src/org/lwes/util/CharacterEncoding.java

   1 package org.lwes.util;
   2
   3 import java.io.UnsupportedEncodingException;
   4 import java.util.HashMap;
   5
   6 /**
   7  * This is a little class to abstract the character encoding strings that Java
   8  * uses into classes which can be checked at compile time.
   9  *
  10  * @author Kevin Scaldeferri
  11  */
  12 public abstract class CharacterEncoding {
  13
  14         // static loading
  15         private static final HashMap<String, CharacterEncoding> ENCODING_HASH = new HashMap<String, CharacterEncoding>();
  16         private static final String CANONICAL_ASCII_NAME = "ASCII";
  17         private static final String CANONICAL_ISO_8859_1_NAME = "ISO8859_1";
  18         private static final String CANONICAL_UTF_8_NAME = "UTF8";
  19         private static final String CANONICAL_SHIFT_JIS_NAME = "SJIS";
  20         private static final String CANONICAL_EUC_JP_NAME = "EUC_JP";
  21         private static final String CANONICAL_EUC_KR_NAME = "EUC_KR";
  22         private static final String[] ASCII_ALIASES = { CANONICAL_ASCII_NAME,
  23                         "US-ASCII", "ISO646-US" };
  24         private static final String[] ISO_8859_1_ALIASES = {
  25                         CANONICAL_ISO_8859_1_NAME, "ISO-8859-1", "ISO-LATIN-1", "8859_1" };
  26         private static final String[] UTF_8_ALIASES = { CANONICAL_UTF_8_NAME,
  27                         "UTF-8" };
  28         private static final String[] SHIFT_JIS_ALIASES = {
  29                         CANONICAL_SHIFT_JIS_NAME, "SHIFTJIS", "SHIFT-JIS", "SHIFT_JIS" };
  30         private static final String[] EUC_JP_ALIASES = { CANONICAL_EUC_JP_NAME,
  31                         "EUC-JP" };
  32         private static final String[] EUC_KR_ALIASES = { CANONICAL_EUC_KR_NAME,
  33                         "EUC-KR" };
  34
  35         /*
  36          * Returns Java's canonical form of the encoding.
  37          */
  38         public abstract String getEncodingString();
  39
  40         /*
  41          * Returns the official IANA name for the encoding. Everything expect Java
  42          * expects this form.
  43          */
  44         public abstract String getIANAString();
  45
  46         public static final CharacterEncoding ISO_8859_1 = new CharacterEncoding() {
  47                 public String getEncodingString() {
  48                         return "ISO-8859-1";
  49                 }
  50
  51                 public String getIANAString() {
  52                         return "ISO-8859-1";
  53                 }
  54         };
  55
  56         public static final CharacterEncoding UTF_8 = new CharacterEncoding() {
  57                 public String getEncodingString() {
  58                         return "UTF-8";
  59                 }
  60
  61                 public String getIANAString() {
  62                         return "UTF-8";
  63                 }
  64         };
  65
  66         public static final CharacterEncoding ASCII = new CharacterEncoding() {
  67                 public String getEncodingString() {
  68                         return "ASCII";
  69                 }
  70
  71                 public String getIANAString() {
  72                         return "US-ASCII";
  73                 }
  74         };
  75
  76         public static final CharacterEncoding SHIFT_JIS = new CharacterEncoding() {
  77                 public String getEncodingString() {
  78                         return "SJIS";
  79                 }
  80
  81                 public String getIANAString() {
  82                         return "Shift_JIS";
  83                 }
  84         };
  85
  86         public static final CharacterEncoding EUC_JP = new CharacterEncoding() {
  87                 public String getEncodingString() {
  88                         return "EUC_JP";
  89                 }
  90
  91                 public String getIANAString() {
  92                         return "EUC-JP";
  93                 }
  94         };
  95
  96         public static final CharacterEncoding EUC_KR = new CharacterEncoding() {
  97                 public String getEncodingString() {
  98                         return "EUC_KR";
  99                 }
 100
 101                 public String getIANAString() {
 102                         return "EUC-KR";
 103                 }
 104         };
 105
 106         /**
 107          * This is a highly limited implementation at the moment, so don't expect
 108          * too much from it.
 109      * @param enc the String representation of the encoding.
 110      * @return CharacterEncoding
 111      * @throws java.io.UnsupportedEncodingException if the encoding doesnt exist.
 112      */
 113         public static CharacterEncoding getInstance(String enc)
 114                         throws UnsupportedEncodingException {
 115                 if (ENCODING_HASH.containsKey(enc.toUpperCase())) {
 116                         return ENCODING_HASH.get(enc.toUpperCase());
 117                 } else {
 118                         throw new UnsupportedEncodingException(enc);
 119                 }
 120         }
 121
 122     @Override
 123     public int hashCode() {
 124         return getEncodingString().hashCode();
 125     }
 126
 127     public boolean equals(Object o) {
 128                 return (o instanceof CharacterEncoding)
 129                                 && getEncodingString().equals(((CharacterEncoding) o).getEncodingString());
 130         }
 131
 132         static {
 133                 int i;
 134                 for (i = 0; i < ASCII_ALIASES.length; i++) {
 135                         ENCODING_HASH.put(ASCII_ALIASES[i], ASCII);
 136                 }
 137                 for (i = 0; i < ISO_8859_1_ALIASES.length; i++) {
 138                         ENCODING_HASH.put(ISO_8859_1_ALIASES[i], ISO_8859_1);
 139                 }
 140                 for (i = 0; i < UTF_8_ALIASES.length; i++) {
 141                         ENCODING_HASH.put(UTF_8_ALIASES[i], UTF_8);
 142                 }
 143                 for (i = 0; i < SHIFT_JIS_ALIASES.length; i++) {
 144                         ENCODING_HASH.put(SHIFT_JIS_ALIASES[i], SHIFT_JIS);
 145                 }
 146                 for (i = 0; i < EUC_JP_ALIASES.length; i++) {
 147                         ENCODING_HASH.put(EUC_JP_ALIASES[i], EUC_JP);
 148                 }
 149                 for (i = 0; i < EUC_KR_ALIASES.length; i++) {
 150                         ENCODING_HASH.put(EUC_KR_ALIASES[i], EUC_KR);
 151                 }
 152         }
 153 }