tikhomirov@320: /*
tikhomirov@526:  * Copyright (c) 2011-2013 TMate Software Ltd
tikhomirov@320:  *  
tikhomirov@320:  * This program is free software; you can redistribute it and/or modify
tikhomirov@320:  * it under the terms of the GNU General Public License as published by
tikhomirov@320:  * the Free Software Foundation; version 2 of the License.
tikhomirov@320:  *
tikhomirov@320:  * This program is distributed in the hope that it will be useful,
tikhomirov@320:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
tikhomirov@320:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
tikhomirov@320:  * GNU General Public License for more details.
tikhomirov@320:  *
tikhomirov@320:  * For information on how to redistribute this software under
tikhomirov@320:  * the terms of a license other than GNU General Public License
tikhomirov@320:  * contact TMate Software at support@hg4j.com
tikhomirov@320:  */
tikhomirov@320: package org.tmatesoft.hg.internal;
tikhomirov@320: 
tikhomirov@456: import static org.tmatesoft.hg.util.LogFacility.Severity.Error;
tikhomirov@456: 
tikhomirov@667: import java.io.UnsupportedEncodingException;
tikhomirov@412: import java.nio.ByteBuffer;
tikhomirov@415: import java.nio.CharBuffer;
tikhomirov@412: import java.nio.charset.CharacterCodingException;
tikhomirov@412: import java.nio.charset.Charset;
tikhomirov@412: import java.nio.charset.CharsetDecoder;
tikhomirov@412: import java.nio.charset.CharsetEncoder;
tikhomirov@320: 
tikhomirov@415: import org.tmatesoft.hg.core.SessionContext;
tikhomirov@667: import org.tmatesoft.hg.repo.HgInvalidStateException;
tikhomirov@415: 
tikhomirov@320: /**
tikhomirov@320:  * Keep all encoding-related issues in the single place
tikhomirov@415:  * NOT thread-safe (encoder and decoder requires synchronized access)
tikhomirov@667:  * 
tikhomirov@667:  * @see http://mercurial.selenic.com/wiki/EncodingStrategy
tikhomirov@667:  * @see http://mercurial.selenic.com/wiki/WindowsUTF8Plan
tikhomirov@667:  * @see http://mercurial.selenic.com/wiki/CharacterEncodingOnWindows
tikhomirov@320:  * @author Artem Tikhomirov
tikhomirov@320:  * @author TMate Software Ltd.
tikhomirov@320:  */
tikhomirov@320: public class EncodingHelper {
tikhomirov@412: 	/*
tikhomirov@412: 	 * To understand what Mercurial thinks of UTF-8 and Unix byte approach to names, see
tikhomirov@412: 	 * http://mercurial.808500.n3.nabble.com/Unicode-support-request-td3430704.html
tikhomirov@412: 	 */
tikhomirov@412: 	
tikhomirov@415: 	private final SessionContext sessionContext;
tikhomirov@412: 	private final CharsetEncoder encoder;
tikhomirov@412: 	private final CharsetDecoder decoder;
tikhomirov@667: 	private final CharsetEncoder utfEncoder;
tikhomirov@667: 	private final CharsetDecoder utfDecoder;
tikhomirov@412: 	
tikhomirov@667: 	EncodingHelper(Charset fsEncoding, SessionContext.Source ctx) {
tikhomirov@667: 		sessionContext = ctx.getSessionContext();
tikhomirov@412: 		decoder = fsEncoding.newDecoder();
tikhomirov@412: 		encoder = fsEncoding.newEncoder();
tikhomirov@667: 		Charset utf8 = getUTF8();
tikhomirov@667: 		if (fsEncoding.equals(utf8)) {
tikhomirov@667: 			utfDecoder = decoder;
tikhomirov@667: 			utfEncoder = encoder;
tikhomirov@667: 		} else {
tikhomirov@667: 			utfDecoder = utf8.newDecoder();
tikhomirov@667: 			utfEncoder = utf8.newEncoder();
tikhomirov@667: 		}
tikhomirov@412: 	}
tikhomirov@320: 
tikhomirov@418: 	/**
tikhomirov@418: 	 * Translate file names from manifest to amazing Unicode string 
tikhomirov@418: 	 */
tikhomirov@412: 	public String fromManifest(byte[] data, int start, int length) {
tikhomirov@418: 		return decodeWithSystemDefaultFallback(data, start, length);
tikhomirov@320: 	}
tikhomirov@418: 	
tikhomirov@415: 	/**
tikhomirov@415: 	 * @return byte representation of the string directly comparable to bytes in manifest
tikhomirov@415: 	 */
tikhomirov@526: 	public byte[] toManifest(CharSequence s) {
tikhomirov@415: 		if (s == null) {
tikhomirov@415: 			// perhaps, can return byte[0] in this case?
tikhomirov@415: 			throw new IllegalArgumentException();
tikhomirov@415: 		}
tikhomirov@667: 		return toArray(encodeWithSystemDefaultFallback(s));
tikhomirov@525: 	}
tikhomirov@525: 
tikhomirov@525: 	/**
tikhomirov@525: 	 * Translate file names from dirstate to amazing Unicode string 
tikhomirov@525: 	 */
tikhomirov@525: 	public String fromDirstate(byte[] data, int start, int length) {
tikhomirov@525: 		return decodeWithSystemDefaultFallback(data, start, length);
tikhomirov@525: 	}
tikhomirov@525: 	
tikhomirov@526: 	public byte[] toDirstate(CharSequence fname) {
tikhomirov@525: 		if (fname == null) {
tikhomirov@525: 			throw new IllegalArgumentException();
tikhomirov@525: 		}
tikhomirov@667: 		return toArray(encodeWithSystemDefaultFallback(fname));
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@667: 	/**
tikhomirov@667: 	 * prepare filename to be serialized into fncache file
tikhomirov@667: 	 */
tikhomirov@667: 	public ByteBuffer toFNCache(CharSequence fname) {
tikhomirov@525: 		return encodeWithSystemDefaultFallback(fname);
tikhomirov@525: 	}
tikhomirov@667: 	
tikhomirov@667: 	public byte[] toBundle(CharSequence fname) {
tikhomirov@667: 		// yes, mercurial transfers filenames in local encoding
tikhomirov@667: 		// so that if your local encoding doesn't match that on server, 
tikhomirov@667: 		// and you use native characters, you'd likely fail
tikhomirov@667: 		return toArray(encodeWithSystemDefaultFallback(fname));
tikhomirov@667: 	}
tikhomirov@667: 	public String fromBundle(byte[] data, int start, int length) {
tikhomirov@667: 		return decodeWithSystemDefaultFallback(data, start, length);
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@667: 	
tikhomirov@667: 	public String userFromChangeset(byte[] data, int start, int length) {
tikhomirov@667: 		return decodeUnicodeWithFallback(data, start, length);
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@667: 	public String commentFromChangeset(byte[] data, int start, int length) {
tikhomirov@667: 		return decodeUnicodeWithFallback(data, start, length);
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@667: 	public String fileFromChangeset(byte[] data, int start, int length) {
tikhomirov@667: 		return decodeWithSystemDefaultFallback(data, start, length);
tikhomirov@667: 	}
tikhomirov@525: 
tikhomirov@667: 	public byte[] userToChangeset(CharSequence user) {
tikhomirov@667: 		return toArray(encodeUnicode(user));
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@667: 	public byte[] commentToChangeset(CharSequence comment) {
tikhomirov@667: 		return toArray(encodeUnicode(comment));
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@667: 	public byte[] fileToChangeset(CharSequence file) {
tikhomirov@667: 		return toArray(encodeWithSystemDefaultFallback(file));
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@525: 	private String decodeWithSystemDefaultFallback(byte[] data, int start, int length) {
tikhomirov@415: 		try {
tikhomirov@525: 			return decoder.decode(ByteBuffer.wrap(data, start, length)).toString();
tikhomirov@525: 		} catch (CharacterCodingException ex) {
tikhomirov@525: 			sessionContext.getLog().dump(getClass(), Error, ex, String.format("Use of charset %s failed, resort to system default", charset().name()));
tikhomirov@525: 			// resort to system-default
tikhomirov@525: 			return new String(data, start, length);
tikhomirov@525: 		}
tikhomirov@525: 	}
tikhomirov@525: 	
tikhomirov@667: 	private ByteBuffer encodeWithSystemDefaultFallback(CharSequence s) {
tikhomirov@525: 		try {
tikhomirov@525: 			// synchronized(encoder) {
tikhomirov@667: 			return encoder.encode(CharBuffer.wrap(s));
tikhomirov@415: 			// }
tikhomirov@415: 		} catch (CharacterCodingException ex) {
tikhomirov@456: 			sessionContext.getLog().dump(getClass(), Error, ex, String.format("Use of charset %s failed, resort to system default", charset().name()));
tikhomirov@415: 			// resort to system-default
tikhomirov@667: 			return ByteBuffer.wrap(s.toString().getBytes());
tikhomirov@667: 		}
tikhomirov@667: 	}
tikhomirov@667: 
tikhomirov@667: 	private byte[] toArray(ByteBuffer bb) {
tikhomirov@667: 		byte[] rv;
tikhomirov@667: 		if (bb.hasArray() && bb.arrayOffset() == 0) {
tikhomirov@667: 			rv = bb.array();
tikhomirov@667: 			if (rv.length == bb.remaining()) {
tikhomirov@667: 				return rv;
tikhomirov@667: 			}
tikhomirov@667: 			// fall through
tikhomirov@667: 		}
tikhomirov@667: 		rv = new byte[bb.remaining()];
tikhomirov@667: 		bb.get(rv, 0, rv.length);
tikhomirov@667: 		return rv;
tikhomirov@667: 	}
tikhomirov@667: 
tikhomirov@667: 	private String decodeUnicodeWithFallback(byte[] data, int start, int length) {
tikhomirov@667: 		try {
tikhomirov@667: 			return utfDecoder.decode(ByteBuffer.wrap(data, start, length)).toString();
tikhomirov@667: 		} catch (CharacterCodingException ex) {
tikhomirov@667: 			// TODO post-1.2 respect ui.fallbackencoding actual setting
tikhomirov@682: 			try {
tikhomirov@682: 				return new String(data, start, length, "ISO-8859-1"); // XXX java5
tikhomirov@682: 			} catch (UnsupportedEncodingException e) {
tikhomirov@682: 				throw new HgInvalidStateException(ex.getMessage());
tikhomirov@682: 			}
tikhomirov@667: 		}
tikhomirov@667: 	}
tikhomirov@667: 	
tikhomirov@667: 	private ByteBuffer encodeUnicode(CharSequence s) {
tikhomirov@667: 		// 
tikhomirov@667: 		try {
tikhomirov@667: 			return utfEncoder.encode(CharBuffer.wrap(s));
tikhomirov@667: 		} catch (CharacterCodingException ex) {
tikhomirov@667: 			byte[] rv;
tikhomirov@667: 			try {
tikhomirov@667: 				rv = s.toString().getBytes(getUTF8().name()); // XXX Java 1.5
tikhomirov@667: 			} catch (UnsupportedEncodingException e) {
tikhomirov@667: 				throw new HgInvalidStateException("Unexpected error trying to get UTF-8 encoding"); 
tikhomirov@667: 			}
tikhomirov@667: 			return ByteBuffer.wrap(rv);
tikhomirov@415: 		}
tikhomirov@415: 	}
tikhomirov@415: 
tikhomirov@418: 	private Charset charset() {
tikhomirov@412: 		return encoder.charset();
tikhomirov@412: 	}
tikhomirov@415: 
tikhomirov@527: 	public static Charset getUTF8() {
tikhomirov@527: 		return Charset.forName("UTF-8");
tikhomirov@527: 	}
tikhomirov@320: }