/*
* @(#)CharToByteUnicode.java 1.19 06/10/10
*
* Copyright 1990-2008 Sun Microsystems, Inc. All Rights Reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER
*
* This program 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.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License version 2 for more details (a copy is
* included at /legal/license.txt).
*
* You should have received a copy of the GNU General Public License
* version 2 along with this work; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
* 02110-1301 USA
*
* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa
* Clara, CA 95054 or visit www.sun.com if you need additional
* information or have any questions.
*
*/
package sun.io;
import java.io.*;
/**
* Convert arrays containing Unicode characters into arrays of bytes, using the
* platform-default byte order.
*
* @version 1.13, 00/02/02
* @author Mark Reinhold
*/
public class CharToByteUnicode extends CharToByteConverter {
static final char BYTE_ORDER_MARK = (char) 0xfeff;
protected boolean usesMark = true; /* A mark should be written */
private boolean markWritten = false; /* A mark has been written */
static final int UNKNOWN = 0;
static final int BIG = 1;
static final int LITTLE = 2;
protected int byteOrder = UNKNOWN;
public CharToByteUnicode() {
String enc = (String) java.security.AccessController.doPrivileged(
new sun.security.action.GetPropertyAction("sun.io.unicode.encoding",
"UnicodeBig"));
if (enc.equals("UnicodeBig"))
byteOrder = BIG;
else if (enc.equals("UnicodeLittle"))
byteOrder = LITTLE;
else
byteOrder = BIG;
}
public CharToByteUnicode(int byteOrder, boolean usesMark) {
this.byteOrder = byteOrder;
this.usesMark = usesMark;
}
public CharToByteUnicode(boolean usesMark) {
this();
this.usesMark = usesMark;
}
public String getCharacterEncoding() {
switch (byteOrder) {
case BIG:
return usesMark ? "UnicodeBig" : "UnicodeBigUnmarked";
case LITTLE:
return usesMark ? "UnicodeLittle" : "UnicodeLittleUnmarked";
default:
return "UnicodeUnknown";
}
}
public int convert(char in[], int inOff, int inEnd,
byte out[], int outOff, int outEnd)
throws ConversionBufferFullException, MalformedInputException
{
charOff = inOff;
byteOff = outOff;
if (inOff >= inEnd)
return 0;
int inI = inOff,
outI = outOff,
outTop = outEnd - 2;
if (usesMark && !markWritten) {
if (outI > outTop)
throw new ConversionBufferFullException();
if (byteOrder == BIG) {
out[outI++] = (byte) (BYTE_ORDER_MARK >> 8);
out[outI++] = (byte) (BYTE_ORDER_MARK & 0xff);
}
else {
out[outI++] = (byte) (BYTE_ORDER_MARK & 0xff);
out[outI++] = (byte) (BYTE_ORDER_MARK >> 8);
}
markWritten = true;
}
if (byteOrder == BIG) {
while (inI < inEnd) {
if (outI > outTop) {
charOff = inI;
byteOff = outI;
throw new ConversionBufferFullException();
}
char c = in[inI++];
out[outI++] = (byte) (c >> 8);
out[outI++] = (byte) (c & 0xff);
}
}
else {
while (inI < inEnd) {
if (outI > outTop) {
charOff = inI;
byteOff = outI;
throw new ConversionBufferFullException();
}
char c = in[inI++];
out[outI++] = (byte) (c & 0xff);
out[outI++] = (byte) (c >> 8);
}
}
charOff = inI;
byteOff = outI;
return outI - outOff;
}
public int flush(byte in[], int inOff, int inEnd) {
byteOff = charOff = 0;
return 0;
}
public void reset () {
byteOff = charOff = 0;
markWritten = false;
}
public int getMaxBytesPerChar() {
return 4; /* To allow for writing the byte-order mark */
}
}