uC9C0\\uC9DC\\uC9F8\\uCA14\\uCA30" + + "\\uCA4C\\uCA68\\uCA84\\uCAA0\\uCABC\\uCAD8\\uCAF4\\uCB10\\uCB2C" + + "\\uCB48\\uCB64\\uCB80\\uCB9C\\uCBB8\\uCBD4\\uCBF0\\uCC0C\\uCC28" + + "\\uCC44\\uCC60\\uCC7C\\uCC98\\uCCB4\\uCCD0\\uCCEC\\uCD08\\uCD24" + + "\\uCD40\\uCD5C\\uCD78\\uCD94\\uCDB0\\uCDCC\\uCDE8\\uCE04\\uCE20" + + "\\uCE3C\\uCE58\\uCE74\\uCE90\\uCEAC\\uCEC8\\uCEE4\\uCF00\\uCF1C" + + "\\uCF38\\uCF54\\uCF70\\uCF8C\\uCFA8\\uCFC4\\uCFE0\\uCFFC\\uD018" + + "\\uD034\\uD050\\uD06C\\uD088\\uD0A4\\uD0C0\\uD0DC\\uD0F8\\uD114" + + "\\uD130\\uD14C\\uD168\\uD184\\uD1A0\\uD1BC\\uD1D8\\uD1F4\\uD210" + + "\\uD22C\\uD248\\uD264\\uD280\\uD29C\\uD2B8\\uD2D4\\uD2F0\\uD30C" + + "\\uD328\\uD344\\uD360\\uD37C\\uD398\\uD3B4\\uD3D0\\uD3EC\\uD408" + + "\\uD424\\uD440\\uD45C\\uD478\\uD494\\uD4B0\\uD4CC\\uD4E8\\uD504" + + "\\uD520\\uD53C\\uD558\\uD574\\uD590\\uD5AC\\uD5C8\\uD5E4\\uD600" + + "\\uD61C\\uD638\\uD654\\uD670\\uD68C\\uD6A8\\uD6C4\\uD6E0\\uD6FC" + + "\\uD718\\uD734\\uD750\\uD76C\\uD788\\uF900-\\uFA0D\\uFA10\\uFA12" + + "\\uFA15-\\uFA1E\\uFA20\\uFA22\\uFA25-\\uFA26\\uFA2A-\\uFA2D" + + "\\uFA30-\\uFA6A\\uFB00-\\uFB06\\uFB13-\\uFB17\\uFB1D-\\uFB36" + + "\\uFB38-\\uFB3C\\uFB3E\\uFB40-\\uFB41\\uFB43-\\uFB44\\uFB46-" + + "\\uFBB1\\uFBD3-\\uFD3D\\uFD50-\\uFD8F\\uFD92-\\uFDC7\\uFDF0-" + + "\\uFDFC\\uFE20-\\uFE23\\uFE30-\\uFE44\\uFE49-\\uFE52\\uFE54-" + + "\\uFE66\\uFE68-\\uFE6B\\uFE70-\\uFE72\\uFE74\\uFE76-\\uFEFC" + + "\\uFF01-\\uFFBE\\uFFC2-\\uFFC7\\uFFCA-\\uFFCF\\uFFD2-\\uFFD7" + + "\\uFFDA-\\uFFDC\\uFFE0-\\uFFE6\\uFFE8-\\uFFEE\\U0001D15E-\\U0001" + + "D169\\U0001D16D-\\U0001D172\\U0001D17B-\\U0001D182\\U0001D185-" + + "\\U0001D18B\\U0001D1AA-\\U0001D1AD\\U0001D1BB-\\U0001D1C0\\U0001" + + "D400-\\U0001D454\\U0001D456-\\U0001D49C\\U0001D49E-\\U0001D49F" + + "\\U0001D4A2\\U0001D4A5-\\U0001D4A6\\U0001D4A9-\\U0001D4AC\\U0001" + + "D4AE-\\U0001D4B9\\U0001D4BB\\U0001D4BD-\\U0001D4C0\\U0001D4C2-" + + "\\U0001D4C3\\U0001D4C5-\\U0001D505\\U0001D507-\\U0001D50A\\U0001" + + "D50D-\\U0001D514\\U0001D516-\\U0001D51C\\U0001D51E-\\U0001D539" + + "\\U0001D53B-\\U0001D53E\\U0001D540-\\U0001D544\\U0001D546\\U0001" + + "D54A-\\U0001D550\\U0001D552-\\U0001D6A3\\U0001D6A8-\\U0001D7C9" + + "\\U0001D7CE-\\U0001D7FF\\U0002F800-\\U0002FA1D]", false); + + return skippables; + } + + public void TestSkippable() { + UnicodeSet starts, diff; + UnicodeSet[] skipSets = new UnicodeSet[]{ + new UnicodeSet(), + new UnicodeSet(), + new UnicodeSet(), + new UnicodeSet() + }; + UnicodeSet[] expectSets = new UnicodeSet[]{ + new UnicodeSet(), + new UnicodeSet(), + new UnicodeSet(), + new UnicodeSet() + }; + StringBuffer s, pattern; + int start, limit, rangeStart, rangeEnd; + int i, range, count; + starts = new UnicodeSet(); + + + /* build NF*Skippable sets from runtime data */ + NormalizerImpl.addPropertyStarts(starts); + count=starts.getRangeCount(); + + start=limit=0; + rangeStart=rangeEnd=0; + range=0; + for(;;) { + if(startrangeEnd) { + if(range=2.2) * 11 flag: not a safe starter for canonical closure * 10 composition exclusion * 9.. 0 index into extraData[] to FC_NFKC_Closure string * (not for lead surrogate), * or lead surrogate offset (for lead surrogate, if 9..0 not zero) - * + * + * Conditions for "NF* Skippable" from Mark Davis' com.ibm.text.UCD.NFSkippable: + * (used in NormalizerTransliterator) + * + * A skippable character is + * a) unassigned, or ALL of the following: + * b) of combining class 0. + * c) not decomposed by this normalization form. + * AND if NFC or NFKC, + * d) can never compose with a previous character. + * e) can never compose with a following character. + * f) can never change if another character is added. + * Example: a-breve might satisfy all but f, but if you + * add an ogonek it changes to a-ogonek + breve + * + * a)..e) must be tested from norm32. + * Since f) is more complicated, the (not-)NFC_Skippable flag (f) is built + * into the auxiliary trie. + * The same bit is used for NFC and NFKC; (c) differs for them. + * As usual, we build the "not skippable" flags so that unassigned + * code points get a 0 bit. + * This bit is only valid after (a)..(e) test FALSE; test NFD_NO before (f) as well. + * Test Hangul LV syllables entirely in code. + * + * * - FC_NFKC_Closure strings in extraData[] * * Strings are either stored as a single code unit or as the length @@ -399,7 +424,7 @@ final class NormalizerDataReader implements ICUBinary.Authenticate { */ private static final byte DATA_FORMAT_ID[] = {(byte)0x4E, (byte)0x6F, (byte)0x72, (byte)0x6D}; - private static final byte DATA_FORMAT_VERSION[] = {(byte)0x2, (byte)0x1, + private static final byte DATA_FORMAT_VERSION[] = {(byte)0x2, (byte)0x2, (byte)0x5, (byte)0x2}; } diff --git a/icu4j/src/com/ibm/icu/impl/NormalizerImpl.java b/icu4j/src/com/ibm/icu/impl/NormalizerImpl.java index 526df9d8aa..94ad143ff7 100644 --- a/icu4j/src/com/ibm/icu/impl/NormalizerImpl.java +++ b/icu4j/src/com/ibm/icu/impl/NormalizerImpl.java @@ -5,17 +5,23 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/impl/NormalizerImpl.java,v $ - * $Date: 2002/09/19 18:12:30 $ - * $Revision: 1.11 $ + * $Date: 2002/11/15 21:04:18 $ + * $Revision: 1.12 $ ******************************************************************************* */ package com.ibm.icu.impl; -import java.io.*; +import java.io.ByteArrayInputStream; +import java.io.IOException; +import java.io.BufferedInputStream; +import java.io.InputStream; import com.ibm.icu.text.Normalizer; import com.ibm.icu.text.UTF16; +import com.ibm.icu.text.UnicodeSet; +import com.ibm.icu.util.RangeValueIterator; import com.ibm.icu.lang.UCharacter; + /** * @version 1.0 * @author Ram Viswanadha @@ -53,7 +59,7 @@ public final class NormalizerImpl { public static final int QC_NFD=4; /* no */ public static final int QC_NFKD=8; /* no */ - private static final int QC_ANY_NO=0xf; + public static final int QC_ANY_NO=0xf; /* quick check flags 4..5 mean "maybe" for their forms; * test flags>=QC_MAYBE @@ -65,10 +71,10 @@ public final class NormalizerImpl { private static final int COMBINES_FWD=0x40; private static final int COMBINES_BACK=0x80; - private static final int COMBINES_ANY=0xc0; + public static final int COMBINES_ANY=0xc0; // UnicodeData.txt combining class in bits 15. private static final int CC_SHIFT=8; - public static final int CC_MASK=0xff00; + public static final int CC_MASK=0xff00; // 16 bits for the index to UChars and other extra data private static final int EXTRA_SHIFT=16; // start of surrogate specials after shift @@ -123,14 +129,16 @@ public final class NormalizerImpl { /* AUX constants */ /* value constants for auxTrie */ - private static final int AUX_UNSAFE_SHIFT = 11; - private static final int AUX_COMP_EX_SHIFT = 10; - - private static final int AUX_MAX_FNC = ((int)1<>(EXTRA_SHIFT-SURROGATE_BLOCK_BITS))& @@ -190,7 +199,7 @@ public final class NormalizerImpl { * the folding offset * @return data offset or 0 if there is no data for the lead surrogate */ - + /* fcdTrie: the folding offset is the lead FCD value itself */ public int getFoldingOffset(int value){ return value; } @@ -205,8 +214,9 @@ public final class NormalizerImpl { * the folding offset * @return data offset or 0 if there is no data for the lead surrogate */ + /* auxTrie: the folding offset is in bits 9..0 of the 16-bit trie result */ public int getFoldingOffset(int value){ - return (value&AUX_FNC_MASK)<<5; + return (int)(value &AUX_FNC_MASK)<=1) ); + isFormatVersion_2_2 =( formatVersion[0]>2 + || + (formatVersion[0]==2 && formatVersion[1]>=2) + ); b.close(); i.close(); } } - + /* ---------------------------------------------------------------------- */ /* Korean Hangul and Jamo constants */ @@ -426,7 +441,10 @@ public final class NormalizerImpl { * */ return fcdTrieImpl.fcdTrie.getTrailValue(fcd16, c2); } - + public static int getFCD16(int c) { + return fcdTrieImpl.fcdTrie.getValue(c); + } + private static int getExtraDataIndex(long norm32) { return (int)(norm32>>EXTRA_SHIFT); } @@ -2502,7 +2520,7 @@ public final class NormalizerImpl { /* found? */ if(c==table[start]) { i=table[start+1]; - if((i&CANON_SET_BMP_MASK)==CANON_SET_BMP_IS_INDEX) { + if((i & CANON_SET_BMP_MASK)==CANON_SET_BMP_IS_INDEX) { /* result 01xxxxxx xxxxxx contains index x to a * USerializedSet */ i&=(CANON_SET_MAX_CANON_SETS-1); @@ -2548,7 +2566,7 @@ public final class NormalizerImpl { * single-code point set {x} in * triplet { 100xxxxx 000hhhhh llllllll llllllll xxxxxxxx xxxxxxxx } */ - i|=((int)h&0x1f00)<<8; /* add high bits from high(c) */ + i|=((int)h & 0x1f00)<<8; /* add high bits from high(c) */ fillSet.setSerializedToOne((int)i); return true; } @@ -2594,6 +2612,84 @@ public final class NormalizerImpl { } } + + /* Is c an NF-skippable code point? See unormimp.h. */ + public static boolean isNFSkippable(int c, Normalizer.Mode mode, long mask) { + long /*unsigned int*/ norm32; + mask = mask & UNSIGNED_INT_MASK; + char aux, fcd; + + /* check conditions (a)..(e), see unormimp.h */ + norm32 = getNorm32(c); + + if((norm32&mask)!=0) { + return false; /* fails (a)..(e), not skippable */ + } + + if(mode == Normalizer.NFD || mode == Normalizer.NFKD || mode == Normalizer.NONE){ + return true; /* NF*D, passed (a)..(c), is skippable */ + } + /* check conditions (a)..(e), see unormimp.h */ + + /* NF*C/FCC, passed (a)..(e) */ + if((norm32& QC_NFD)==0) { + return true; /* no canonical decomposition, is skippable */ + } + + /* check Hangul syllables algorithmically */ + if(isNorm32HangulOrJamo(norm32)) { + /* Jamo passed (a)..(e) above, must be Hangul */ + return !isHangulWithoutJamoT((char)c); /* LVT are skippable, LV are not */ + } + + /* if(mode<=UNORM_NFKC) { -- enable when implementing FCC */ + /* NF*C, test (f) flag */ + if(!isFormatVersion_2_2) { + return false; /* no (f) data, say not skippable to be safe */ + } + + + aux = auxTrieImpl.auxTrie.getCodePointValue(c); + return (aux&AUX_NFC_SKIP_F_MASK)==0; /* TRUE=skippable if the (f) flag is not set */ + + /* } else { FCC, test fcd<=1 instead of the above } */ + } + /* + private static final boolean + _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 limit, uint32_t value) { + // add the start code point to the USet + uset_add((USet *)context, start); + return TRUE; + } + */ + + public static UnicodeSet addPropertyStarts(UnicodeSet set) { + int c; + + /* add the start code point of each same-value range of each trie */ + //utrie_enum(&normTrie, NULL, _enumPropertyStartsRange, set); + TrieIterator normIter = new TrieIterator(normTrieImpl.normTrie); + RangeValueIterator.Element result = new RangeValueIterator.Element(); + + while(normIter.next(result)){ + set.add(result.start); + } + + TrieIterator fcdIter = new TrieIterator(fcdTrieImpl.fcdTrie); + //utrie_enum(&fcdTrie, NULL, _enumPropertyStartsRange, set); + while(normIter.next(result)){ + set.add(result.start); + } + + /* add Hangul LV syllables and LV+1 because of skippables */ + for(c=HANGUL_BASE; c1); + } }; @@ -2010,7 +2035,7 @@ public final class Normalizer implements Cloneable{ return buffer.length-startIndex[0]; } - private static int previous(UCharacterIterator src, + protected static int previous(UCharacterIterator src, char[] dest, int destStart, int destLimit, Mode mode, /*int options,*/ @@ -2238,7 +2263,7 @@ public final class Normalizer implements Cloneable{ return bufferIndex; } - private static int next(UCharacterIterator src, + protected static int next(UCharacterIterator src, char[] dest, int destStart, int destLimit, Normalizer.Mode mode, /*int options,*/ boolean doNormalize, boolean[] pNeededToNormalize){ @@ -2363,5 +2388,12 @@ public final class Normalizer implements Cloneable{ return buffer[index]; } - + + /** + * Internal API + * @internal + */ + public static boolean isNFSkippable(int c, Mode mode){ + return mode.isNFSkippable(c); + } }