You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@lucene.apache.org by rm...@apache.org on 2011/02/09 18:47:15 UTC

svn commit: r1068997 [1/3] - in /lucene/dev/branches/branch_3x: ./ lucene/ lucene/contrib/icu/src/data/uax29/ lucene/contrib/icu/src/java/org/apache/lucene/analysis/icu/segmentation/ lucene/contrib/icu/src/resources/org/apache/lucene/analysis/icu/segme...

Author: rmuir
Date: Wed Feb  9 17:47:13 2011
New Revision: 1068997

URL: http://svn.apache.org/viewvc?rev=1068997&view=rev
Log:
LUCENE-2911: synchronize grammar/token types across StandardTokenizer, UAX29EmailURLTokenizer, ICUTokenizer; add CJK types

Added:
    lucene/dev/branches/branch_3x/lucene/contrib/icu/src/data/uax29/Default.rbbi
      - copied unchanged from r1068979, lucene/dev/trunk/modules/analysis/icu/src/data/uax29/Default.rbbi
    lucene/dev/branches/branch_3x/lucene/contrib/icu/src/resources/org/apache/lucene/analysis/icu/segmentation/Default.brk
      - copied unchanged from r1068979, lucene/dev/trunk/modules/analysis/icu/src/resources/org/apache/lucene/analysis/icu/segmentation/Default.brk
Modified:
    lucene/dev/branches/branch_3x/   (props changed)
    lucene/dev/branches/branch_3x/lucene/   (props changed)
    lucene/dev/branches/branch_3x/lucene/CHANGES.txt
    lucene/dev/branches/branch_3x/lucene/contrib/icu/src/java/org/apache/lucene/analysis/icu/segmentation/DefaultICUTokenizerConfig.java
    lucene/dev/branches/branch_3x/lucene/contrib/icu/src/test/org/apache/lucene/analysis/icu/segmentation/TestICUTokenizer.java
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ClassicTokenizerImpl.java
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/SUPPLEMENTARY.jflex-macro
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizer.java
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.java
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.jflex
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/UAX29URLEmailTokenizer.java
    lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/UAX29URLEmailTokenizer.jflex
    lucene/dev/branches/branch_3x/lucene/src/test/org/apache/lucene/analysis/TestStandardAnalyzer.java
    lucene/dev/branches/branch_3x/lucene/src/test/org/apache/lucene/analysis/TestUAX29URLEmailTokenizer.java
    lucene/dev/branches/branch_3x/solr/   (props changed)

Modified: lucene/dev/branches/branch_3x/lucene/CHANGES.txt
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/CHANGES.txt?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/CHANGES.txt (original)
+++ lucene/dev/branches/branch_3x/lucene/CHANGES.txt Wed Feb  9 17:47:13 2011
@@ -509,6 +509,13 @@ New features
   IndexReader, to allow apps that maintain external per-segment caches
   to evict entries when a segment is finished.  (Shay Banon, Yonik
   Seeley, Mike McCandless)
+
+* LUCENE-2911: The new StandardTokenizer, UAX29URLEmailTokenizer, and
+  the ICUTokenizer in contrib now all tag types with a consistent set
+  of token types (defined in StandardTokenizer). Tokens in the major
+  CJK types are explicitly marked to allow for custom downstream handling: 
+  <IDEOGRAPHIC>, <HANGUL>, <KATAKANA>, and <HIRAGANA>.  
+  (Robert Muir, Steven Rowe)
   
 Optimizations
 

Modified: lucene/dev/branches/branch_3x/lucene/contrib/icu/src/java/org/apache/lucene/analysis/icu/segmentation/DefaultICUTokenizerConfig.java
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/contrib/icu/src/java/org/apache/lucene/analysis/icu/segmentation/DefaultICUTokenizerConfig.java?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/contrib/icu/src/java/org/apache/lucene/analysis/icu/segmentation/DefaultICUTokenizerConfig.java (original)
+++ lucene/dev/branches/branch_3x/lucene/contrib/icu/src/java/org/apache/lucene/analysis/icu/segmentation/DefaultICUTokenizerConfig.java Wed Feb  9 17:47:13 2011
@@ -20,6 +20,8 @@ package org.apache.lucene.analysis.icu.s
 import java.io.IOException;
 import java.io.InputStream;
 
+import org.apache.lucene.analysis.standard.StandardTokenizer;
+
 import com.ibm.icu.lang.UScript;
 import com.ibm.icu.text.BreakIterator;
 import com.ibm.icu.text.RuleBasedBreakIterator;
@@ -44,20 +46,24 @@ import com.ibm.icu.util.ULocale;
  */
 public class DefaultICUTokenizerConfig extends ICUTokenizerConfig {
   /** Token type for words containing ideographic characters */
-  public static final String WORD_IDEO = "<IDEOGRAPHIC>";
-  /** Token type for words containing Japanese kana */
-  public static final String WORD_KANA = "<KANA>";
+  public static final String WORD_IDEO = StandardTokenizer.TOKEN_TYPES[StandardTokenizer.IDEOGRAPHIC];
+  /** Token type for words containing Japanese hiragana */
+  public static final String WORD_HIRAGANA = StandardTokenizer.TOKEN_TYPES[StandardTokenizer.HIRAGANA];
+  /** Token type for words containing Japanese katakana */
+  public static final String WORD_KATAKANA = StandardTokenizer.TOKEN_TYPES[StandardTokenizer.KATAKANA];
+  /** Token type for words containing Korean hangul  */
+  public static final String WORD_HANGUL = StandardTokenizer.TOKEN_TYPES[StandardTokenizer.HANGUL];
   /** Token type for words that contain letters */
-  public static final String WORD_LETTER = "<ALPHANUM>";
+  public static final String WORD_LETTER = StandardTokenizer.TOKEN_TYPES[StandardTokenizer.ALPHANUM];
   /** Token type for words that appear to be numbers */
-  public static final String WORD_NUMBER = "<NUM>";
+  public static final String WORD_NUMBER = StandardTokenizer.TOKEN_TYPES[StandardTokenizer.NUM];
   
   /*
    * the default breakiterators in use. these can be expensive to
    * instantiate, cheap to clone.
    */  
   private static final BreakIterator rootBreakIterator = 
-    BreakIterator.getWordInstance(ULocale.ROOT);
+    readBreakIterator("Default.brk");
   private static final BreakIterator thaiBreakIterator = 
     BreakIterator.getWordInstance(new ULocale("th_TH"));
   private static final BreakIterator hebrewBreakIterator = 
@@ -87,9 +93,9 @@ public class DefaultICUTokenizerConfig e
       case RuleBasedBreakIterator.WORD_IDEO:
         return WORD_IDEO;
       case RuleBasedBreakIterator.WORD_KANA:
-        return WORD_KANA;
+        return script == UScript.HIRAGANA ? WORD_HIRAGANA : WORD_KATAKANA;
       case RuleBasedBreakIterator.WORD_LETTER:
-        return WORD_LETTER;
+        return script == UScript.HANGUL ? WORD_HANGUL : WORD_LETTER;
       case RuleBasedBreakIterator.WORD_NUMBER:
         return WORD_NUMBER;
       default: /* some other custom code */

Modified: lucene/dev/branches/branch_3x/lucene/contrib/icu/src/test/org/apache/lucene/analysis/icu/segmentation/TestICUTokenizer.java
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/contrib/icu/src/test/org/apache/lucene/analysis/icu/segmentation/TestICUTokenizer.java?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/contrib/icu/src/test/org/apache/lucene/analysis/icu/segmentation/TestICUTokenizer.java (original)
+++ lucene/dev/branches/branch_3x/lucene/contrib/icu/src/test/org/apache/lucene/analysis/icu/segmentation/TestICUTokenizer.java Wed Feb  9 17:47:13 2011
@@ -128,11 +128,10 @@ public class TestICUTokenizer extends Ba
   
   /*
    * For chinese, tokenize as char (these can later form bigrams or whatever)
-   * TODO: why do full-width numerics have no word-break prop?
    */
   public void testChinese() throws Exception {
     assertAnalyzesTo(a, "我是中国人。 1234 Tests ",
-        new String[] { "我", "是", "中", "国", "人", "tests"});
+        new String[] { "我", "是", "中", "国", "人", "1234", "tests"});
   }
   
   public void testEmpty() throws Exception {
@@ -221,4 +220,16 @@ public class TestICUTokenizer extends Ba
         new String[] {"david", "has", "5000", "bones"},
         new String[] { "<ALPHANUM>", "<ALPHANUM>", "<NUM>", "<ALPHANUM>" });
   }
+  
+  public void testKorean() throws Exception {
+    BaseTokenStreamTestCase.assertAnalyzesTo(a, "훈민정음",
+        new String[] { "훈민정음" },
+        new String[] { "<HANGUL>" });
+  }
+  
+  public void testJapanese() throws Exception {
+    BaseTokenStreamTestCase.assertAnalyzesTo(a, "仮名遣い カタカナ",
+        new String[] { "仮", "名", "遣", "い", "カタカナ" },
+        new String[] { "<IDEOGRAPHIC>", "<IDEOGRAPHIC>", "<IDEOGRAPHIC>", "<HIRAGANA>", "<KATAKANA>" });
+  }
 }

Modified: lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro (original)
+++ lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro Wed Feb  9 17:47:13 2011
@@ -15,8 +15,8 @@
  */
 
 // Generated from IANA Root Zone Database <http://www.internic.net/zones/root.zone>
-// file version from Thursday, January 6, 2011 12:34:09 PM UTC
-// generated on Thursday, January 6, 2011 2:34:39 PM UTC
+// file version from Wednesday, February 9, 2011 12:34:10 PM UTC
+// generated on Wednesday, February 9, 2011 4:45:18 PM UTC
 // by org.apache.lucene.analysis.standard.GenerateJflexTLDMacros
 
 ASCIITLD = "." (
@@ -285,13 +285,19 @@ ASCIITLD = "." (
 	| [wW][sS]
 	| [xX][nN]--0[zZ][wW][mM]56[dD]
 	| [xX][nN]--11[bB]5[bB][sS]3[aA]9[aA][jJ]6[gG]
+	| [xX][nN]--3[eE]0[bB]707[eE]
+	| [xX][nN]--45[bB][rR][jJ]9[cC]
 	| [xX][nN]--80[aA][kK][hH][bB][yY][kK][nN][jJ]4[fF]
 	| [xX][nN]--9[tT]4[bB]11[yY][iI]5[aA]
+	| [xX][nN]--[cC][lL][cC][hH][cC]0[eE][aA]0[bB]2[gG]2[aA]9[gG][cC][dD]
 	| [xX][nN]--[dD][eE][bB][aA]0[aA][dD]
 	| [xX][nN]--[fF][iI][qQ][sS]8[sS]
 	| [xX][nN]--[fF][iI][qQ][zZ]9[sS]
+	| [xX][nN]--[fF][pP][cC][rR][jJ]9[cC]3[dD]
 	| [xX][nN]--[fF][zZ][cC]2[cC]9[eE]2[cC]
 	| [xX][nN]--[gG]6[wW]251[dD]
+	| [xX][nN]--[gG][eE][cC][rR][jJ]9[cC]
+	| [xX][nN]--[hH]2[bB][rR][jJ]9[cC]
 	| [xX][nN]--[hH][gG][bB][kK]6[aA][jJ]7[fF]53[bB][bB][aA]
 	| [xX][nN]--[hH][lL][cC][jJ]6[aA][yY][aA]9[eE][sS][cC]7[aA]
 	| [xX][nN]--[jJ]6[wW]193[gG]
@@ -301,13 +307,18 @@ ASCIITLD = "." (
 	| [xX][nN]--[kK][pP][rR][yY]57[dD]
 	| [xX][nN]--[mM][gG][bB][aA][aA][mM]7[aA]8[hH]
 	| [xX][nN]--[mM][gG][bB][aA][yY][hH]7[gG][pP][aA]
+	| [xX][nN]--[mM][gG][bB][bB][hH]1[aA]71[eE]
 	| [xX][nN]--[mM][gG][bB][eE][rR][pP]4[aA]5[dD]4[aA][rR]
 	| [xX][nN]--[oO]3[cC][wW]4[hH]
+	| [xX][nN]--[oO][gG][bB][pP][fF]8[fF][lL]
 	| [xX][nN]--[pP]1[aA][iI]
 	| [xX][nN]--[pP][gG][bB][sS]0[dD][hH]
+	| [xX][nN]--[sS]9[bB][rR][jJ]9[cC]
 	| [xX][nN]--[wW][gG][bB][hH]1[cC]
 	| [xX][nN]--[wW][gG][bB][lL]6[aA]
 	| [xX][nN]--[xX][kK][cC]2[aA][lL]3[hH][yY][eE]2[aA]
+	| [xX][nN]--[xX][kK][cC]2[dD][lL]3[aA]5[eE][eE]0[hH]
+	| [xX][nN]--[yY][fF][rR][oO]4[iI]67[oO]
 	| [xX][nN]--[yY][gG][bB][iI]2[aA][mM][mM][xX]
 	| [xX][nN]--[zZ][cC][kK][zZ][aA][hH]
 	| [yY][eE]

Modified: lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ClassicTokenizerImpl.java
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ClassicTokenizerImpl.java?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ClassicTokenizerImpl.java (original)
+++ lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ClassicTokenizerImpl.java Wed Feb  9 17:47:13 2011
@@ -1,4 +1,4 @@
-/* The following code was generated by JFlex 1.5.0-SNAPSHOT on 12/7/10 11:53 AM */
+/* The following code was generated by JFlex 1.5.0-SNAPSHOT on 2/9/11 11:45 AM */
 
 package org.apache.lucene.analysis.standard;
 
@@ -33,8 +33,8 @@ import org.apache.lucene.analysis.tokena
 /**
  * This class is a scanner generated by 
  * <a href="http://www.jflex.de/">JFlex</a> 1.5.0-SNAPSHOT
- * on 12/7/10 11:53 AM from the specification file
- * <tt>C:/cygwin/home/us/svn/lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/ClassicTokenizerImpl.jflex</tt>
+ * on 2/9/11 11:45 AM from the specification file
+ * <tt>C:/Users/rmuir/workspace/lucene-2911/modules/analysis/common/src/java/org/apache/lucene/analysis/standard/ClassicTokenizerImpl.jflex</tt>
  */
 class ClassicTokenizerImpl implements StandardTokenizerInterface {
 
@@ -355,11 +355,6 @@ public static final int EMAIL           
 public static final int HOST              = StandardTokenizer.HOST;
 public static final int NUM               = StandardTokenizer.NUM;
 public static final int CJ                = StandardTokenizer.CJ;
-/**
- * @deprecated this solves a bug where HOSTs that end with '.' are identified
- *             as ACRONYMs.
- */
-@Deprecated
 public static final int ACRONYM_DEP       = StandardTokenizer.ACRONYM_DEP;
 
 public static final String [] TOKEN_TYPES = StandardTokenizer.TOKEN_TYPES;
@@ -687,44 +682,44 @@ public final void getText(CharTermAttrib
       zzMarkedPos = zzMarkedPosL;
 
       switch (zzAction < 0 ? zzAction : ZZ_ACTION[zzAction]) {
-        case 5: 
-          { return NUM;
+        case 10: 
+          { return EMAIL;
           }
         case 11: break;
-        case 9: 
-          { return ACRONYM;
+        case 2: 
+          { return ALPHANUM;
           }
         case 12: break;
-        case 7: 
-          { return COMPANY;
+        case 4: 
+          { return HOST;
           }
         case 13: break;
-        case 10: 
-          { return EMAIL;
-          }
-        case 14: break;
         case 1: 
           { /* ignore */
           }
+        case 14: break;
+        case 8: 
+          { return ACRONYM_DEP;
+          }
         case 15: break;
-        case 6: 
-          { return APOSTROPHE;
+        case 5: 
+          { return NUM;
           }
         case 16: break;
-        case 3: 
-          { return CJ;
+        case 9: 
+          { return ACRONYM;
           }
         case 17: break;
-        case 8: 
-          { return ACRONYM_DEP;
+        case 7: 
+          { return COMPANY;
           }
         case 18: break;
-        case 2: 
-          { return ALPHANUM;
+        case 6: 
+          { return APOSTROPHE;
           }
         case 19: break;
-        case 4: 
-          { return HOST;
+        case 3: 
+          { return CJ;
           }
         case 20: break;
         default: 

Modified: lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/SUPPLEMENTARY.jflex-macro
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/SUPPLEMENTARY.jflex-macro?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/SUPPLEMENTARY.jflex-macro (original)
+++ lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/SUPPLEMENTARY.jflex-macro Wed Feb  9 17:47:13 2011
@@ -14,22 +14,22 @@
  * limitations under the License.
  */
 
-// Generated using ICU4J 4.6.0.0 on Thursday, January 6, 2011 7:32:04 PM UTC
+// Generated using ICU4J 4.6.0.0 on Wednesday, February 9, 2011 4:45:11 PM UTC
 // by org.apache.lucene.analysis.icu.GenerateJFlexSupplementaryMacros
 
 
 ALetterSupp = (
-	  ([\ud804][\uDC03-\uDC37\uDC83-\uDCAF])
+	  ([\ud80d][\uDC00-\uDC2E])
 	| ([\ud80c][\uDC00-\uDFFF])
-	| ([\ud801][\uDC00-\uDC9D])
-	| ([\ud800][\uDC00-\uDC0B\uDC0D-\uDC26\uDC28-\uDC3A\uDC3C\uDC3D\uDC3F-\uDC4D\uDC50-\uDC5D\uDC80-\uDCFA\uDD40-\uDD74\uDE80-\uDE9C\uDEA0-\uDED0\uDF00-\uDF1E\uDF30-\uDF4A\uDF80-\uDF9D\uDFA0-\uDFC3\uDFC8-\uDFCF\uDFD1-\uDFD5])
-	| ([\ud81a][\uDC00-\uDE38])
 	| ([\ud809][\uDC00-\uDC62])
+	| ([\ud808][\uDC00-\uDF6E])
+	| ([\ud81a][\uDC00-\uDE38])
+	| ([\ud804][\uDC03-\uDC37\uDC83-\uDCAF])
 	| ([\ud835][\uDC00-\uDC54\uDC56-\uDC9C\uDC9E\uDC9F\uDCA2\uDCA5\uDCA6\uDCA9-\uDCAC\uDCAE-\uDCB9\uDCBB\uDCBD-\uDCC3\uDCC5-\uDD05\uDD07-\uDD0A\uDD0D-\uDD14\uDD16-\uDD1C\uDD1E-\uDD39\uDD3B-\uDD3E\uDD40-\uDD44\uDD46\uDD4A-\uDD50\uDD52-\uDEA5\uDEA8-\uDEC0\uDEC2-\uDEDA\uDEDC-\uDEFA\uDEFC-\uDF14\uDF16-\uDF34\uDF36-\uDF4E\uDF50-\uDF6E\uDF70-\uDF88\uDF8A-\uDFA8\uDFAA-\uDFC2\uDFC4-\uDFCB])
+	| ([\ud801][\uDC00-\uDC9D])
+	| ([\ud800][\uDC00-\uDC0B\uDC0D-\uDC26\uDC28-\uDC3A\uDC3C\uDC3D\uDC3F-\uDC4D\uDC50-\uDC5D\uDC80-\uDCFA\uDD40-\uDD74\uDE80-\uDE9C\uDEA0-\uDED0\uDF00-\uDF1E\uDF30-\uDF4A\uDF80-\uDF9D\uDFA0-\uDFC3\uDFC8-\uDFCF\uDFD1-\uDFD5])
 	| ([\ud803][\uDC00-\uDC48])
-	| ([\ud808][\uDC00-\uDF6E])
 	| ([\ud802][\uDC00-\uDC05\uDC08\uDC0A-\uDC35\uDC37\uDC38\uDC3C\uDC3F-\uDC55\uDD00-\uDD15\uDD20-\uDD39\uDE00\uDE10-\uDE13\uDE15-\uDE17\uDE19-\uDE33\uDE60-\uDE7C\uDF00-\uDF35\uDF40-\uDF55\uDF60-\uDF72])
-	| ([\ud80d][\uDC00-\uDC2E])
 )
 FormatSupp = (
 	  ([\ud804][\uDCBD])
@@ -45,8 +45,8 @@ ExtendSupp = (
 )
 NumericSupp = (
 	  ([\ud804][\uDC66-\uDC6F])
-	| ([\ud801][\uDCA0-\uDCA9])
 	| ([\ud835][\uDFCE-\uDFFF])
+	| ([\ud801][\uDCA0-\uDCA9])
 )
 KatakanaSupp = (
 	  ([\ud82c][\uDC00])
@@ -70,56 +70,56 @@ ComplexContextSupp = (
 	  []
 )
 HanSupp = (
-	  ([\ud858][\uDC00-\uDFFF])
-	| ([\ud865][\uDC00-\uDFFF])
-	| ([\ud87e][\uDC00-\uDE1D])
-	| ([\ud85f][\uDC00-\uDFFF])
+	  ([\ud87e][\uDC00-\uDE1D])
+	| ([\ud86b][\uDC00-\uDFFF])
+	| ([\ud86a][\uDC00-\uDFFF])
+	| ([\ud869][\uDC00-\uDED6\uDF00-\uDFFF])
+	| ([\ud868][\uDC00-\uDFFF])
+	| ([\ud86e][\uDC00-\uDC1D])
+	| ([\ud86d][\uDC00-\uDF34\uDF40-\uDFFF])
 	| ([\ud86c][\uDC00-\uDFFF])
-	| ([\ud84c][\uDC00-\uDFFF])
-	| ([\ud857][\uDC00-\uDFFF])
-	| ([\ud841][\uDC00-\uDFFF])
-	| ([\ud860][\uDC00-\uDFFF])
-	| ([\ud849][\uDC00-\uDFFF])
-	| ([\ud840][\uDC00-\uDFFF])
-	| ([\ud853][\uDC00-\uDFFF])
-	| ([\ud859][\uDC00-\uDFFF])
-	| ([\ud844][\uDC00-\uDFFF])
-	| ([\ud856][\uDC00-\uDFFF])
-	| ([\ud84a][\uDC00-\uDFFF])
 	| ([\ud863][\uDC00-\uDFFF])
-	| ([\ud843][\uDC00-\uDFFF])
-	| ([\ud86e][\uDC00-\uDC1D])
-	| ([\ud866][\uDC00-\uDFFF])
-	| ([\ud86b][\uDC00-\uDFFF])
-	| ([\ud84b][\uDC00-\uDFFF])
-	| ([\ud85e][\uDC00-\uDFFF])
-	| ([\ud851][\uDC00-\uDFFF])
-	| ([\ud842][\uDC00-\uDFFF])
 	| ([\ud862][\uDC00-\uDFFF])
-	| ([\ud86d][\uDC00-\uDF34\uDF40-\uDFFF])
-	| ([\ud845][\uDC00-\uDFFF])
-	| ([\ud85b][\uDC00-\uDFFF])
 	| ([\ud861][\uDC00-\uDFFF])
-	| ([\ud84e][\uDC00-\uDFFF])
-	| ([\ud868][\uDC00-\uDFFF])
-	| ([\ud848][\uDC00-\uDFFF])
-	| ([\ud854][\uDC00-\uDFFF])
-	| ([\ud869][\uDC00-\uDED6\uDF00-\uDFFF])
-	| ([\ud84f][\uDC00-\uDFFF])
-	| ([\ud85c][\uDC00-\uDFFF])
+	| ([\ud860][\uDC00-\uDFFF])
 	| ([\ud867][\uDC00-\uDFFF])
-	| ([\ud847][\uDC00-\uDFFF])
+	| ([\ud866][\uDC00-\uDFFF])
+	| ([\ud865][\uDC00-\uDFFF])
 	| ([\ud864][\uDC00-\uDFFF])
-	| ([\ud850][\uDC00-\uDFFF])
-	| ([\ud86a][\uDC00-\uDFFF])
+	| ([\ud858][\uDC00-\uDFFF])
+	| ([\ud859][\uDC00-\uDFFF])
+	| ([\ud85a][\uDC00-\uDFFF])
+	| ([\ud85b][\uDC00-\uDFFF])
+	| ([\ud85c][\uDC00-\uDFFF])
 	| ([\ud85d][\uDC00-\uDFFF])
+	| ([\ud85e][\uDC00-\uDFFF])
+	| ([\ud85f][\uDC00-\uDFFF])
+	| ([\ud850][\uDC00-\uDFFF])
+	| ([\ud851][\uDC00-\uDFFF])
 	| ([\ud852][\uDC00-\uDFFF])
+	| ([\ud853][\uDC00-\uDFFF])
+	| ([\ud854][\uDC00-\uDFFF])
 	| ([\ud855][\uDC00-\uDFFF])
-	| ([\ud846][\uDC00-\uDFFF])
-	| ([\ud85a][\uDC00-\uDFFF])
+	| ([\ud856][\uDC00-\uDFFF])
+	| ([\ud857][\uDC00-\uDFFF])
+	| ([\ud849][\uDC00-\uDFFF])
+	| ([\ud848][\uDC00-\uDFFF])
+	| ([\ud84b][\uDC00-\uDFFF])
+	| ([\ud84a][\uDC00-\uDFFF])
 	| ([\ud84d][\uDC00-\uDFFF])
+	| ([\ud84c][\uDC00-\uDFFF])
+	| ([\ud84f][\uDC00-\uDFFF])
+	| ([\ud84e][\uDC00-\uDFFF])
+	| ([\ud841][\uDC00-\uDFFF])
+	| ([\ud840][\uDC00-\uDFFF])
+	| ([\ud843][\uDC00-\uDFFF])
+	| ([\ud842][\uDC00-\uDFFF])
+	| ([\ud845][\uDC00-\uDFFF])
+	| ([\ud844][\uDC00-\uDFFF])
+	| ([\ud847][\uDC00-\uDFFF])
+	| ([\ud846][\uDC00-\uDFFF])
 )
 HiraganaSupp = (
-	  ([\ud82c][\uDC01])
-	| ([\ud83c][\uDE00])
+	  ([\ud83c][\uDE00])
+	| ([\ud82c][\uDC01])
 )

Modified: lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizer.java
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizer.java?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizer.java (original)
+++ lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizer.java Wed Feb  9 17:47:13 2011
@@ -81,6 +81,8 @@ public final class StandardTokenizer ext
   public static final int SOUTHEAST_ASIAN = 9;
   public static final int IDEOGRAPHIC = 10;
   public static final int HIRAGANA = 11;
+  public static final int KATAKANA = 12;
+  public static final int HANGUL = 13;
   
   /** String token types that correspond to token type int constants */
   public static final String [] TOKEN_TYPES = new String [] {
@@ -95,7 +97,9 @@ public final class StandardTokenizer ext
     "<ACRONYM_DEP>",
     "<SOUTHEAST_ASIAN>",
     "<IDEOGRAPHIC>",
-    "<HIRAGANA>"
+    "<HIRAGANA>",
+    "<KATAKANA>",
+    "<HANGUL>"
   };
 
   private boolean replaceInvalidAcronym;

Modified: lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.java
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.java?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.java (original)
+++ lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.java Wed Feb  9 17:47:13 2011
@@ -1,4 +1,4 @@
-/* The following code was generated by JFlex 1.5.0-SNAPSHOT on 1/6/11 9:34 AM */
+/* The following code was generated by JFlex 1.5.0-SNAPSHOT on 2/9/11 11:45 AM */
 
 package org.apache.lucene.analysis.standard;
 
@@ -116,84 +116,85 @@ public final class StandardTokenizerImpl
     "\1\133\71\0\53\142\24\143\1\142\12\134\6\0\6\142\4\143\4\142"+
     "\3\143\1\142\3\143\2\142\7\143\3\142\4\143\15\142\14\143\1\142"+
     "\1\143\12\134\4\143\2\142\46\132\12\0\53\132\1\0\1\132\3\0"+
-    "\u0149\132\1\0\4\132\2\0\7\132\1\0\1\132\1\0\4\132\2\0"+
-    "\51\132\1\0\4\132\2\0\41\132\1\0\4\132\2\0\7\132\1\0"+
-    "\1\132\1\0\4\132\2\0\17\132\1\0\71\132\1\0\4\132\2\0"+
-    "\103\132\2\0\3\133\40\0\20\132\20\0\125\132\14\0\u026c\132\2\0"+
-    "\21\132\1\0\32\132\5\0\113\132\3\0\3\132\17\0\15\132\1\0"+
-    "\4\132\3\133\13\0\22\132\3\133\13\0\22\132\2\133\14\0\15\132"+
-    "\1\0\3\132\1\0\2\133\14\0\64\142\2\143\36\143\3\0\1\142"+
-    "\4\0\1\142\1\143\2\0\12\134\41\0\3\133\2\0\12\134\6\0"+
-    "\130\132\10\0\51\132\1\133\1\132\5\0\106\132\12\0\35\132\3\0"+
-    "\14\133\4\0\14\133\12\0\12\134\36\142\2\0\5\142\13\0\54\142"+
-    "\4\0\21\143\7\142\2\143\6\0\12\134\1\142\3\0\2\142\40\0"+
-    "\27\132\5\133\4\0\65\142\12\143\1\0\35\143\2\0\1\133\12\134"+
-    "\6\0\12\134\6\0\16\142\122\0\5\133\57\132\21\133\7\132\4\0"+
-    "\12\134\21\0\11\133\14\0\3\133\36\132\12\133\3\0\2\132\12\134"+
-    "\6\0\46\132\16\133\14\0\44\132\24\133\10\0\12\134\3\0\3\132"+
-    "\12\134\44\132\122\0\3\133\1\0\25\133\4\132\1\133\4\132\1\133"+
-    "\15\0\300\132\47\133\25\0\4\133\u0116\132\2\0\6\132\2\0\46\132"+
-    "\2\0\6\132\2\0\10\132\1\0\1\132\1\0\1\132\1\0\1\132"+
-    "\1\0\37\132\2\0\65\132\1\0\7\132\1\0\1\132\3\0\3\132"+
-    "\1\0\7\132\3\0\4\132\2\0\6\132\4\0\15\132\5\0\3\132"+
-    "\1\0\7\132\17\0\2\133\2\133\10\0\2\140\12\0\1\140\2\0"+
-    "\1\136\2\0\5\133\20\0\2\141\3\0\1\137\17\0\1\141\13\0"+
-    "\5\133\5\0\6\133\1\0\1\132\15\0\1\132\20\0\15\132\63\0"+
-    "\41\133\21\0\1\132\4\0\1\132\2\0\12\132\1\0\1\132\3\0"+
-    "\5\132\6\0\1\132\1\0\1\132\1\0\1\132\1\0\4\132\1\0"+
-    "\13\132\2\0\4\132\5\0\5\132\4\0\1\132\21\0\51\132\u032d\0"+
-    "\64\132\u0716\0\57\132\1\0\57\132\1\0\205\132\6\0\4\132\3\133"+
-    "\16\0\46\132\12\0\66\132\11\0\1\132\17\0\1\133\27\132\11\0"+
-    "\7\132\1\0\7\132\1\0\7\132\1\0\7\132\1\0\7\132\1\0"+
-    "\7\132\1\0\7\132\1\0\7\132\1\0\40\133\57\0\1\132\120\0"+
-    "\32\144\1\0\131\144\14\0\326\144\57\0\1\132\1\0\1\144\31\0"+
-    "\11\144\6\133\1\0\5\135\2\0\3\144\1\132\1\132\4\0\126\145"+
-    "\2\0\2\133\2\135\3\145\133\135\1\0\4\135\5\0\51\132\3\0"+
-    "\136\132\21\0\33\132\65\0\20\135\320\0\57\135\1\0\130\135\250\0"+
-    "\u19b6\144\112\0\u51cc\144\64\0\u048d\132\103\0\56\132\2\0\u010d\132\3\0"+
-    "\20\132\12\134\2\132\24\0\57\132\4\133\11\0\2\133\1\0\31\132"+
-    "\10\0\120\132\2\133\45\0\11\132\2\0\147\132\2\0\4\132\1\0"+
-    "\2\132\16\0\12\132\120\0\10\132\1\133\3\132\1\133\4\132\1\133"+
-    "\27\132\5\133\30\0\64\132\14\0\2\133\62\132\21\133\13\0\12\134"+
-    "\6\0\22\133\6\132\3\0\1\132\4\0\12\134\34\132\10\133\2\0"+
-    "\27\132\15\133\14\0\35\132\3\0\4\133\57\132\16\133\16\0\1\132"+
-    "\12\134\46\0\51\132\16\133\11\0\3\132\1\133\10\132\2\133\2\0"+
-    "\12\134\6\0\33\142\1\143\4\0\60\142\1\143\1\142\3\143\2\142"+
-    "\2\143\5\142\2\143\1\142\1\143\1\142\30\0\5\142\41\0\6\132"+
-    "\2\0\6\132\2\0\6\132\11\0\7\132\1\0\7\132\221\0\43\132"+
-    "\10\133\1\0\2\133\2\0\12\134\6\0\u2ba4\132\14\0\27\132\4\0"+
-    "\61\132\4\0\1\10\1\5\1\43\1\33\1\1\3\0\1\36\1\17"+
-    "\2\0\1\3\1\60\14\0\1\14\21\0\1\112\7\0\1\65\1\23"+
-    "\6\0\1\130\3\0\1\114\1\114\1\114\1\114\1\114\1\114\1\114"+
-    "\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114"+
-    "\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114"+
-    "\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114\1\114"+
-    "\1\114\1\114\1\114\1\114\1\126\1\114\1\114\1\114\1\123\1\121"+
-    "\17\0\1\115\u02c1\0\1\70\277\0\1\113\1\73\1\63\3\122\2\61"+
-    "\1\122\1\61\2\122\1\6\21\122\2\62\7\72\1\71\7\72\7\47"+
-    "\1\21\1\47\1\75\2\35\1\34\1\75\1\35\1\34\10\75\2\53"+
-    "\5\52\2\46\5\52\1\22\10\32\5\20\3\7\12\106\20\7\3\13"+
-    "\32\27\1\12\2\24\2\107\1\110\2\107\2\110\2\107\1\110\3\24"+
-    "\1\2\2\24\12\64\1\74\1\31\1\16\1\64\6\31\1\16\66\31"+
-    "\5\116\6\103\1\45\4\103\2\45\10\103\1\45\7\101\1\15\2\101"+
-    "\32\103\1\15\4\101\1\15\5\102\1\100\1\102\3\100\7\102\1\100"+
-    "\23\102\5\67\3\102\6\67\2\67\6\66\10\66\2\101\7\66\36\101"+
-    "\4\66\102\101\15\116\1\77\2\116\1\131\3\120\1\116\2\120\5\116"+
-    "\4\120\4\117\1\116\3\117\1\116\5\117\26\51\4\30\1\104\2\105"+
-    "\4\127\1\105\2\127\3\76\33\127\35\56\3\127\35\124\3\127\6\124"+
-    "\2\40\31\124\1\40\17\124\6\127\4\41\1\37\37\41\1\37\4\41"+
-    "\25\57\1\125\11\57\21\56\5\57\1\55\12\42\13\57\4\56\1\54"+
-    "\6\56\12\127\17\56\1\44\3\50\15\25\11\26\1\11\24\26\2\25"+
-    "\11\26\1\11\31\26\1\11\4\25\4\26\2\11\2\111\1\4\5\111"+
-    "\52\4\u1900\0\u012e\144\2\0\76\144\2\0\152\144\46\0\7\132\14\0"+
-    "\5\132\5\0\1\132\1\133\12\132\1\0\15\132\1\0\5\132\1\0"+
-    "\1\132\1\0\2\132\1\0\2\132\1\0\154\132\41\0\u016b\132\22\0"+
-    "\100\132\2\0\66\132\50\0\14\132\4\0\20\133\1\137\2\0\1\136"+
-    "\1\137\13\0\7\133\14\0\2\141\30\0\3\141\1\137\1\0\1\140"+
-    "\1\0\1\137\1\136\32\0\5\132\1\0\207\132\2\0\1\133\7\0"+
-    "\1\140\4\0\1\137\1\0\1\140\1\0\12\134\1\136\1\137\5\0"+
-    "\32\132\4\0\1\141\1\0\32\132\13\0\70\135\2\133\37\132\3\0"+
-    "\6\132\2\0\6\132\2\0\6\132\2\0\3\132\34\0\3\133\4\0";
+    "\u0100\146\111\132\1\0\4\132\2\0\7\132\1\0\1\132\1\0\4\132"+
+    "\2\0\51\132\1\0\4\132\2\0\41\132\1\0\4\132\2\0\7\132"+
+    "\1\0\1\132\1\0\4\132\2\0\17\132\1\0\71\132\1\0\4\132"+
+    "\2\0\103\132\2\0\3\133\40\0\20\132\20\0\125\132\14\0\u026c\132"+
+    "\2\0\21\132\1\0\32\132\5\0\113\132\3\0\3\132\17\0\15\132"+
+    "\1\0\4\132\3\133\13\0\22\132\3\133\13\0\22\132\2\133\14\0"+
+    "\15\132\1\0\3\132\1\0\2\133\14\0\64\142\2\143\36\143\3\0"+
+    "\1\142\4\0\1\142\1\143\2\0\12\134\41\0\3\133\2\0\12\134"+
+    "\6\0\130\132\10\0\51\132\1\133\1\132\5\0\106\132\12\0\35\132"+
+    "\3\0\14\133\4\0\14\133\12\0\12\134\36\142\2\0\5\142\13\0"+
+    "\54\142\4\0\21\143\7\142\2\143\6\0\12\134\1\142\3\0\2\142"+
+    "\40\0\27\132\5\133\4\0\65\142\12\143\1\0\35\143\2\0\1\133"+
+    "\12\134\6\0\12\134\6\0\16\142\122\0\5\133\57\132\21\133\7\132"+
+    "\4\0\12\134\21\0\11\133\14\0\3\133\36\132\12\133\3\0\2\132"+
+    "\12\134\6\0\46\132\16\133\14\0\44\132\24\133\10\0\12\134\3\0"+
+    "\3\132\12\134\44\132\122\0\3\133\1\0\25\133\4\132\1\133\4\132"+
+    "\1\133\15\0\300\132\47\133\25\0\4\133\u0116\132\2\0\6\132\2\0"+
+    "\46\132\2\0\6\132\2\0\10\132\1\0\1\132\1\0\1\132\1\0"+
+    "\1\132\1\0\37\132\2\0\65\132\1\0\7\132\1\0\1\132\3\0"+
+    "\3\132\1\0\7\132\3\0\4\132\2\0\6\132\4\0\15\132\5\0"+
+    "\3\132\1\0\7\132\17\0\2\133\2\133\10\0\2\140\12\0\1\140"+
+    "\2\0\1\136\2\0\5\133\20\0\2\141\3\0\1\137\17\0\1\141"+
+    "\13\0\5\133\5\0\6\133\1\0\1\132\15\0\1\132\20\0\15\132"+
+    "\63\0\41\133\21\0\1\132\4\0\1\132\2\0\12\132\1\0\1\132"+
+    "\3\0\5\132\6\0\1\132\1\0\1\132\1\0\1\132\1\0\4\132"+
+    "\1\0\13\132\2\0\4\132\5\0\5\132\4\0\1\132\21\0\51\132"+
+    "\u032d\0\64\132\u0716\0\57\132\1\0\57\132\1\0\205\132\6\0\4\132"+
+    "\3\133\16\0\46\132\12\0\66\132\11\0\1\132\17\0\1\133\27\132"+
+    "\11\0\7\132\1\0\7\132\1\0\7\132\1\0\7\132\1\0\7\132"+
+    "\1\0\7\132\1\0\7\132\1\0\7\132\1\0\40\133\57\0\1\132"+
+    "\120\0\32\144\1\0\131\144\14\0\326\144\57\0\1\132\1\0\1\144"+
+    "\31\0\11\144\4\133\2\133\1\0\5\135\2\0\3\144\1\132\1\132"+
+    "\4\0\126\145\2\0\2\133\2\135\3\145\133\135\1\0\4\135\5\0"+
+    "\51\132\3\0\136\146\21\0\33\132\65\0\20\135\37\0\101\0\37\0"+
+    "\121\0\57\135\1\0\130\135\250\0\u19b6\144\112\0\u51cc\144\64\0\u048d\132"+
+    "\103\0\56\132\2\0\u010d\132\3\0\20\132\12\134\2\132\24\0\57\132"+
+    "\4\133\11\0\2\133\1\0\31\132\10\0\120\132\2\133\45\0\11\132"+
+    "\2\0\147\132\2\0\4\132\1\0\2\132\16\0\12\132\120\0\10\132"+
+    "\1\133\3\132\1\133\4\132\1\133\27\132\5\133\30\0\64\132\14\0"+
+    "\2\133\62\132\21\133\13\0\12\134\6\0\22\133\6\132\3\0\1\132"+
+    "\4\0\12\134\34\132\10\133\2\0\27\132\15\133\14\0\35\146\3\0"+
+    "\4\133\57\132\16\133\16\0\1\132\12\134\46\0\51\132\16\133\11\0"+
+    "\3\132\1\133\10\132\2\133\2\0\12\134\6\0\33\142\1\143\4\0"+
+    "\60\142\1\143\1\142\3\143\2\142\2\143\5\142\2\143\1\142\1\143"+
+    "\1\142\30\0\5\142\41\0\6\132\2\0\6\132\2\0\6\132\11\0"+
+    "\7\132\1\0\7\132\221\0\43\132\10\133\1\0\2\133\2\0\12\134"+
+    "\6\0\u2ba4\146\14\0\27\146\4\0\61\146\4\0\1\31\1\25\1\46"+
+    "\1\43\1\13\3\0\1\7\1\5\2\0\1\3\1\1\14\0\1\11"+
+    "\21\0\1\112\7\0\1\65\1\17\6\0\1\130\3\0\1\120\1\120"+
+    "\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120"+
+    "\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120"+
+    "\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120"+
+    "\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\120\1\121"+
+    "\1\120\1\120\1\120\1\125\1\123\17\0\1\114\u02c1\0\1\70\277\0"+
+    "\1\113\1\71\1\2\3\124\2\35\1\124\1\35\2\124\1\14\21\124"+
+    "\2\60\7\73\1\72\7\73\7\52\1\15\1\52\1\75\2\45\1\44"+
+    "\1\75\1\45\1\44\10\75\2\63\5\61\2\54\5\61\1\6\10\37"+
+    "\5\21\3\27\12\106\20\27\3\42\32\30\1\26\2\24\2\110\1\111"+
+    "\2\110\2\111\2\110\1\111\3\24\1\16\2\24\12\64\1\74\1\41"+
+    "\1\34\1\64\6\41\1\34\66\41\5\115\6\103\1\51\4\103\2\51"+
+    "\10\103\1\51\7\100\1\12\2\100\32\103\1\12\4\100\1\12\5\102"+
+    "\1\101\1\102\3\101\7\102\1\101\23\102\5\67\3\102\6\67\2\67"+
+    "\6\66\10\66\2\100\7\66\36\100\4\66\102\100\15\115\1\77\2\115"+
+    "\1\131\3\117\1\115\2\117\5\115\4\117\4\116\1\115\3\116\1\115"+
+    "\5\116\26\56\4\23\1\105\2\104\4\122\1\104\2\122\3\76\33\122"+
+    "\35\55\3\122\35\126\3\122\6\126\2\33\31\126\1\33\17\126\6\122"+
+    "\4\22\1\10\37\22\1\10\4\22\25\62\1\127\11\62\21\55\5\62"+
+    "\1\57\12\40\13\62\4\55\1\50\6\55\12\122\17\55\1\47\3\53"+
+    "\15\20\11\36\1\32\24\36\2\20\11\36\1\32\31\36\1\32\4\20"+
+    "\4\36\2\32\2\107\1\4\5\107\52\4\u1900\0\u012e\144\2\0\76\144"+
+    "\2\0\152\144\46\0\7\132\14\0\5\132\5\0\1\132\1\133\12\132"+
+    "\1\0\15\132\1\0\5\132\1\0\1\132\1\0\2\132\1\0\2\132"+
+    "\1\0\154\132\41\0\u016b\132\22\0\100\132\2\0\66\132\50\0\14\132"+
+    "\4\0\20\133\1\137\2\0\1\136\1\137\13\0\7\133\14\0\2\141"+
+    "\30\0\3\141\1\137\1\0\1\140\1\0\1\137\1\136\32\0\5\132"+
+    "\1\0\207\132\2\0\1\133\7\0\1\140\4\0\1\137\1\0\1\140"+
+    "\1\0\12\134\1\136\1\137\5\0\32\132\4\0\1\141\1\0\32\132"+
+    "\13\0\70\135\2\133\37\146\3\0\6\146\2\0\6\146\2\0\6\146"+
+    "\2\0\3\146\34\0\3\133\4\0";
 
   /** 
    * Translates characters to character classes
@@ -206,11 +207,12 @@ public final class StandardTokenizerImpl
   private static final int [] ZZ_ACTION = zzUnpackAction();
 
   private static final String ZZ_ACTION_PACKED_0 =
-    "\1\0\23\1\1\2\1\3\1\2\1\1\1\4\1\5"+
-    "\1\6\15\0\1\2\1\0\1\2\10\0\1\3\61\0";
+    "\1\0\23\1\1\2\1\3\1\4\1\1\1\5\1\6"+
+    "\1\7\1\10\15\0\1\2\1\0\1\2\10\0\1\3"+
+    "\15\0\1\2\57\0";
 
   private static int [] zzUnpackAction() {
-    int [] result = new int[101];
+    int [] result = new int[114];
     int offset = 0;
     offset = zzUnpackAction(ZZ_ACTION_PACKED_0, offset, result);
     return result;
@@ -235,22 +237,24 @@ public final class StandardTokenizerImpl
   private static final int [] ZZ_ROWMAP = zzUnpackRowMap();
 
   private static final String ZZ_ROWMAP_PACKED_0 =
-    "\0\0\0\146\0\314\0\u0132\0\u0198\0\u01fe\0\u0264\0\u02ca"+
-    "\0\u0330\0\u0396\0\u03fc\0\u0462\0\u04c8\0\u052e\0\u0594\0\u05fa"+
-    "\0\u0660\0\u06c6\0\u072c\0\u0792\0\u07f8\0\u085e\0\u08c4\0\u092a"+
-    "\0\u0990\0\146\0\146\0\u09f6\0\u0132\0\u0a5c\0\u0ac2\0\u0264"+
-    "\0\u02ca\0\u0b28\0\u0396\0\u03fc\0\u0b8e\0\u04c8\0\u0bf4\0\u0c5a"+
-    "\0\u0cc0\0\u0d26\0\u0d8c\0\u0df2\0\u0198\0\u0e58\0\u0330\0\u0ebe"+
-    "\0\u0f24\0\u0f8a\0\u0ff0\0\u1056\0\u10bc\0\u1122\0\u1188\0\u11ee"+
-    "\0\u1254\0\u12ba\0\u1320\0\u1386\0\u13ec\0\u1452\0\u14b8\0\u092a"+
-    "\0\u151e\0\u1584\0\u15ea\0\u1650\0\u16b6\0\u171c\0\u1782\0\u17e8"+
-    "\0\u184e\0\u18b4\0\u191a\0\u1980\0\u19e6\0\u1a4c\0\u1ab2\0\u1b18"+
-    "\0\u1b7e\0\u1be4\0\u1c4a\0\u1cb0\0\u1d16\0\u1d7c\0\u1de2\0\u1e48"+
-    "\0\u1eae\0\u1f14\0\u1f7a\0\u1fe0\0\u2046\0\u20ac\0\u2112\0\u2178"+
-    "\0\u21de\0\u2244\0\u22aa\0\u2310\0\u2376";
+    "\0\0\0\147\0\316\0\u0135\0\u019c\0\u0203\0\u026a\0\u02d1"+
+    "\0\u0338\0\u039f\0\u0406\0\u046d\0\u04d4\0\u053b\0\u05a2\0\u0609"+
+    "\0\u0670\0\u06d7\0\u073e\0\u07a5\0\u080c\0\u0873\0\u08da\0\u0941"+
+    "\0\u09a8\0\147\0\147\0\u0a0f\0\316\0\u0135\0\u019c\0\u0203"+
+    "\0\u026a\0\u0a76\0\u0add\0\u0b44\0\u0bab\0\u046d\0\u0c12\0\u0c79"+
+    "\0\u0ce0\0\u0d47\0\u0dae\0\u0e15\0\u0e7c\0\u0338\0\u039f\0\u0ee3"+
+    "\0\u0f4a\0\u0fb1\0\u1018\0\u107f\0\u10e6\0\u114d\0\u11b4\0\u121b"+
+    "\0\u1282\0\u12e9\0\u1350\0\u13b7\0\u141e\0\u1485\0\u14ec\0\u1553"+
+    "\0\u15ba\0\u0941\0\u1621\0\u1688\0\u16ef\0\u1756\0\u17bd\0\u1824"+
+    "\0\u188b\0\u18f2\0\u1959\0\u19c0\0\u1a27\0\u1a8e\0\u1af5\0\u1b5c"+
+    "\0\u1bc3\0\u1c2a\0\u1c91\0\u1cf8\0\u1d5f\0\u1dc6\0\u1e2d\0\u1e94"+
+    "\0\u1efb\0\u1f62\0\u1fc9\0\u2030\0\u2097\0\u20fe\0\u2165\0\u21cc"+
+    "\0\u2233\0\u229a\0\u2301\0\u2368\0\u23cf\0\u2436\0\u249d\0\u2504"+
+    "\0\u256b\0\u25d2\0\u2639\0\u26a0\0\u2707\0\u276e\0\u27d5\0\u283c"+
+    "\0\u28a3\0\u290a";
 
   private static int [] zzUnpackRowMap() {
-    int [] result = new int[101];
+    int [] result = new int[114];
     int offset = 0;
     offset = zzUnpackRowMap(ZZ_ROWMAP_PACKED_0, offset, result);
     return result;
@@ -273,267 +277,313 @@ public final class StandardTokenizerImpl
   private static final int [] ZZ_TRANS = zzUnpackTrans();
 
   private static final String ZZ_TRANS_PACKED_0 =
-    "\1\2\1\3\1\2\1\4\1\2\1\5\2\2\1\6"+
-    "\3\2\1\7\2\2\1\10\3\2\1\11\7\2\1\12"+
-    "\2\2\1\13\4\2\1\14\14\2\1\15\31\2\1\16"+
-    "\1\2\1\17\1\20\3\2\1\21\1\2\1\22\2\2"+
-    "\1\23\1\2\1\24\1\2\1\25\1\2\1\26\1\27"+
-    "\3\2\1\30\2\31\1\32\1\33\150\0\1\25\3\0"+
-    "\1\25\3\0\1\25\6\0\1\25\2\0\1\25\2\0"+
-    "\1\25\17\0\1\25\11\0\2\25\6\0\2\25\13\0"+
-    "\1\26\2\25\11\0\1\25\25\0\1\25\1\0\1\25"+
-    "\1\0\2\25\1\0\3\25\1\0\2\25\1\0\3\25"+
-    "\1\0\7\25\1\0\2\25\1\0\4\25\1\0\14\25"+
-    "\1\0\4\25\1\0\2\25\1\0\21\25\1\0\1\25"+
-    "\2\0\3\25\1\0\1\25\1\0\2\25\1\0\1\25"+
-    "\1\0\1\25\22\0\2\25\2\0\2\25\4\0\3\25"+
-    "\4\0\1\25\2\0\1\25\1\0\2\25\10\0\2\25"+
-    "\2\0\2\25\5\0\3\25\5\0\3\25\1\0\1\25"+
-    "\10\0\1\25\2\26\2\0\1\25\6\0\1\25\25\0"+
-    "\1\25\6\0\3\25\2\0\1\25\2\0\2\25\1\0"+
-    "\1\25\1\0\2\25\1\0\2\25\2\0\1\25\2\0"+
-    "\1\25\1\0\1\25\4\0\1\25\2\0\2\25\1\0"+
-    "\1\25\1\0\1\25\1\0\4\25\2\0\1\25\2\0"+
-    "\4\25\2\0\1\25\1\0\1\25\4\0\3\25\1\0"+
-    "\1\25\6\0\1\25\1\0\2\25\22\0\1\25\3\0"+
-    "\2\25\2\0\2\25\1\0\2\25\1\0\3\25\1\0"+
-    "\1\25\2\0\4\25\1\0\2\25\7\0\3\25\1\0"+
-    "\3\25\5\0\4\25\1\0\2\25\1\0\5\25\1\0"+
-    "\6\25\1\0\3\25\2\0\1\25\2\0\3\25\1\0"+
-    "\1\25\6\0\1\25\22\0\1\25\11\0\3\25\7\0"+
-    "\1\25\1\0\2\25\10\0\2\25\2\0\2\25\5\0"+
-    "\3\25\5\0\3\25\1\0\1\25\15\0\1\25\6\0"+
-    "\1\25\27\0\1\26\1\0\2\25\3\0\1\25\4\0"+
-    "\2\25\2\0\7\25\1\0\2\25\3\0\2\25\3\0"+
-    "\6\25\2\0\2\25\1\0\4\25\1\0\2\25\1\0"+
-    "\3\25\1\0\3\25\1\0\6\25\1\0\1\25\1\26"+
-    "\1\0\1\25\2\0\3\25\1\0\1\25\1\0\1\25"+
-    "\2\0\1\25\1\0\1\25\22\0\1\25\12\0\1\25"+
-    "\12\0\2\25\11\0\1\25\3\0\1\25\5\0\3\25"+
-    "\5\0\3\25\1\0\1\25\15\0\1\25\6\0\1\25"+
-    "\25\0\1\25\3\0\2\25\2\0\2\25\1\0\2\25"+
-    "\1\0\3\25\1\0\1\25\2\0\4\25\1\0\2\25"+
-    "\1\0\4\25\2\0\3\25\1\0\7\25\1\0\4\25"+
-    "\1\0\2\25\1\0\20\25\2\0\1\25\2\0\3\25"+
-    "\1\0\1\25\1\0\2\25\1\0\1\25\1\0\1\25"+
-    "\22\0\1\25\13\0\1\25\21\0\14\25\2\0\2\25"+
-    "\5\0\3\25\1\0\1\25\5\0\1\25\7\0\1\25"+
-    "\3\0\1\25\2\0\1\25\2\0\1\25\3\0\1\25"+
-    "\22\0\1\25\52\0\3\25\5\0\3\25\17\0\1\25"+
-    "\6\0\1\25\116\0\1\33\17\0\1\27\34\0\1\32"+
-    "\1\0\1\32\1\0\2\32\1\0\3\32\1\0\2\32"+
-    "\1\0\3\32\1\0\7\32\1\0\2\32\1\0\4\32"+
-    "\1\0\14\32\1\0\4\32\1\0\2\32\1\0\21\32"+
-    "\1\0\1\32\2\0\3\32\1\0\1\32\1\0\2\32"+
-    "\1\0\1\32\1\0\1\32\16\0\1\32\3\0\2\32"+
-    "\2\0\2\32\1\0\2\32\1\0\3\32\1\0\1\32"+
-    "\2\0\1\32\1\0\2\32\1\0\2\32\7\0\3\32"+
-    "\2\0\2\32\5\0\4\32\1\0\2\32\1\0\5\32"+
-    "\1\0\5\32\2\0\3\32\2\0\1\32\2\0\3\32"+
-    "\1\0\1\32\6\0\1\32\22\0\1\32\52\0\1\32"+
-    "\1\0\1\32\7\0\1\32\17\0\1\32\6\0\1\32"+
-    "\25\0\1\32\1\0\1\32\1\0\2\32\1\0\3\32"+
-    "\1\0\2\32\1\0\3\32\1\0\7\32\1\0\2\32"+
-    "\1\0\3\32\2\0\11\32\1\0\2\32\1\0\4\32"+
-    "\1\0\2\32\1\0\21\32\1\0\1\32\2\0\3\32"+
-    "\1\0\1\32\1\0\2\32\1\0\1\32\1\0\1\32"+
-    "\16\0\1\32\1\0\1\32\1\0\2\32\1\0\3\32"+
-    "\1\0\2\32\1\0\3\32\1\0\7\32\1\0\2\32"+
-    "\2\0\1\32\1\0\1\32\1\0\14\32\1\0\4\32"+
-    "\1\0\2\32\1\0\21\32\1\0\1\32\2\0\3\32"+
-    "\1\0\1\32\1\0\2\32\1\0\1\32\1\0\1\32"+
-    "\145\0\1\33\15\0\1\34\1\0\1\35\1\0\1\36"+
-    "\2\0\1\37\3\0\1\40\2\0\1\41\3\0\1\42"+
-    "\7\0\1\43\2\0\1\44\4\0\1\45\14\0\1\46"+
-    "\4\0\1\47\2\0\1\50\41\0\2\25\1\51\1\0"+
-    "\1\52\1\0\1\52\1\53\1\0\1\25\3\0\1\54"+
-    "\1\0\1\35\1\0\1\55\2\0\1\56\3\0\1\40"+
-    "\2\0\1\41\3\0\1\57\7\0\1\43\2\0\1\44"+
-    "\4\0\1\60\14\0\1\46\4\0\1\61\2\0\1\62"+
-    "\41\0\1\25\2\26\2\0\2\63\1\64\1\0\1\26"+
-    "\3\0\1\65\6\0\1\66\32\0\1\67\21\0\1\70"+
-    "\2\0\1\71\21\0\1\72\20\0\1\27\1\0\1\27"+
-    "\3\0\1\53\1\0\1\27\3\0\1\73\1\0\1\35"+
-    "\1\0\1\55\2\0\1\74\3\0\1\40\2\0\1\41"+
-    "\3\0\1\57\7\0\1\43\2\0\1\44\4\0\1\75"+
-    "\14\0\1\46\4\0\1\76\2\0\1\77\21\0\1\72"+
-    "\17\0\1\25\1\100\1\26\1\27\3\0\1\100\1\0"+
-    "\1\100\144\0\2\31\4\0\1\25\3\0\1\25\3\0"+
-    "\2\25\5\0\1\25\2\0\1\25\2\0\1\25\4\0"+
-    "\2\25\11\0\1\25\11\0\4\25\4\0\5\25\10\0"+
-    "\1\51\2\25\2\0\1\25\6\0\1\25\31\0\2\25"+
-    "\2\0\2\25\4\0\3\25\4\0\1\25\2\0\1\25"+
-    "\1\0\2\25\10\0\2\25\2\0\2\25\5\0\3\25"+
-    "\5\0\3\25\1\0\1\25\10\0\1\25\2\51\2\0"+
-    "\1\25\6\0\1\25\25\0\1\25\6\0\3\25\2\0"+
-    "\1\25\2\0\2\25\1\0\1\25\1\0\2\25\1\0"+
-    "\2\25\2\0\1\25\2\0\1\25\1\0\1\25\4\0"+
-    "\1\25\2\0\2\25\1\0\1\25\1\0\1\25\1\0"+
-    "\4\25\2\0\1\25\2\0\4\25\1\0\2\25\1\0"+
-    "\1\25\4\0\3\25\1\0\1\25\6\0\1\25\1\0"+
-    "\2\25\24\0\1\51\1\0\2\25\3\0\1\25\4\0"+
-    "\2\25\2\0\7\25\1\0\2\25\3\0\2\25\3\0"+
-    "\6\25\2\0\2\25\1\0\4\25\1\0\2\25\1\0"+
-    "\3\25\1\0\3\25\1\0\6\25\1\0\1\25\1\51"+
-    "\1\0\1\25\2\0\3\25\1\0\1\25\1\0\1\25"+
-    "\2\0\1\25\1\0\1\25\22\0\1\25\13\0\1\25"+
-    "\21\0\14\25\2\0\2\25\5\0\3\25\1\0\1\25"+
-    "\5\0\3\25\5\0\1\25\3\0\2\25\1\0\1\25"+
-    "\2\0\1\25\3\0\1\25\102\0\2\25\6\0\1\25"+
-    "\56\0\1\25\5\0\1\25\2\0\3\25\7\0\1\25"+
-    "\1\0\2\25\7\0\3\25\2\0\2\25\12\0\2\25"+
-    "\1\0\3\25\1\0\1\25\2\0\4\25\2\0\1\25"+
-    "\40\0\1\101\1\0\1\35\1\0\1\36\2\0\1\102"+
-    "\3\0\1\40\2\0\1\41\3\0\1\42\7\0\1\43"+
-    "\2\0\1\44\4\0\1\103\14\0\1\46\4\0\1\104"+
-    "\2\0\1\105\41\0\1\25\2\51\2\0\2\106\1\53"+
-    "\1\0\1\51\3\0\1\107\1\0\1\35\1\0\1\110"+
-    "\2\0\1\111\3\0\1\40\2\0\1\41\3\0\1\112"+
-    "\7\0\1\43\2\0\1\44\4\0\1\113\14\0\1\46"+
-    "\4\0\1\114\2\0\1\115\41\0\1\25\1\52\7\0"+
-    "\1\52\3\0\1\116\1\0\1\35\1\0\1\36\2\0"+
-    "\1\117\3\0\1\40\2\0\1\41\3\0\1\42\7\0"+
-    "\1\43\2\0\1\44\4\0\1\120\14\0\1\46\4\0"+
-    "\1\121\2\0\1\122\21\0\1\72\17\0\1\25\1\53"+
-    "\1\51\1\27\3\0\1\53\1\0\1\53\4\0\1\25"+
-    "\3\0\1\25\3\0\1\25\1\26\5\0\1\25\2\0"+
-    "\1\25\2\0\1\25\4\0\2\26\11\0\1\25\11\0"+
-    "\2\25\2\26\4\0\2\25\3\26\10\0\1\26\2\25"+
-    "\2\0\1\26\6\0\1\25\25\0\1\25\6\0\3\25"+
-    "\2\0\1\25\2\0\2\25\1\0\1\25\1\0\2\25"+
-    "\1\0\2\25\2\0\1\25\2\0\1\25\1\0\1\25"+
-    "\4\0\1\25\2\0\2\25\1\0\1\25\1\0\1\25"+
-    "\1\0\4\25\2\0\1\25\2\0\4\25\1\0\1\26"+
-    "\1\25\1\0\1\25\4\0\3\25\1\0\1\25\6\0"+
-    "\1\25\1\0\2\25\26\0\1\25\13\0\1\25\21\0"+
-    "\14\25\2\0\2\25\5\0\3\25\1\0\1\25\5\0"+
-    "\1\25\2\26\5\0\1\25\3\0\1\25\1\26\1\0"+
-    "\1\25\2\0\1\25\3\0\1\25\102\0\2\26\6\0"+
-    "\1\26\56\0\1\26\5\0\1\26\2\0\3\26\7\0"+
-    "\1\26\1\0\2\26\7\0\3\26\2\0\2\26\12\0"+
-    "\2\26\1\0\3\26\1\0\1\26\2\0\4\26\2\0"+
-    "\1\26\40\0\1\123\3\0\1\124\2\0\1\125\12\0"+
-    "\1\126\17\0\1\127\21\0\1\130\2\0\1\131\42\0"+
-    "\1\63\1\26\6\0\1\63\3\0\1\132\1\0\1\35"+
-    "\1\0\1\55\2\0\1\133\3\0\1\40\2\0\1\41"+
-    "\3\0\1\57\7\0\1\43\2\0\1\44\4\0\1\134"+
-    "\14\0\1\46\4\0\1\135\2\0\1\136\21\0\1\72"+
-    "\17\0\1\25\1\64\1\26\1\27\3\0\1\64\1\0"+
-    "\1\64\15\0\1\27\20\0\2\27\25\0\2\27\6\0"+
-    "\3\27\15\0\1\27\131\0\1\27\152\0\2\27\12\0"+
-    "\1\27\113\0\2\27\6\0\1\27\56\0\1\27\5\0"+
-    "\1\27\2\0\3\27\7\0\1\27\1\0\2\27\7\0"+
-    "\3\27\2\0\2\27\12\0\2\27\1\0\3\27\1\0"+
-    "\1\27\2\0\4\27\2\0\1\27\152\0\1\27\34\0"+
-    "\1\25\3\0\1\25\3\0\1\25\1\100\5\0\1\25"+
-    "\2\0\1\25\2\0\1\25\4\0\2\100\11\0\1\25"+
-    "\11\0\2\25\2\100\4\0\2\25\3\100\10\0\1\26"+
-    "\2\25\2\0\1\100\6\0\1\25\25\0\1\25\6\0"+
-    "\3\25\2\0\1\25\2\0\2\25\1\0\1\25\1\0"+
-    "\2\25\1\0\2\25\2\0\1\25\2\0\1\25\1\0"+
-    "\1\25\4\0\1\25\2\0\2\25\1\0\1\25\1\0"+
-    "\1\25\1\0\4\25\2\0\1\25\2\0\4\25\1\0"+
-    "\1\100\1\25\1\0\1\25\4\0\3\25\1\0\1\25"+
-    "\6\0\1\25\1\0\2\25\26\0\1\25\13\0\1\25"+
-    "\21\0\14\25\2\0\2\25\5\0\3\25\1\0\1\25"+
-    "\5\0\1\25\2\100\5\0\1\25\3\0\1\25\1\100"+
-    "\1\0\1\25\2\0\1\25\3\0\1\25\102\0\2\100"+
-    "\6\0\1\100\56\0\1\100\5\0\1\100\2\0\3\100"+
-    "\7\0\1\100\1\0\2\100\7\0\3\100\2\0\2\100"+
-    "\12\0\2\100\1\0\3\100\1\0\1\100\2\0\4\100"+
-    "\2\0\1\100\41\0\1\25\3\0\1\25\3\0\1\25"+
-    "\1\51\5\0\1\25\2\0\1\25\2\0\1\25\4\0"+
-    "\2\51\11\0\1\25\11\0\2\25\2\51\4\0\2\25"+
-    "\3\51\10\0\1\51\2\25\2\0\1\51\6\0\1\25"+
-    "\25\0\1\25\6\0\3\25\2\0\1\25\2\0\2\25"+
-    "\1\0\1\25\1\0\2\25\1\0\2\25\2\0\1\25"+
-    "\2\0\1\25\1\0\1\25\4\0\1\25\2\0\2\25"+
-    "\1\0\1\25\1\0\1\25\1\0\4\25\2\0\1\25"+
-    "\2\0\4\25\1\0\1\51\1\25\1\0\1\25\4\0"+
-    "\3\25\1\0\1\25\6\0\1\25\1\0\2\25\26\0"+
-    "\1\25\13\0\1\25\21\0\14\25\2\0\2\25\5\0"+
-    "\3\25\1\0\1\25\5\0\1\25\2\51\5\0\1\25"+
-    "\3\0\1\25\1\51\1\0\1\25\2\0\1\25\3\0"+
-    "\1\25\102\0\2\51\6\0\1\51\56\0\1\51\5\0"+
-    "\1\51\2\0\3\51\7\0\1\51\1\0\2\51\7\0"+
-    "\3\51\2\0\2\51\12\0\2\51\1\0\3\51\1\0"+
-    "\1\51\2\0\4\51\2\0\1\51\40\0\1\137\3\0"+
-    "\1\140\2\0\1\141\12\0\1\142\17\0\1\143\21\0"+
-    "\1\144\2\0\1\145\42\0\1\106\1\51\6\0\1\106"+
-    "\4\0\1\25\3\0\1\25\3\0\1\25\1\52\5\0"+
-    "\1\25\2\0\1\25\2\0\1\25\4\0\2\52\11\0"+
-    "\1\25\11\0\2\25\2\52\4\0\2\25\3\52\11\0"+
-    "\2\25\2\0\1\52\6\0\1\25\31\0\2\25\2\0"+
-    "\2\25\4\0\3\25\4\0\1\25\2\0\1\25\1\0"+
-    "\2\25\10\0\2\25\2\0\2\25\5\0\3\25\5\0"+
-    "\3\25\1\0\1\25\10\0\1\25\4\0\1\25\6\0"+
-    "\1\25\25\0\1\25\6\0\3\25\2\0\1\25\2\0"+
-    "\2\25\1\0\1\25\1\0\2\25\1\0\2\25\2\0"+
-    "\1\25\2\0\1\25\1\0\1\25\4\0\1\25\2\0"+
-    "\2\25\1\0\1\25\1\0\1\25\1\0\4\25\2\0"+
-    "\1\25\2\0\4\25\1\0\1\52\1\25\1\0\1\25"+
-    "\4\0\3\25\1\0\1\25\6\0\1\25\1\0\2\25"+
-    "\26\0\2\25\3\0\1\25\4\0\2\25\2\0\7\25"+
-    "\1\0\2\25\3\0\2\25\3\0\6\25\2\0\2\25"+
-    "\1\0\4\25\1\0\2\25\1\0\3\25\1\0\3\25"+
-    "\1\0\6\25\1\0\1\25\2\0\1\25\2\0\3\25"+
-    "\1\0\1\25\1\0\1\25\2\0\1\25\1\0\1\25"+
-    "\22\0\1\25\13\0\1\25\21\0\14\25\2\0\2\25"+
-    "\5\0\3\25\1\0\1\25\5\0\1\25\2\52\5\0"+
-    "\1\25\3\0\1\25\1\52\1\0\1\25\2\0\1\25"+
-    "\3\0\1\25\102\0\2\52\6\0\1\52\56\0\1\52"+
-    "\5\0\1\52\2\0\3\52\7\0\1\52\1\0\2\52"+
-    "\7\0\3\52\2\0\2\52\12\0\2\52\1\0\3\52"+
-    "\1\0\1\52\2\0\4\52\2\0\1\52\41\0\1\25"+
-    "\3\0\1\25\3\0\1\25\1\53\5\0\1\25\2\0"+
-    "\1\25\2\0\1\25\4\0\2\53\11\0\1\25\11\0"+
-    "\2\25\2\53\4\0\2\25\3\53\10\0\1\51\2\25"+
-    "\2\0\1\53\6\0\1\25\25\0\1\25\6\0\3\25"+
-    "\2\0\1\25\2\0\2\25\1\0\1\25\1\0\2\25"+
-    "\1\0\2\25\2\0\1\25\2\0\1\25\1\0\1\25"+
-    "\4\0\1\25\2\0\2\25\1\0\1\25\1\0\1\25"+
-    "\1\0\4\25\2\0\1\25\2\0\4\25\1\0\1\53"+
-    "\1\25\1\0\1\25\4\0\3\25\1\0\1\25\6\0"+
-    "\1\25\1\0\2\25\26\0\1\25\13\0\1\25\21\0"+
-    "\14\25\2\0\2\25\5\0\3\25\1\0\1\25\5\0"+
-    "\1\25\2\53\5\0\1\25\3\0\1\25\1\53\1\0"+
-    "\1\25\2\0\1\25\3\0\1\25\102\0\2\53\6\0"+
-    "\1\53\56\0\1\53\5\0\1\53\2\0\3\53\7\0"+
-    "\1\53\1\0\2\53\7\0\3\53\2\0\2\53\12\0"+
+    "\1\2\1\3\1\2\1\4\1\2\1\5\1\2\1\6"+
+    "\1\2\1\7\1\2\1\10\3\2\1\11\5\2\1\12"+
+    "\3\2\1\13\11\2\1\14\2\2\1\15\43\2\1\16"+
+    "\1\2\1\17\3\2\1\20\1\21\1\2\1\22\1\2"+
+    "\1\23\2\2\1\24\1\2\1\25\1\2\1\26\1\27"+
+    "\3\2\1\30\2\31\1\32\1\33\1\34\151\0\1\25"+
+    "\11\0\1\25\20\0\1\25\22\0\1\25\10\0\3\25"+
+    "\17\0\1\25\10\0\1\25\24\0\1\25\1\0\1\25"+
+    "\1\0\1\25\1\0\1\25\1\0\1\25\1\0\3\25"+
+    "\1\0\5\25\1\0\3\25\1\0\11\25\1\0\2\25"+
+    "\1\0\16\25\1\0\2\25\1\0\21\25\1\0\1\25"+
+    "\1\0\3\25\2\0\1\25\1\0\1\25\1\0\2\25"+
+    "\1\0\1\25\17\0\1\25\3\0\1\25\5\0\2\25"+
+    "\3\0\1\25\13\0\1\25\1\0\1\25\4\0\2\25"+
+    "\4\0\1\25\1\0\1\25\3\0\2\25\1\0\1\25"+
+    "\5\0\3\25\1\0\1\25\15\0\1\25\10\0\1\25"+
+    "\24\0\1\25\3\0\1\25\1\0\1\25\1\0\1\25"+
+    "\1\0\3\25\2\0\4\25\1\0\3\25\2\0\3\25"+
+    "\1\0\4\25\1\0\2\25\2\0\3\25\1\0\11\25"+
+    "\1\0\2\25\1\0\16\25\1\0\2\25\1\0\1\25"+
+    "\1\0\3\25\2\0\1\25\1\0\1\25\1\0\2\25"+
+    "\1\0\1\25\17\0\1\25\3\0\1\25\3\0\1\25"+
+    "\1\0\3\25\2\0\1\25\1\0\2\25\1\0\3\25"+
+    "\3\0\2\25\1\0\1\25\1\0\2\25\1\0\2\25"+
+    "\3\0\2\25\1\0\1\25\1\0\1\25\1\0\2\25"+
+    "\1\0\2\25\1\0\2\25\1\0\5\25\1\0\5\25"+
+    "\1\0\2\25\1\0\2\25\1\0\1\25\1\0\3\25"+
+    "\4\0\1\25\4\0\1\25\31\0\3\25\5\0\1\25"+
+    "\1\0\1\25\1\0\1\25\4\0\1\25\14\0\1\25"+
+    "\5\0\1\25\11\0\2\25\12\0\1\26\1\0\2\25"+
+    "\12\0\1\25\24\0\1\25\1\0\1\26\7\0\2\25"+
+    "\2\0\5\25\2\0\2\25\4\0\6\25\1\0\2\25"+
+    "\4\0\5\25\1\0\5\25\1\0\2\25\1\0\3\25"+
+    "\1\0\4\25\1\0\5\25\1\26\1\0\1\25\1\0"+
+    "\1\25\1\0\3\25\2\0\1\25\1\0\1\25\1\0"+
+    "\1\25\2\0\1\25\17\0\1\25\3\0\1\25\5\0"+
+    "\2\25\3\0\1\25\4\0\3\25\4\0\1\25\1\0"+
+    "\1\25\2\0\1\25\1\0\2\25\4\0\1\25\1\0"+
+    "\1\25\3\0\2\25\1\0\1\25\5\0\3\25\1\0"+
+    "\1\25\10\0\1\25\1\0\2\26\1\0\1\25\10\0"+
+    "\1\25\24\0\1\25\3\0\1\25\6\0\2\25\5\0"+
+    "\1\25\1\0\1\25\1\0\1\25\1\0\11\25\2\0"+
+    "\1\25\4\0\1\25\4\0\6\25\2\0\1\25\1\0"+
+    "\1\25\1\0\3\25\3\0\2\25\4\0\3\25\1\0"+
+    "\1\25\10\0\1\25\1\0\2\25\21\0\1\25\11\0"+
+    "\2\25\17\0\1\25\6\0\2\25\4\0\1\25\5\0"+
+    "\1\25\2\0\1\25\5\0\3\25\1\0\1\25\15\0"+
+    "\1\25\10\0\1\25\24\0\1\25\3\0\1\25\5\0"+
+    "\1\25\32\0\15\25\5\0\3\25\1\0\1\25\5\0"+
+    "\1\25\7\0\1\25\2\0\1\25\5\0\1\25\2\0"+
+    "\1\25\1\0\1\25\106\0\1\33\21\0\1\27\35\0"+
+    "\1\32\3\0\1\32\3\0\1\32\1\0\3\32\2\0"+
+    "\1\32\2\0\1\32\1\0\3\32\3\0\2\32\1\0"+
+    "\1\32\1\0\2\32\1\0\2\32\3\0\2\32\1\0"+
+    "\1\32\3\0\2\32\1\0\2\32\1\0\2\32\1\0"+
+    "\5\32\1\0\5\32\2\0\1\32\1\0\2\32\1\0"+
+    "\1\32\1\0\3\32\4\0\1\32\4\0\1\32\17\0"+
+    "\1\32\1\0\1\32\1\0\1\32\1\0\1\32\1\0"+
+    "\1\32\1\0\3\32\1\0\5\32\1\0\3\32\1\0"+
+    "\11\32\1\0\2\32\1\0\16\32\1\0\2\32\1\0"+
+    "\21\32\1\0\1\32\1\0\3\32\2\0\1\32\1\0"+
+    "\1\32\1\0\2\32\1\0\1\32\17\0\1\32\1\0"+
+    "\1\32\1\0\1\32\3\0\1\32\1\0\3\32\1\0"+
+    "\2\32\1\0\2\32\1\0\3\32\1\0\11\32\1\0"+
+    "\2\32\1\0\16\32\1\0\2\32\1\0\21\32\1\0"+
+    "\1\32\1\0\3\32\2\0\1\32\1\0\1\32\1\0"+
+    "\2\32\1\0\1\32\17\0\1\32\11\0\1\32\20\0"+
+    "\1\32\33\0\1\32\21\0\1\32\10\0\1\32\24\0"+
+    "\1\32\1\0\1\32\1\0\1\32\1\0\1\32\1\0"+
+    "\1\32\1\0\3\32\1\0\5\32\1\0\3\32\1\0"+
+    "\6\32\1\0\2\32\1\0\2\32\1\0\10\32\1\0"+
+    "\5\32\1\0\2\32\1\0\21\32\1\0\1\32\1\0"+
+    "\3\32\2\0\1\32\1\0\1\32\1\0\2\32\1\0"+
+    "\1\32\146\0\1\33\16\0\1\35\1\0\1\36\1\0"+
+    "\1\37\1\0\1\40\1\0\1\41\1\0\1\42\3\0"+
+    "\1\43\5\0\1\44\3\0\1\45\11\0\1\46\2\0"+
+    "\1\47\16\0\1\50\2\0\1\51\41\0\2\25\1\52"+
+    "\1\0\1\53\1\0\1\53\1\54\1\0\1\25\2\0"+
+    "\1\25\1\0\1\35\1\0\1\36\1\0\1\37\1\0"+
+    "\1\40\1\0\1\41\1\0\1\55\3\0\1\56\5\0"+
+    "\1\57\3\0\1\60\11\0\1\46\2\0\1\61\16\0"+
+    "\1\62\2\0\1\63\41\0\1\25\2\26\2\0\2\64"+
+    "\1\65\1\0\1\26\2\0\1\25\13\0\1\66\15\0"+
+    "\1\67\14\0\1\70\16\0\1\71\2\0\1\72\21\0"+
+    "\1\73\20\0\1\27\1\0\1\27\3\0\1\54\1\0"+
+    "\1\27\4\0\1\35\1\0\1\36\1\0\1\37\1\0"+
+    "\1\40\1\0\1\41\1\0\1\74\3\0\1\56\5\0"+
+    "\1\57\3\0\1\75\11\0\1\46\2\0\1\76\16\0"+
+    "\1\77\2\0\1\100\21\0\1\101\17\0\1\25\1\102"+
+    "\1\26\1\103\3\0\1\102\1\0\1\102\2\0\1\25"+
+    "\142\0\2\31\4\0\1\35\1\0\1\36\1\0\1\37"+
+    "\1\0\1\40\1\0\1\41\1\0\1\104\3\0\1\43"+
+    "\5\0\1\44\3\0\1\105\11\0\1\46\2\0\1\106"+
+    "\16\0\1\107\2\0\1\110\41\0\1\25\1\34\1\52"+
+    "\1\0\1\53\1\0\1\53\1\54\1\0\1\34\2\0"+
+    "\1\34\2\0\1\25\11\0\3\25\5\0\1\25\1\0"+
+    "\1\25\1\0\1\25\4\0\1\25\4\0\1\25\1\0"+
+    "\2\25\4\0\1\25\5\0\1\25\3\0\1\25\4\0"+
+    "\5\25\10\0\1\52\1\0\2\25\1\0\1\25\10\0"+
+    "\1\25\24\0\1\25\1\0\1\52\7\0\2\25\2\0"+
+    "\5\25\2\0\2\25\4\0\6\25\1\0\2\25\4\0"+
+    "\5\25\1\0\5\25\1\0\2\25\1\0\3\25\1\0"+
+    "\4\25\1\0\5\25\1\52\1\0\1\25\1\0\1\25"+
+    "\1\0\3\25\2\0\1\25\1\0\1\25\1\0\1\25"+
+    "\2\0\1\25\17\0\1\25\3\0\1\25\5\0\2\25"+
+    "\3\0\1\25\4\0\3\25\4\0\1\25\1\0\1\25"+
+    "\2\0\1\25\1\0\2\25\4\0\1\25\1\0\1\25"+
+    "\3\0\2\25\1\0\1\25\5\0\3\25\1\0\1\25"+
+    "\10\0\1\25\1\0\2\52\1\0\1\25\10\0\1\25"+
+    "\24\0\1\25\3\0\1\25\6\0\2\25\5\0\1\25"+
+    "\1\0\1\25\1\0\1\25\1\0\11\25\2\0\1\25"+
+    "\4\0\1\25\4\0\6\25\2\0\1\25\1\0\1\25"+
+    "\1\0\3\25\1\0\1\25\1\0\2\25\4\0\3\25"+
+    "\1\0\1\25\10\0\1\25\1\0\2\25\21\0\1\25"+
+    "\3\0\1\25\5\0\1\25\32\0\15\25\5\0\3\25"+
+    "\1\0\1\25\5\0\3\25\5\0\1\25\2\0\2\25"+
+    "\4\0\1\25\2\0\1\25\1\0\1\25\103\0\2\25"+
+    "\6\0\1\25\56\0\1\25\3\0\1\25\2\0\1\25"+
+    "\3\0\1\25\5\0\1\25\7\0\1\25\4\0\2\25"+
+    "\3\0\2\25\1\0\1\25\4\0\1\25\1\0\1\25"+
+    "\2\0\2\25\1\0\3\25\1\0\1\25\2\0\4\25"+
+    "\2\0\1\25\41\0\1\35\1\0\1\36\1\0\1\37"+
+    "\1\0\1\40\1\0\1\41\1\0\1\111\3\0\1\43"+
+    "\5\0\1\44\3\0\1\112\11\0\1\46\2\0\1\113"+
+    "\16\0\1\114\2\0\1\115\41\0\1\25\2\52\2\0"+
+    "\2\116\1\54\1\0\1\52\2\0\1\25\1\0\1\35"+
+    "\1\0\1\36\1\0\1\37\1\0\1\40\1\0\1\41"+
+    "\1\0\1\117\3\0\1\120\5\0\1\121\3\0\1\122"+
+    "\11\0\1\46\2\0\1\123\16\0\1\124\2\0\1\125"+
+    "\41\0\1\25\1\53\7\0\1\53\2\0\1\25\1\0"+
+    "\1\35\1\0\1\36\1\0\1\37\1\0\1\40\1\0"+
+    "\1\41\1\0\1\126\3\0\1\43\5\0\1\44\3\0"+
+    "\1\127\11\0\1\46\2\0\1\130\16\0\1\131\2\0"+
+    "\1\132\21\0\1\101\17\0\1\25\1\54\1\52\1\103"+
+    "\3\0\1\54\1\0\1\54\2\0\1\25\2\0\1\26"+
+    "\11\0\3\25\5\0\1\25\1\0\1\25\1\0\1\25"+
+    "\4\0\1\25\4\0\1\26\1\0\2\26\4\0\1\25"+
+    "\5\0\1\25\3\0\1\26\4\0\1\26\2\25\2\26"+
+    "\10\0\1\26\1\0\2\25\1\0\1\26\10\0\1\25"+
+    "\24\0\1\25\3\0\1\25\6\0\2\25\5\0\1\25"+
+    "\1\0\1\25\1\0\1\25\1\0\11\25\2\0\1\25"+
+    "\4\0\1\25\4\0\6\25\2\0\1\25\1\0\1\25"+
+    "\1\0\3\25\1\0\1\26\1\0\2\25\4\0\3\25"+
+    "\1\0\1\25\10\0\1\25\1\0\2\25\21\0\1\25"+
+    "\3\0\1\25\5\0\1\25\32\0\15\25\5\0\3\25"+
+    "\1\0\1\25\5\0\1\25\2\26\5\0\1\25\2\0"+
+    "\1\25\1\26\4\0\1\25\2\0\1\25\1\0\1\25"+
+    "\103\0\2\26\6\0\1\26\56\0\1\26\3\0\1\26"+
+    "\2\0\1\26\3\0\1\26\5\0\1\26\7\0\1\26"+
+    "\4\0\2\26\3\0\2\26\1\0\1\26\4\0\1\26"+
+    "\1\0\1\26\2\0\2\26\1\0\3\26\1\0\1\26"+
+    "\2\0\4\26\2\0\1\26\53\0\1\133\3\0\1\134"+
+    "\5\0\1\135\3\0\1\136\14\0\1\137\16\0\1\140"+
+    "\2\0\1\141\42\0\1\64\1\26\6\0\1\64\4\0"+
+    "\1\35\1\0\1\36\1\0\1\37\1\0\1\40\1\0"+
+    "\1\41\1\0\1\142\3\0\1\56\5\0\1\57\3\0"+
+    "\1\143\11\0\1\46\2\0\1\144\16\0\1\145\2\0"+
+    "\1\146\21\0\1\101\17\0\1\25\1\65\1\26\1\103"+
+    "\3\0\1\65\1\0\1\65\2\0\1\25\2\0\1\27"+
+    "\37\0\1\27\1\0\2\27\16\0\1\27\4\0\1\27"+
+    "\2\0\2\27\15\0\1\27\132\0\1\27\153\0\2\27"+
+    "\11\0\1\27\115\0\2\27\6\0\1\27\56\0\1\27"+
+    "\3\0\1\27\2\0\1\27\3\0\1\27\5\0\1\27"+
+    "\7\0\1\27\4\0\2\27\3\0\2\27\1\0\1\27"+
+    "\4\0\1\27\1\0\1\27\2\0\2\27\1\0\3\27"+
+    "\1\0\1\27\2\0\4\27\2\0\1\27\153\0\1\27"+
+    "\35\0\1\102\11\0\3\25\5\0\1\25\1\0\1\25"+
+    "\1\0\1\25\4\0\1\25\4\0\1\102\1\0\2\102"+
+    "\4\0\1\25\5\0\1\25\3\0\1\102\4\0\1\102"+
+    "\2\25\2\102\10\0\1\26\1\0\2\25\1\0\1\102"+
+    "\10\0\1\25\24\0\1\25\3\0\1\25\6\0\2\25"+
+    "\5\0\1\25\1\0\1\25\1\0\1\25\1\0\11\25"+
+    "\2\0\1\25\4\0\1\25\4\0\6\25\2\0\1\25"+
+    "\1\0\1\25\1\0\3\25\1\0\1\102\1\0\2\25"+
+    "\4\0\3\25\1\0\1\25\10\0\1\25\1\0\2\25"+
+    "\21\0\1\25\3\0\1\25\5\0\1\25\32\0\15\25"+
+    "\5\0\3\25\1\0\1\25\5\0\1\25\2\102\5\0"+
+    "\1\25\2\0\1\25\1\102\4\0\1\25\2\0\1\25"+
+    "\1\0\1\25\103\0\2\102\6\0\1\102\56\0\1\102"+
+    "\3\0\1\102\2\0\1\102\3\0\1\102\5\0\1\102"+
+    "\7\0\1\102\4\0\2\102\3\0\2\102\1\0\1\102"+
+    "\4\0\1\102\1\0\1\102\2\0\2\102\1\0\3\102"+
+    "\1\0\1\102\2\0\4\102\2\0\1\102\153\0\1\103"+
+    "\46\0\1\147\15\0\1\150\14\0\1\151\16\0\1\152"+
+    "\2\0\1\153\21\0\1\101\20\0\1\103\1\0\1\103"+
+    "\3\0\1\54\1\0\1\103\5\0\1\34\11\0\3\25"+
+    "\5\0\1\25\1\0\1\25\1\0\1\25\4\0\1\25"+
+    "\4\0\1\34\1\0\2\34\4\0\1\25\5\0\1\25"+
+    "\3\0\1\34\4\0\1\34\2\25\2\34\10\0\1\52"+
+    "\1\0\2\25\1\0\1\34\10\0\1\25\24\0\1\25"+
+    "\3\0\1\25\6\0\2\25\5\0\1\25\1\0\1\25"+
+    "\1\0\1\25\1\0\11\25\2\0\1\25\4\0\1\25"+
+    "\4\0\6\25\2\0\1\25\1\0\1\25\1\0\3\25"+
+    "\1\0\1\34\1\0\2\25\4\0\3\25\1\0\1\25"+
+    "\10\0\1\25\1\0\2\25\21\0\1\25\3\0\1\25"+
+    "\5\0\1\25\32\0\15\25\5\0\3\25\1\0\1\25"+
+    "\5\0\1\25\2\34\5\0\1\25\2\0\1\25\1\34"+
+    "\4\0\1\25\2\0\1\25\1\0\1\25\103\0\2\34"+
+    "\6\0\1\34\56\0\1\34\3\0\1\34\2\0\1\34"+
+    "\3\0\1\34\5\0\1\34\7\0\1\34\4\0\2\34"+
+    "\3\0\2\34\1\0\1\34\4\0\1\34\1\0\1\34"+
+    "\2\0\2\34\1\0\3\34\1\0\1\34\2\0\4\34"+
+    "\2\0\1\34\42\0\1\52\11\0\3\25\5\0\1\25"+
+    "\1\0\1\25\1\0\1\25\4\0\1\25\4\0\1\52"+
+    "\1\0\2\52\4\0\1\25\5\0\1\25\3\0\1\52"+
+    "\4\0\1\52\2\25\2\52\10\0\1\52\1\0\2\25"+
+    "\1\0\1\52\10\0\1\25\24\0\1\25\3\0\1\25"+
+    "\6\0\2\25\5\0\1\25\1\0\1\25\1\0\1\25"+
+    "\1\0\11\25\2\0\1\25\4\0\1\25\4\0\6\25"+
+    "\2\0\1\25\1\0\1\25\1\0\3\25\1\0\1\52"+
+    "\1\0\2\25\4\0\3\25\1\0\1\25\10\0\1\25"+
+    "\1\0\2\25\21\0\1\25\3\0\1\25\5\0\1\25"+
+    "\32\0\15\25\5\0\3\25\1\0\1\25\5\0\1\25"+
+    "\2\52\5\0\1\25\2\0\1\25\1\52\4\0\1\25"+
+    "\2\0\1\25\1\0\1\25\103\0\2\52\6\0\1\52"+
+    "\56\0\1\52\3\0\1\52\2\0\1\52\3\0\1\52"+
+    "\5\0\1\52\7\0\1\52\4\0\2\52\3\0\2\52"+
+    "\1\0\1\52\4\0\1\52\1\0\1\52\2\0\2\52"+
+    "\1\0\3\52\1\0\1\52\2\0\4\52\2\0\1\52"+
+    "\53\0\1\154\3\0\1\155\5\0\1\156\3\0\1\157"+
+    "\14\0\1\160\16\0\1\161\2\0\1\162\42\0\1\116"+
+    "\1\52\6\0\1\116\5\0\1\53\11\0\3\25\5\0"+
+    "\1\25\1\0\1\25\1\0\1\25\4\0\1\25\4\0"+
+    "\1\53\1\0\2\53\4\0\1\25\5\0\1\25\3\0"+
+    "\1\53\4\0\1\53\2\25\2\53\12\0\2\25\1\0"+
+    "\1\53\10\0\1\25\24\0\1\25\11\0\2\25\2\0"+
+    "\5\25\2\0\2\25\4\0\6\25\1\0\2\25\4\0"+
+    "\5\25\1\0\5\25\1\0\2\25\1\0\3\25\1\0"+
+    "\4\25\1\0\5\25\2\0\1\25\1\0\1\25\1\0"+
+    "\3\25\2\0\1\25\1\0\1\25\1\0\1\25\2\0"+
+    "\1\25\17\0\1\25\3\0\1\25\5\0\2\25\3\0"+
+    "\1\25\4\0\3\25\4\0\1\25\1\0\1\25\2\0"+
+    "\1\25\1\0\2\25\4\0\1\25\1\0\1\25\3\0"+
+    "\2\25\1\0\1\25\5\0\3\25\1\0\1\25\10\0"+
+    "\1\25\4\0\1\25\10\0\1\25\24\0\1\25\3\0"+
+    "\1\25\6\0\2\25\5\0\1\25\1\0\1\25\1\0"+
+    "\1\25\1\0\11\25\2\0\1\25\4\0\1\25\4\0"+
+    "\6\25\2\0\1\25\1\0\1\25\1\0\3\25\1\0"+
+    "\1\53\1\0\2\25\4\0\3\25\1\0\1\25\10\0"+
+    "\1\25\1\0\2\25\21\0\1\25\3\0\1\25\5\0"+
+    "\1\25\32\0\15\25\5\0\3\25\1\0\1\25\5\0"+
+    "\1\25\2\53\5\0\1\25\2\0\1\25\1\53\4\0"+
+    "\1\25\2\0\1\25\1\0\1\25\103\0\2\53\6\0"+
+    "\1\53\56\0\1\53\3\0\1\53\2\0\1\53\3\0"+
+    "\1\53\5\0\1\53\7\0\1\53\4\0\2\53\3\0"+
+    "\2\53\1\0\1\53\4\0\1\53\1\0\1\53\2\0"+
     "\2\53\1\0\3\53\1\0\1\53\2\0\4\53\2\0"+
-    "\1\53\52\0\1\63\20\0\2\63\25\0\2\63\6\0"+
-    "\3\63\10\0\1\26\4\0\1\63\141\0\2\26\134\0"+
-    "\1\63\52\0\1\26\104\0\1\26\140\0\2\63\12\0"+
-    "\1\63\113\0\2\63\6\0\1\63\56\0\1\63\5\0"+
-    "\1\63\2\0\3\63\7\0\1\63\1\0\2\63\7\0"+
-    "\3\63\2\0\2\63\12\0\2\63\1\0\3\63\1\0"+
-    "\1\63\2\0\4\63\2\0\1\63\41\0\1\25\3\0"+
-    "\1\25\3\0\1\25\1\64\5\0\1\25\2\0\1\25"+
-    "\2\0\1\25\4\0\2\64\11\0\1\25\11\0\2\25"+
-    "\2\64\4\0\2\25\3\64\10\0\1\26\2\25\2\0"+
-    "\1\64\6\0\1\25\25\0\1\25\6\0\3\25\2\0"+
-    "\1\25\2\0\2\25\1\0\1\25\1\0\2\25\1\0"+
-    "\2\25\2\0\1\25\2\0\1\25\1\0\1\25\4\0"+
-    "\1\25\2\0\2\25\1\0\1\25\1\0\1\25\1\0"+
-    "\4\25\2\0\1\25\2\0\4\25\1\0\1\64\1\25"+
-    "\1\0\1\25\4\0\3\25\1\0\1\25\6\0\1\25"+
-    "\1\0\2\25\26\0\1\25\13\0\1\25\21\0\14\25"+
-    "\2\0\2\25\5\0\3\25\1\0\1\25\5\0\1\25"+
-    "\2\64\5\0\1\25\3\0\1\25\1\64\1\0\1\25"+
-    "\2\0\1\25\3\0\1\25\102\0\2\64\6\0\1\64"+
-    "\56\0\1\64\5\0\1\64\2\0\3\64\7\0\1\64"+
-    "\1\0\2\64\7\0\3\64\2\0\2\64\12\0\2\64"+
-    "\1\0\3\64\1\0\1\64\2\0\4\64\2\0\1\64"+
-    "\52\0\1\106\20\0\2\106\25\0\2\106\6\0\3\106"+
-    "\10\0\1\51\4\0\1\106\141\0\2\51\134\0\1\106"+
-    "\52\0\1\51\104\0\1\51\140\0\2\106\12\0\1\106"+
-    "\113\0\2\106\6\0\1\106\56\0\1\106\5\0\1\106"+
-    "\2\0\3\106\7\0\1\106\1\0\2\106\7\0\3\106"+
-    "\2\0\2\106\12\0\2\106\1\0\3\106\1\0\1\106"+
-    "\2\0\4\106\2\0\1\106\37\0";
+    "\1\53\42\0\1\54\11\0\3\25\5\0\1\25\1\0"+
+    "\1\25\1\0\1\25\4\0\1\25\4\0\1\54\1\0"+
+    "\2\54\4\0\1\25\5\0\1\25\3\0\1\54\4\0"+
+    "\1\54\2\25\2\54\10\0\1\52\1\0\2\25\1\0"+
+    "\1\54\10\0\1\25\24\0\1\25\3\0\1\25\6\0"+
+    "\2\25\5\0\1\25\1\0\1\25\1\0\1\25\1\0"+
+    "\11\25\2\0\1\25\4\0\1\25\4\0\6\25\2\0"+
+    "\1\25\1\0\1\25\1\0\3\25\1\0\1\54\1\0"+
+    "\2\25\4\0\3\25\1\0\1\25\10\0\1\25\1\0"+
+    "\2\25\21\0\1\25\3\0\1\25\5\0\1\25\32\0"+
+    "\15\25\5\0\3\25\1\0\1\25\5\0\1\25\2\54"+
+    "\5\0\1\25\2\0\1\25\1\54\4\0\1\25\2\0"+
+    "\1\25\1\0\1\25\103\0\2\54\6\0\1\54\56\0"+
+    "\1\54\3\0\1\54\2\0\1\54\3\0\1\54\5\0"+
+    "\1\54\7\0\1\54\4\0\2\54\3\0\2\54\1\0"+
+    "\1\54\4\0\1\54\1\0\1\54\2\0\2\54\1\0"+
+    "\3\54\1\0\1\54\2\0\4\54\2\0\1\54\42\0"+
+    "\1\64\37\0\1\64\1\0\2\64\16\0\1\64\4\0"+
+    "\1\64\2\0\2\64\10\0\1\26\4\0\1\64\37\0"+
+    "\1\26\102\0\1\26\147\0\2\26\134\0\1\64\153\0"+
+    "\2\64\11\0\1\64\115\0\2\64\6\0\1\64\56\0"+
+    "\1\64\3\0\1\64\2\0\1\64\3\0\1\64\5\0"+
+    "\1\64\7\0\1\64\4\0\2\64\3\0\2\64\1\0"+
+    "\1\64\4\0\1\64\1\0\1\64\2\0\2\64\1\0"+
+    "\3\64\1\0\1\64\2\0\4\64\2\0\1\64\42\0"+
+    "\1\65\11\0\3\25\5\0\1\25\1\0\1\25\1\0"+
+    "\1\25\4\0\1\25\4\0\1\65\1\0\2\65\4\0"+
+    "\1\25\5\0\1\25\3\0\1\65\4\0\1\65\2\25"+
+    "\2\65\10\0\1\26\1\0\2\25\1\0\1\65\10\0"+
+    "\1\25\24\0\1\25\3\0\1\25\6\0\2\25\5\0"+
+    "\1\25\1\0\1\25\1\0\1\25\1\0\11\25\2\0"+
+    "\1\25\4\0\1\25\4\0\6\25\2\0\1\25\1\0"+
+    "\1\25\1\0\3\25\1\0\1\65\1\0\2\25\4\0"+
+    "\3\25\1\0\1\25\10\0\1\25\1\0\2\25\21\0"+
+    "\1\25\3\0\1\25\5\0\1\25\32\0\15\25\5\0"+
+    "\3\25\1\0\1\25\5\0\1\25\2\65\5\0\1\25"+
+    "\2\0\1\25\1\65\4\0\1\25\2\0\1\25\1\0"+
+    "\1\25\103\0\2\65\6\0\1\65\56\0\1\65\3\0"+
+    "\1\65\2\0\1\65\3\0\1\65\5\0\1\65\7\0"+
+    "\1\65\4\0\2\65\3\0\2\65\1\0\1\65\4\0"+
+    "\1\65\1\0\1\65\2\0\2\65\1\0\3\65\1\0"+
+    "\1\65\2\0\4\65\2\0\1\65\42\0\1\103\37\0"+
+    "\1\103\1\0\2\103\16\0\1\103\4\0\1\103\2\0"+
+    "\2\103\15\0\1\103\132\0\1\103\153\0\2\103\11\0"+
+    "\1\103\115\0\2\103\6\0\1\103\56\0\1\103\3\0"+
+    "\1\103\2\0\1\103\3\0\1\103\5\0\1\103\7\0"+
+    "\1\103\4\0\2\103\3\0\2\103\1\0\1\103\4\0"+
+    "\1\103\1\0\1\103\2\0\2\103\1\0\3\103\1\0"+
+    "\1\103\2\0\4\103\2\0\1\103\42\0\1\116\37\0"+
+    "\1\116\1\0\2\116\16\0\1\116\4\0\1\116\2\0"+
+    "\2\116\10\0\1\52\4\0\1\116\37\0\1\52\102\0"+
+    "\1\52\147\0\2\52\134\0\1\116\153\0\2\116\11\0"+
+    "\1\116\115\0\2\116\6\0\1\116\56\0\1\116\3\0"+
+    "\1\116\2\0\1\116\3\0\1\116\5\0\1\116\7\0"+
+    "\1\116\4\0\2\116\3\0\2\116\1\0\1\116\4\0"+
+    "\1\116\1\0\1\116\2\0\2\116\1\0\3\116\1\0"+
+    "\1\116\2\0\4\116\2\0\1\116\40\0";
 
   private static int [] zzUnpackTrans() {
-    int [] result = new int[9180];
+    int [] result = new int[10609];
     int offset = 0;
     offset = zzUnpackTrans(ZZ_TRANS_PACKED_0, offset, result);
     return result;
@@ -571,11 +621,11 @@ public final class StandardTokenizerImpl
   private static final int [] ZZ_ATTRIBUTE = zzUnpackAttribute();
 
   private static final String ZZ_ATTRIBUTE_PACKED_0 =
-    "\1\0\1\11\27\1\2\11\15\0\1\1\1\0\1\1"+
-    "\10\0\1\1\61\0";
+    "\1\0\1\11\27\1\2\11\1\1\15\0\1\1\1\0"+
+    "\1\1\10\0\1\1\15\0\1\1\57\0";
 
   private static int [] zzUnpackAttribute() {
-    int [] result = new int[101];
+    int [] result = new int[114];
     int offset = 0;
     offset = zzUnpackAttribute(ZZ_ATTRIBUTE_PACKED_0, offset, result);
     return result;
@@ -662,6 +712,10 @@ public final class StandardTokenizerImpl
   public static final int IDEOGRAPHIC_TYPE = StandardTokenizer.IDEOGRAPHIC;
   
   public static final int HIRAGANA_TYPE = StandardTokenizer.HIRAGANA;
+  
+  public static final int KATAKANA_TYPE = StandardTokenizer.KATAKANA;
+  
+  public static final int HANGUL_TYPE = StandardTokenizer.HANGUL;
 
   public final int yychar()
   {
@@ -706,7 +760,7 @@ public final class StandardTokenizerImpl
     char [] map = new char[0x10000];
     int i = 0;  /* index in packed string  */
     int j = 0;  /* index in unpacked array */
-    while (i < 2640) {
+    while (i < 2650) {
       int  count = packed.charAt(i++);
       char value = packed.charAt(i++);
       do map[j++] = value; while (--count > 0);
@@ -985,30 +1039,38 @@ public final class StandardTokenizerImpl
       zzMarkedPos = zzMarkedPosL;
 
       switch (zzAction < 0 ? zzAction : ZZ_ACTION[zzAction]) {
-        case 1: 
-          { /* Not numeric, word, ideographic, hiragana, or SE Asian -- ignore it. */
-          }
-        case 7: break;
-        case 6: 
-          { return HIRAGANA_TYPE;
-          }
-        case 8: break;
         case 2: 
           { return WORD_TYPE;
           }
         case 9: break;
         case 5: 
-          { return IDEOGRAPHIC_TYPE;
+          { return SOUTH_EAST_ASIAN_TYPE;
           }
         case 10: break;
-        case 3: 
-          { return NUMERIC_TYPE;
+        case 4: 
+          { return KATAKANA_TYPE;
           }
         case 11: break;
-        case 4: 
-          { return SOUTH_EAST_ASIAN_TYPE;
+        case 6: 
+          { return IDEOGRAPHIC_TYPE;
           }
         case 12: break;
+        case 1: 
+          { /* Not numeric, word, ideographic, hiragana, or SE Asian -- ignore it. */
+          }
+        case 13: break;
+        case 8: 
+          { return HANGUL_TYPE;
+          }
+        case 14: break;
+        case 3: 
+          { return NUMERIC_TYPE;
+          }
+        case 15: break;
+        case 7: 
+          { return HIRAGANA_TYPE;
+          }
+        case 16: break;
         default: 
           if (zzInput == YYEOF && zzStartRead == zzCurrentPos) {
             zzAtEOF = true;

Modified: lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.jflex
URL: http://svn.apache.org/viewvc/lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.jflex?rev=1068997&r1=1068996&r2=1068997&view=diff
==============================================================================
--- lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.jflex (original)
+++ lucene/dev/branches/branch_3x/lucene/src/java/org/apache/lucene/analysis/standard/StandardTokenizerImpl.jflex Wed Feb  9 17:47:13 2011
@@ -59,6 +59,8 @@ ComplexContext = ([\p{LB:Complex_Context
 Han = ([\p{Script:Han}] | {HanSupp})
 Hiragana = ([\p{Script:Hiragana}] | {HiraganaSupp})
 
+// Script=Hangul & Aletter
+HangulEx       = (!(!\p{Script:Hangul}|!\p{WB:ALetter})) ({Format} | {Extend})*
 // UAX#29 WB4. X (Extend | Format)* --> X
 //
 ALetterEx      = {ALetter}                     ({Format} | {Extend})*
@@ -90,6 +92,10 @@ ExtendNumLetEx = {ExtendNumLet}         
   public static final int IDEOGRAPHIC_TYPE = StandardTokenizer.IDEOGRAPHIC;
   
   public static final int HIRAGANA_TYPE = StandardTokenizer.HIRAGANA;
+  
+  public static final int KATAKANA_TYPE = StandardTokenizer.KATAKANA;
+  
+  public static final int HANGUL_TYPE = StandardTokenizer.HANGUL;
 
   public final int yychar()
   {
@@ -123,6 +129,12 @@ ExtendNumLetEx = {ExtendNumLet}         
 {ExtendNumLetEx}* 
   { return NUMERIC_TYPE; }
 
+// subset of the below for typing purposes only!
+{HangulEx}+
+  { return HANGUL_TYPE; }
+  
+{KatakanaEx}+
+  { return KATAKANA_TYPE; }
 
 // UAX#29 WB5.   ALetter × ALetter
 //        WB6.   ALetter × (MidLetter | MidNumLet) ALetter