< prev index next >
src/jdk.charsets/share/classes/sun/nio/cs/ext/IBM29626C.java.template
Print this page
@@ -1,7 +1,7 @@
/*
- * Copyright (c) 2002, 2012, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2018, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation. Oracle designates this
@@ -37,28 +37,25 @@
import sun.nio.cs.Surrogate;
import sun.nio.cs.SingleByte;
import sun.nio.cs.*;
import static sun.nio.cs.CharsetMapping.*;
-public class EUC_JP
+public class IBM29626C
extends Charset
implements HistoricallyNamedCharset
{
- public EUC_JP() {
- super("EUC-JP", $ALIASES$);
+ public IBM29626C() {
+ super("x-IBM29626C", $ALIASES$);
}
public String historicalName() {
- return "EUC_JP";
+ return "Cp29626C";
}
public boolean contains(Charset cs) {
return ((cs.name().equals("US-ASCII"))
- || (cs instanceof JIS_X_0201)
- || (cs instanceof JIS_X_0208)
- || (cs instanceof JIS_X_0212)
- || (cs instanceof EUC_JP));
+ || (cs instanceof IBM29626C));
}
public CharsetDecoder newDecoder() {
return new Decoder(this);
}
@@ -77,14 +74,65 @@
(DoubleByte.Decoder)new JIS_X_0208().newDecoder();
final static DoubleByte.Decoder DEC0212 =
(DoubleByte.Decoder)new JIS_X_0212().newDecoder();
+ final static DoubleByte.Encoder ibm943 =
+ (DoubleByte.Encoder)new IBM943().newEncoder();
+
private final SingleByte.Decoder dec0201;
private final DoubleByte.Decoder dec0208;
private final DoubleByte.Decoder dec0212;
+ private final String G2_b =
+ "\uA1F1\uA1F2\uA2CC\uADA1\uADA2\uADA3\uADA4\uADA5\uADA6\uADA7"+
+ "\uADA8\uADA9\uADAA\uADAB\uADAC\uADAD\uADAE\uADAF\uADB0\uADB1"+
+ "\uADB2\uADB3\uADB4\uADB5\uADB6\uADB7\uADB8\uADB9\uADBA\uADBB"+
+ "\uADBC\uADBD\uADBE\uADC0\uADC1\uADC2\uADC3\uADC4\uADC5\uADC6"+
+ "\uADC7\uADC8\uADC9\uADCA\uADCB\uADCC\uADCD\uADCE\uADCF\uADD0"+
+ "\uADD1\uADD2\uADD3\uADD4\uADD5\uADD6\uADDF\uADE0\uADE1\uADE2"+
+ "\uADE3\uADE4\uADE5\uADE6\uADE7\uADE8\uADE9\uADEA\uADEB\uADEC"+
+ "\uADED\uADEE\uADEF\uADF0\uADF1\uADF2\uADF3\uADF4\uADF5\uADF6"+
+ "\uADF7\uADF8\uADF9\uADFA\uADFB\uADFC";
+
+ private final String G2_c =
+ "\uFFE0\uFFE1\uFFE2\u2460\u2461\u2462\u2463\u2464\u2465\u2466"+
+ "\u2467\u2468\u2469\u246A\u246B\u246C\u246D\u246E\u246F\u2470"+
+ "\u2471\u2472\u2473\u2160\u2161\u2162\u2163\u2164\u2165\u2166"+
+ "\u2167\u2168\u2169\u3349\u3314\u3322\u334D\u3318\u3327\u3303"+
+ "\u3336\u3351\u3357\u330D\u3326\u3323\u332B\u334A\u333B\u339C"+
+ "\u339D\u339E\u338E\u338F\u33C4\u33A1\u337B\u301D\u301F\u2116"+
+ "\u33CD\u2121\u32A4\u32A5\u32A6\u32A7\u32A8\u3231\u3232\u3239"+
+ "\u337E\u337D\u337C\u2252\u2261\u222B\u222E\u2211\u221A\u22A5"+
+ "\u2220\u221F\u22BF\u2235\u2229\u222A";
+
+ private final String G3_b =
+ "\uF3B8\uF3B9\uF3AB\uF3AC\uF3AD\uF3AE\uF3AF\uF3B0\uF3B1\uF3B2"+
+ "\uF3B3\uF3B4\uF3A1\uF3A2\uF3A3\uF3A4\uF3A5\uF3A6\uF3A7\uF3A8"+
+ "\uF3A9\uF3AA\uF3B7\uF3B8\uF4A2\uF4A3\uF4A4\uF4A5\uF4A6\uF4A8"+
+ "\uF4A9\uF4AC\uF4AE\uF4AF\uF4B0\uF4B2\uF4B3\uF4B4\uF4B5\uF4B6"+
+ "\uF4B7\uF4BA\uF4BD\uF4BE\uF4C0\uF4BF\uF4C2\uF4A1\uF4C6\uF4C7"+
+ "\uF4C8\uF4CB\uF4D0\uF4D4\uF4D5\uF4D7\uF4D9\uF4DC\uF4DF\uF4E0"+
+ "\uF4E1\uF4E5\uF4E7\uF4EA\uF4ED\uF4EE\uF4EF\uF4F4\uF4F5\uF4F6"+
+ "\uF4F8\uF3B8\uF4B9\uF4EB\uF4A7\uF4AA\uF4AB\uF4B1\uF4B8\uF4BB"+
+ "\uF4BC\uF4C4\uF4C5\uF4C9\uF4CC\uF4CD\uF4CE\uF4CF\uF4D1\uF4D3"+
+ "\uF4D6\uF4D8\uF4DA\uF4DB\uF4DE\uF4E2\uF4E3\uF4E4\uF4E6\uF4E8"+
+ "\uF4E9\uF4EC\uF4F1\uF4F2\uF4F3\uF4F7\uF3B6\uF3B5";
+
+ private final String G3_c =
+ "\u2116\u2121\u2160\u2161\u2162\u2163\u2164\u2165\u2166\u2167"+
+ "\u2168\u2169\u2170\u2171\u2172\u2173\u2174\u2175\u2176\u2177"+
+ "\u2178\u2179\u3231\u00A6\u4EFC\u50F4\u51EC\u5307\u5324\u548A"+
+ "\u5759\u589E\u5BEC\u5CF5\u5D53\u5FB7\u6085\u6120\u654E\u663B"+
+ "\u6665\u6801\u6A6B\u6AE2\u6DF2\u6DF8\u7028\u70BB\u7501\u7682"+
+ "\u769E\u7930\u7AE7\u7DA0\u7DD6\u8362\u85B0\u8807\u8B7F\u8CF4"+
+ "\u8D76\u90DE\u9115\u9592\u973B\u974D\u9751\u999E\u9AD9\u9B72"+
+ "\u9ED1\uF86F\uF929\uF9DC\uFA0E\uFA0F\uFA10\uFA11\uFA12\uFA13"+
+ "\uFA14\uFA15\uFA16\uFA17\uFA18\uFA19\uFA1A\uFA1B\uFA1C\uFA1D"+
+ "\uFA1E\uFA1F\uFA20\uFA21\uFA22\uFA23\uFA24\uFA25\uFA26\uFA27"+
+ "\uFA28\uFA29\uFA2A\uFA2B\uFA2C\uFA2D\uFF02\uFF07";
+
protected Decoder(Charset cs) {
this(cs, 0.5f, 1.0f, DEC0201, DEC0208, DEC0212);
}
protected Decoder(Charset cs, float avgCpb, float maxCpb,
@@ -96,17 +144,68 @@
this.dec0208 = dec0208;
this.dec0212 = dec0212;
}
+ protected char decodeSingle(int b) {
+ if (b < 0x8e)
+ return (char) b;
+ if (b < 0x90)
+ return UNMAPPABLE_DECODING;
+ if (b < 0xa0)
+ return (char) b;
+ return UNMAPPABLE_DECODING;
+ }
+
+ protected char decodeUDC(int byte1, int byte2, int offset) {
+ if ((byte1 >= 0xf5 && byte1 <= 0xfe)
+ && (byte2 >= 0xa1 && byte2 <= 0xfe)) {
+ return (char)((byte1 - 0xf5) * 94 + (byte2 - 0xa1) + offset);
+ }
+ return UNMAPPABLE_DECODING;
+ }
+
+ final static String g1_c = "\u00a2\u00a3\u00ac\\\u007e";
+
protected char decodeDouble(int byte1, int byte2) {
if (byte1 == 0x8e) {
if (byte2 < 0x80)
return UNMAPPABLE_DECODING;
- return dec0201.decode((byte)byte2);
- }
- return dec0208.decodeDouble(byte1 - 0x80, byte2 - 0x80);
+ char c = dec0201.decode((byte)byte2);
+ if (byte2 >= 0xe0 && byte2 <= 0xe4)
+ c = g1_c.charAt(byte2 - 0xe0);
+ return c;
+ }
+ if ((byte1 >= 0xa1 && byte1 <= 0xfe)
+ && (byte2 >= 0xa1 && byte2 <= 0xfe)) {
+ char c = (char)((byte1 << 8) + byte2);
+ int idx = G2_b.indexOf(c);
+ if (idx > -1)
+ return G2_c.charAt(idx);
+ }
+ char ch = dec0208.decodeDouble(byte1 - 0x80, byte2 - 0x80);
+ if (ch == UNMAPPABLE_DECODING)
+ ch = decodeUDC(byte1, byte2, 0xe000);
+ return ch;
+ }
+
+ protected char decodeDoubleG3(int byte1, int byte2) {
+ if ((byte1 >= 0xa1 && byte1 <= 0xfe)
+ && (byte2 >= 0xa1 && byte2 <= 0xfe)) {
+ char c = (char)((byte1 << 8) + byte2);
+ int idx = G3_b.indexOf(c);
+ if (idx > -1)
+ return G3_c.charAt(idx);
+ }
+ char ch = dec0212.decodeDouble(byte1 - 0x80, byte2 - 0x80);
+ if (ch == '\u2116')
+ ch = UNMAPPABLE_DECODING;
+ if (ch != UNMAPPABLE_DECODING)
+ ch = ibm943.canEncode(ch) ? ch : UNMAPPABLE_DECODING;
+ if (ch == UNMAPPABLE_DECODING)
+ ch = decodeUDC(byte1, byte2, 0xe3ac);
+ return ch;
}
private CoderResult decodeArrayLoop(ByteBuffer src,
CharBuffer dst)
{
@@ -128,22 +227,19 @@
try {
while (sp < sl) {
b1 = sa[sp] & 0xff;
inputSize = 1;
- if ((b1 & 0x80) == 0) {
- outputChar = (char)b1;
- } else { // Multibyte char
+ outputChar = decodeSingle(b1);
+ if (outputChar == UNMAPPABLE_DECODING) { // Multibyte char
if (b1 == 0x8f) { // JIS0212
if (sp + 3 > sl)
return CoderResult.UNDERFLOW;
b1 = sa[sp + 1] & 0xff;
b2 = sa[sp + 2] & 0xff;
inputSize += 2;
- if (dec0212 == null) // JIS02012 not supported
- return CoderResult.unmappableForLength(inputSize);
- outputChar = dec0212.decodeDouble(b1-0x80, b2-0x80);
+ outputChar = decodeDoubleG3(b1, b2);
} else { // JIS0201, JIS0208
if (sp + 2 > sl)
return CoderResult.UNDERFLOW;
b2 = sa[sp + 1] & 0xff;
inputSize++;
@@ -175,22 +271,19 @@
try {
while (src.hasRemaining()) {
b1 = src.get() & 0xff;
inputSize = 1;
- if ((b1 & 0x80) == 0) {
- outputChar = (char)b1;
- } else { // Multibyte char
+ outputChar = decodeSingle(b1);
+ if (outputChar == UNMAPPABLE_DECODING) { // Multibyte char
if (b1 == 0x8f) { // JIS0212
if (src.remaining() < 2)
return CoderResult.UNDERFLOW;
b1 = src.get() & 0xff;
b2 = src.get() & 0xff;
inputSize += 2;
- if (dec0212 == null) // JIS02012 not supported
- return CoderResult.unmappableForLength(inputSize);
- outputChar = dec0212.decodeDouble(b1-0x80, b2-0x80);
+ outputChar = decodeDoubleG3(b1, b2);
} else { // JIS0201 JIS0208
if (src.remaining() < 1)
return CoderResult.UNDERFLOW;
b2 = src.get() & 0xff;
inputSize++;
@@ -236,17 +329,71 @@
(DoubleByte.Encoder)new JIS_X_0208().newEncoder();
final static DoubleByte.Encoder ENC0212 =
(DoubleByte.Encoder)new JIS_X_0212().newEncoder();
- private final Surrogate.Parser sgp = new Surrogate.Parser();
+ final static DoubleByte.Encoder ibm943 =
+ (DoubleByte.Encoder)new IBM943().newEncoder();
+ private final Surrogate.Parser sgp = new Surrogate.Parser();
private final SingleByte.Encoder enc0201;
private final DoubleByte.Encoder enc0208;
private final DoubleByte.Encoder enc0212;
+ private final String G2_c =
+ "\u2015\u2211\u221F\u2225\u222E\u22BF\u2460\u2461\u2462\u2463"+
+ "\u2464\u2465\u2466\u2467\u2468\u2469\u246A\u246B\u246C\u246D"+
+ "\u246E\u246F\u2470\u2471\u2472\u2473\u301D\u301F\u3232\u3239"+
+ "\u32A4\u32A5\u32A6\u32A7\u32A8\u3303\u330D\u3314\u3318\u3322"+
+ "\u3323\u3326\u3327\u332B\u3336\u333B\u3349\u334A\u334D\u3351"+
+ "\u3357\u337B\u337C\u337D\u337E\u338E\u338F\u339C\u339D\u339E"+
+ "\u33A1\u33C4\u33CD\u4FE0\u525D\u555E\u5699\u56CA\u5861\u5C5B"+
+ "\u5C62\u6414\u6451\u6522\u6805\u688E\u6F51\u7006\u7130\u7626"+
+ "\u79B1\u7C1E\u7E48\u7E61\u7E6B\u8141\u8346\u840A\u8523\u87EC"+
+ "\u881F\u8EC0\u91AC\u91B1\u9830\u9839\u985A\u9A52\u9DD7\u9E7C"+
+ "\u9EB4\u9EB5\uFF0D\uFF5E\uFFE0\uFFE1\uFFE2";
+
+ private final String G2_b =
+ "\uA1BD\uADF4\uADF8\uA1C2\uADF3\uADF9\uADA1\uADA2\uADA3\uADA4"+
+ "\uADA5\uADA6\uADA7\uADA8\uADA9\uADAA\uADAB\uADAC\uADAD\uADAE"+
+ "\uADAF\uADB0\uADB1\uADB2\uADB3\uADB4\uADE0\uADE1\uADEB\uADEC"+
+ "\uADE5\uADE6\uADE7\uADE8\uADE9\uADC6\uADCA\uADC1\uADC4\uADC2"+
+ "\uADCC\uADCB\uADC5\uADCD\uADC7\uADCF\uADC0\uADCE\uADC3\uADC8"+
+ "\uADC9\uADDF\uADEF\uADEE\uADED\uADD3\uADD4\uADD0\uADD1\uADD2"+
+ "\uADD6\uADD5\uADE3\uB6A2\uC7ED\uB0A2\uB3FA\uC7B9\uC5B6\uD6A2"+
+ "\uBCC8\uC1DF\uC4CF\uDAB9\uBAF4\uDBF4\uC8AE\uC6C2\uB1EB\uC1E9"+
+ "\uC5F8\uC3BD\uE5DA\uBDAB\uB7D2\uE7A6\uB7D5\uCDE9\uBED5\uC0E6"+
+ "\uCFB9\uB6ED\uBEDF\uC8B0\uCBCB\uF0F8\uC5BF\uC2CD\uB2AA\uB8B4"+
+ "\uB9ED\uCCCD\uA1DD\uA1C1\uA1F1\uA1F2\uA2CC";
+
+ private final String G3_c =
+ "\u2116\u2121\u2160\u2161\u2162\u2163\u2164\u2165\u2166\u2167"+
+ "\u2168\u2169\u2170\u2171\u2172\u2173\u2174\u2175\u2176\u2177"+
+ "\u2178\u2179\u3231\u4EFC\u50F4\u51EC\u5307\u5324\u548A\u5759"+
+ "\u589E\u5BEC\u5CF5\u5D53\u5FB7\u6085\u6120\u654E\u663B\u6665"+
+ "\u6801\u6A6B\u6AE2\u6DF2\u6DF8\u7028\u70BB\u7501\u7682\u769E"+
+ "\u7930\u7AE7\u7DA0\u7DD6\u8362\u85B0\u8807\u8B7F\u8CF4\u8D76"+
+ "\u90DE\u9115\u9592\u973B\u974D\u9751\u999E\u9AD9\u9B72\u9ED1"+
+ "\uF86F\uF929\uF9DC\uFA0E\uFA0F\uFA10\uFA11\uFA12\uFA13\uFA14"+
+ "\uFA15\uFA16\uFA17\uFA18\uFA19\uFA1A\uFA1B\uFA1C\uFA1D\uFA1E"+
+ "\uFA1F\uFA20\uFA21\uFA22\uFA23\uFA24\uFA25\uFA26\uFA27\uFA28"+
+ "\uFA29\uFA2A\uFA2B\uFA2C\uFA2D\uFF02\uFF07\uFFE4";
+
+ private final String G3_b =
+ "\uF3B8\uF3B9\uF3AB\uF3AC\uF3AD\uF3AE\uF3AF\uF3B0\uF3B1\uF3B2"+
+ "\uF3B3\uF3B4\uF3A1\uF3A2\uF3A3\uF3A4\uF3A5\uF3A6\uF3A7\uF3A8"+
+ "\uF3A9\uF3AA\uF3B7\uF4A2\uF4A3\uF4A4\uF4A5\uF4A6\uF4A8\uF4A9"+
+ "\uF4AC\uF4AE\uF4AF\uF4B0\uF4B2\uF4B3\uF4B4\uF4B5\uF4B6\uF4B7"+
+ "\uF4BA\uF4BD\uF4BE\uF4C0\uF4BF\uF4C2\uF4A1\uF4C6\uF4C7\uF4C8"+
+ "\uF4CB\uF4D0\uF4D4\uF4D5\uF4D7\uF4D9\uF4DC\uF4DF\uF4E0\uF4E1"+
+ "\uF4E5\uF4E7\uF4EA\uF4ED\uF4EE\uF4EF\uF4F4\uF4F5\uF4F6\uF4F8"+
+ "\uF3B8\uF4B9\uF4EB\uF4A7\uF4AA\uF4AB\uF4B1\uF4B8\uF4BB\uF4BC"+
+ "\uF4C4\uF4C5\uF4C9\uF4CC\uF4CD\uF4CE\uF4CF\uF4D1\uF4D3\uF4D6"+
+ "\uF4D8\uF4DA\uF4DB\uF4DE\uF4E2\uF4E3\uF4E4\uF4E6\uF4E8\uF4E9"+
+ "\uF4EC\uF4F1\uF4F2\uF4F3\uF4F7\uF3B6\uF3B5\uA2C3";
+
protected Encoder(Charset cs) {
this(cs, 3.0f, 3.0f, ENC0201, ENC0208, ENC0212);
}
protected Encoder(Charset cs, float avgBpc, float maxBpc,
@@ -263,12 +410,27 @@
byte[] encodedBytes = new byte[3];
return encodeSingle(c, encodedBytes) != 0 ||
encodeDouble(c) != UNMAPPABLE_ENCODING;
}
+ private final static String G1_c = "\u00A2\u00A3\u00AC";
+
protected int encodeSingle(char inputChar, byte[] outputByte) {
+ if (inputChar >= 0x80 && inputChar < 0x8e) {
+ outputByte[0] = (byte)inputChar;
+ return 1;
+ }
+ if (inputChar >= 0x90 && inputChar < 0xa0) {
+ outputByte[0] = (byte)inputChar;
+ return 1;
+ }
int b = enc0201.encode(inputChar);
+ if (b == UNMAPPABLE_ENCODING) {
+ int idx = G1_c.indexOf(inputChar);
+ if (idx > -1)
+ b = 0xe0 + idx;
+ }
if (b == UNMAPPABLE_ENCODING)
return 0;
if (b >= 0 && b < 128) {
outputByte[0] = (byte)b;
return 1;
@@ -276,18 +438,44 @@
outputByte[0] = (byte)0x8e;
outputByte[1] = (byte)b;
return 2;
}
+ protected int encodeUDC(char ch) {
+ if (ch >= '\ue000' && ch <= '\ue757') {
+ if (ch < '\ue3ac') {
+ int offset = (int)ch - 0xe000;
+ int b = ((offset / 94) << 8) + (offset % 94);
+ return b + 0xf5a1;
+ } else {
+ int offset = (int)ch - 0xe3ac;
+ int b = ((offset / 94) << 8) + (offset % 94);
+ return b + 0x8ff5a1;
+ }
+ }
+ return UNMAPPABLE_ENCODING;
+ }
+
protected int encodeDouble(char ch) {
+ int idx = G2_c.indexOf(ch);
+ if (idx > -1)
+ return (int)G2_b.charAt(idx);
+ idx = G3_c.indexOf(ch);
+ if (idx > -1)
+ return (int)G3_b.charAt(idx) + 0x8f0000;
int b = enc0208.encodeChar(ch);
if (b != UNMAPPABLE_ENCODING)
return b + 0x8080;
- if (enc0212 != null) {
- b = enc0212.encodeChar(ch);
+ b = encodeUDC(ch);
if (b != UNMAPPABLE_ENCODING)
+ return b;
+ if (ibm943.canEncode(ch)) {
+ b = enc0212.encodeChar(ch);
+ if (b != UNMAPPABLE_ENCODING) {
b += 0x8F8080;
+ return b;
+ }
}
return b;
}
private CoderResult encodeArrayLoop(CharBuffer src,
< prev index next >