jars/icu4j-52_1/tools/misc/src/com/ibm/icu/dev/tool/compression/CompressionTableGenerator.java

   1 /*
   2  *******************************************************************************
   3  * Copyright (C) 1996-2005, International Business Machines Corporation and    *
   4  * others. All Rights Reserved.                                                *
   5  *******************************************************************************
   6  */
   7 package com.ibm.icu.dev.tool.compression;
   8
   9
  10 /**
  11  * Utility class to generate the tables used by the SCSU interface and
  12  * the UnicodeCompressor class.
  13  *
  14  * @author Stephen F. Booth
  15  * @version 1.0 08 Mar 99
  16  */
  17 class CompressionTableGenerator
  18 {
  19     // duplicate of constants in SCSU
  20
  21     final static int LATININDEX                  = 0xF9;
  22     final static int IPAEXTENSIONINDEX           = 0xFA;
  23     final static int GREEKINDEX                  = 0xFB;
  24     final static int ARMENIANINDEX               = 0xFC;
  25     final static int HIRAGANAINDEX               = 0xFD;
  26     final static int KATAKANAINDEX               = 0xFE;
  27     final static int HALFWIDTHKATAKANAINDEX      = 0xFF;
  28
  29     final static int SDEFINEX                    = 0x0B;
  30     final static int SRESERVED                   = 0x0C;  // reserved value
  31     final static int SQUOTEU                     = 0x0E;
  32     final static int SCHANGEU                    = 0x0F;
  33
  34     final static int SQUOTE0                     = 0x01;
  35     final static int SQUOTE1                     = 0x02;
  36     final static int SQUOTE2                     = 0x03;
  37     final static int SQUOTE3                     = 0x04;
  38     final static int SQUOTE4                     = 0x05;
  39     final static int SQUOTE5                     = 0x06;
  40     final static int SQUOTE6                     = 0x07;
  41     final static int SQUOTE7                     = 0x08;
  42
  43     final static int SCHANGE0                    = 0x10;
  44     final static int SCHANGE1                    = 0x11;
  45     final static int SCHANGE2                    = 0x12;
  46     final static int SCHANGE3                    = 0x13;
  47     final static int SCHANGE4                    = 0x14;
  48     final static int SCHANGE5                    = 0x15;
  49     final static int SCHANGE6                    = 0x16;
  50     final static int SCHANGE7                    = 0x17;
  51
  52     final static int SDEFINE0                    = 0x18;
  53     final static int SDEFINE1                    = 0x19;
  54     final static int SDEFINE2                    = 0x1A;
  55     final static int SDEFINE3                    = 0x1B;
  56     final static int SDEFINE4                    = 0x1C;
  57     final static int SDEFINE5                    = 0x1D;
  58     final static int SDEFINE6                    = 0x1E;
  59     final static int SDEFINE7                    = 0x1F;
  60
  61     //==========================
  62     // Unicode mode tags
  63     //==========================
  64     final static int UCHANGE0                    = 0xE0;
  65     final static int UCHANGE1                    = 0xE1;
  66     final static int UCHANGE2                    = 0xE2;
  67     final static int UCHANGE3                    = 0xE3;
  68     final static int UCHANGE4                    = 0xE4;
  69     final static int UCHANGE5                    = 0xE5;
  70     final static int UCHANGE6                    = 0xE6;
  71     final static int UCHANGE7                    = 0xE7;
  72
  73     final static int UDEFINE0                    = 0xE8;
  74     final static int UDEFINE1                    = 0xE9;
  75     final static int UDEFINE2                    = 0xEA;
  76     final static int UDEFINE3                    = 0xEB;
  77     final static int UDEFINE4                    = 0xEC;
  78     final static int UDEFINE5                    = 0xED;
  79     final static int UDEFINE6                    = 0xEE;
  80     final static int UDEFINE7                    = 0xEF;
  81
  82     final static int UQUOTEU                     = 0xF0;
  83     final static int UDEFINEX                    = 0xF1;
  84     final static int URESERVED                   = 0xF2;  // reserved value
  85
  86     final static int BLOCKSIZE = 0xFF;
  87
  88     /**
  89      * Generate the table used as sOffsetTable in SCSU.
  90      * This table contains preformed indices so we can do array lookups
  91      * instead of calculations for speed during decompression.
  92      */
  93     static void printOffsetTable()
  94     {
  95         int     i           = 0;
  96     int []    offsetTable = new int [ BLOCKSIZE + 1 ];
  97
  98         // 0x00 is reserved
  99
 100         // half blocks from U+0080 to U+3380
 101         for( i = 0x01; i < 0x68; i++ )
 102             offsetTable[i] = i * 0x80;
 103
 104         // half blocks from U+E000 to U+FF80
 105         for( i = 0x68; i < 0xA8; i++ )
 106             offsetTable[i] = (i * 0x80) + 0xAC00;
 107
 108         // 0xA8..0xF8 is reserved
 109
 110         offsetTable[ LATININDEX ] = 0x00C0;
 111         offsetTable[ IPAEXTENSIONINDEX ] = 0x0250;
 112         offsetTable[ GREEKINDEX ] = 0x0370;
 113         offsetTable[ ARMENIANINDEX ] = 0x0530;
 114         offsetTable[ HIRAGANAINDEX ] = 0x3040;
 115         offsetTable[ KATAKANAINDEX ] = 0x30A0;
 116         offsetTable[ HALFWIDTHKATAKANAINDEX ] = 0xFF60;
 117
 118         // dump the generated table
 119     System.out.println("static int [] sOffsetTable = {");
 120         for(i = 0; i < offsetTable.length - 1; i++)
 121             System.out.print("0x" + Integer.toHexString(offsetTable[i])
 122                  + ", ");
 123         for(i = offsetTable.length - 1; i < offsetTable.length; i++)
 124             System.out.print("0x" + Integer.toHexString(offsetTable[i]));
 125         System.out.println();
 126         System.out.println("};");
 127     }
 128
 129     /**
 130      * Generate the table used as sSingleTagTable in UnicodeCompressor.
 131      * This table contains boolean values indicating if a byte is a
 132      * single-byte mode tag.
 133      */
 134     static void printSingleTagTable()
 135     {
 136         int        i              = 0;
 137     boolean [] singleTagTable = new boolean  [ BLOCKSIZE + 1 ];
 138
 139         for( i = 0x00; i <= BLOCKSIZE; i++ ) {
 140             switch( i ) {
 141
 142         case SQUOTEU:  case SCHANGEU:
 143         case SDEFINEX: case SRESERVED:
 144         case SQUOTE0:  case SQUOTE1:
 145         case SQUOTE2:  case SQUOTE3:
 146         case SQUOTE4:  case SQUOTE5:
 147         case SQUOTE6:  case SQUOTE7:
 148         case SCHANGE0: case SCHANGE1:
 149         case SCHANGE2: case SCHANGE3:
 150         case SCHANGE4: case SCHANGE5:
 151         case SCHANGE6: case SCHANGE7:
 152         case SDEFINE0: case SDEFINE1:
 153         case SDEFINE2: case SDEFINE3:
 154         case SDEFINE4: case SDEFINE5:
 155         case SDEFINE6: case SDEFINE7:
 156         singleTagTable[i] = true;
 157                 break;
 158
 159         default:
 160         singleTagTable[i] = false;
 161                 break;
 162             }
 163         }
 164
 165         // dump the generated table
 166         System.out.println("private static boolean [] sSingleTagTable = {");
 167         for(i = 0; i < singleTagTable.length - 1; i++)
 168             System.out.print(singleTagTable[i] + ", ");
 169         for(i = singleTagTable.length - 1; i < singleTagTable.length; i++)
 170             System.out.print(singleTagTable[i]);
 171         System.out.println();
 172         System.out.println("};");
 173     }
 174
 175
 176     /**
 177      * Generate the table used as sUnicodeTagTable in
 178      * This table contains boolean values indicating if a byte is a
 179      * unicode mode tag.
 180      */
 181     static void printUnicodeTagTable()
 182     {
 183         int        i               = 0;
 184     boolean [] unicodeTagTable = new boolean  [ BLOCKSIZE + 1 ];
 185
 186         for( i = 0x00; i <= BLOCKSIZE; i++ ) {
 187             switch( i ) {
 188         case UQUOTEU:  case UDEFINEX:
 189         case URESERVED:
 190         case UCHANGE0: case UCHANGE1:
 191         case UCHANGE2: case UCHANGE3:
 192         case UCHANGE4: case UCHANGE5:
 193         case UCHANGE6: case UCHANGE7:
 194         case UDEFINE0: case UDEFINE1:
 195         case UDEFINE2: case UDEFINE3:
 196         case UDEFINE4: case UDEFINE5:
 197         case UDEFINE6: case UDEFINE7:
 198         unicodeTagTable[i] = true;
 199                 break;
 200
 201         default:
 202         unicodeTagTable[i] = false;
 203                 break;
 204             }
 205         }
 206
 207         // dump the generated table
 208         System.out.println("private static boolean [] sUnicodeTagTable = {");
 209         for(i = 0; i < unicodeTagTable.length - 1; i++)
 210             System.out.print(unicodeTagTable[i] + ", ");
 211         for(i = unicodeTagTable.length - 1; i < unicodeTagTable.length; i++)
 212             System.out.print(unicodeTagTable[i]);
 213         System.out.println();
 214         System.out.println("};");
 215     }
 216
 217     public static void main(String[] argv)
 218     {
 219         printOffsetTable();
 220         printSingleTagTable();
 221         printUnicodeTagTable();
 222     }
 223 }