blob: 77cc3fa756cc38a491fec7f6ff3ea3b3551e578d [file] [log] [blame]
/*
* Copyright (c) 1997, 2012, 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
* particular file as subject to the "Classpath" exception as provided
* by Oracle in the LICENSE file that accompanied this code.
*
* This code 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 in the LICENSE file that
* accompanied this code).
*
* 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 Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
/* FROM mail.jar */
package com.sun.xml.internal.org.jvnet.mimepull;
import java.io.*;
/**
* This class implements a UUDecoder. It is implemented as
* a FilterInputStream, so one can just wrap this class around
* any input stream and read bytes from this filter. The decoding
* is done as the bytes are read out.
*
* @author John Mani
* @author Bill Shannon
*/
final class UUDecoderStream extends FilterInputStream {
private String name;
private int mode;
private byte[] buffer = new byte[45]; // max decoded chars in a line = 45
private int bufsize = 0; // size of the cache
private int index = 0; // index into the cache
private boolean gotPrefix = false;
private boolean gotEnd = false;
private LineInputStream lin;
private boolean ignoreErrors;
private boolean ignoreMissingBeginEnd;
private String readAhead;
/**
* Create a UUdecoder that decodes the specified input stream.
* The System property <code>mail.mime.uudecode.ignoreerrors</code>
* controls whether errors in the encoded data cause an exception
* or are ignored. The default is false (errors cause exception).
* The System property <code>mail.mime.uudecode.ignoremissingbeginend</code>
* controls whether a missing begin or end line cause an exception
* or are ignored. The default is false (errors cause exception).
* @param in the input stream
*/
public UUDecoderStream(InputStream in) {
super(in);
lin = new LineInputStream(in);
// default to false
ignoreErrors = PropUtil.getBooleanSystemProperty(
"mail.mime.uudecode.ignoreerrors", false);
// default to false
ignoreMissingBeginEnd = PropUtil.getBooleanSystemProperty(
"mail.mime.uudecode.ignoremissingbeginend", false);
}
/**
* Create a UUdecoder that decodes the specified input stream.
* @param in the input stream
* @param ignoreErrors ignore errors?
* @param ignoreMissingBeginEnd ignore missing begin or end?
*/
public UUDecoderStream(InputStream in, boolean ignoreErrors,
boolean ignoreMissingBeginEnd) {
super(in);
lin = new LineInputStream(in);
this.ignoreErrors = ignoreErrors;
this.ignoreMissingBeginEnd = ignoreMissingBeginEnd;
}
/**
* Read the next decoded byte from this input stream. The byte
* is returned as an <code>int</code> in the range <code>0</code>
* to <code>255</code>. If no byte is available because the end of
* the stream has been reached, the value <code>-1</code> is returned.
* This method blocks until input data is available, the end of the
* stream is detected, or an exception is thrown.
*
* @return next byte of data, or <code>-1</code> if the end of
* stream is reached.
* @exception IOException if an I/O error occurs.
* @see java.io.FilterInputStream#in
*/
@Override
public int read() throws IOException {
if (index >= bufsize) {
readPrefix();
if (!decode()) {
return -1;
}
index = 0; // reset index into buffer
}
return buffer[index++] & 0xff; // return lower byte
}
@Override
public int read(byte[] buf, int off, int len) throws IOException {
int i, c;
for (i = 0; i < len; i++) {
if ((c = read()) == -1) {
if (i == 0) {// At end of stream, so we should
i = -1; // return -1, NOT 0.
}
break;
}
buf[off+i] = (byte)c;
}
return i;
}
@Override
public boolean markSupported() {
return false;
}
@Override
public int available() throws IOException {
// This is only an estimate, since in.available()
// might include CRLFs too ..
return ((in.available() * 3)/4 + (bufsize-index));
}
/**
* Get the "name" field from the prefix. This is meant to
* be the pathname of the decoded file
*
* @return name of decoded file
* @exception IOException if an I/O error occurs.
*/
public String getName() throws IOException {
readPrefix();
return name;
}
/**
* Get the "mode" field from the prefix. This is the permission
* mode of the source file.
*
* @return permission mode of source file
* @exception IOException if an I/O error occurs.
*/
public int getMode() throws IOException {
readPrefix();
return mode;
}
/**
* UUencoded streams start off with the line:
* "begin <mode> <filename>"
* Search for this prefix and gobble it up.
*/
private void readPrefix() throws IOException {
if (gotPrefix) {
return;
}
mode = 0666; // defaults, overridden below
name = "encoder.buf"; // same default used by encoder
String line;
for (;;) {
// read till we get the prefix: "begin MODE FILENAME"
line = lin.readLine(); // NOTE: readLine consumes CRLF pairs too
if (line == null) {
if (!ignoreMissingBeginEnd) {
throw new DecodingException("UUDecoder: Missing begin");
}
// at EOF, fake it
gotPrefix = true;
gotEnd = true;
break;
}
if (line.regionMatches(false, 0, "begin", 0, 5)) {
try {
mode = Integer.parseInt(line.substring(6,9));
} catch (NumberFormatException ex) {
if (!ignoreErrors) {
throw new DecodingException(
"UUDecoder: Error in mode: " + ex.toString());
}
}
if (line.length() > 10) {
name = line.substring(10);
} else {
if (!ignoreErrors) {
throw new DecodingException(
"UUDecoder: Missing name: " + line);
}
}
gotPrefix = true;
break;
} else if (ignoreMissingBeginEnd && line.length() != 0) {
int count = line.charAt(0);
count = (count - ' ') & 0x3f;
int need = ((count * 8)+5)/6;
if (need == 0 || line.length() >= need + 1) {
/*
* Looks like a legitimate encoded line.
* Pretend we saw the "begin" line and
* save this line for later processing in
* decode().
*/
readAhead = line;
gotPrefix = true; // fake it
break;
}
}
}
}
private boolean decode() throws IOException {
if (gotEnd) {
return false;
}
bufsize = 0;
int count = 0;
String line;
for (;;) {
/*
* If we ignored a missing "begin", the first line
* will be saved in readAhead.
*/
if (readAhead != null) {
line = readAhead;
readAhead = null;
} else {
line = lin.readLine();
}
/*
* Improperly encoded data sometimes omits the zero length
* line that starts with a space character, we detect the
* following "end" line here.
*/
if (line == null) {
if (!ignoreMissingBeginEnd) {
throw new DecodingException(
"UUDecoder: Missing end at EOF");
}
gotEnd = true;
return false;
}
if (line.equals("end")) {
gotEnd = true;
return false;
}
if (line.length() == 0) {
continue;
}
count = line.charAt(0);
if (count < ' ') {
if (!ignoreErrors) {
throw new DecodingException(
"UUDecoder: Buffer format error");
}
continue;
}
/*
* The first character in a line is the number of original (not
* the encoded atoms) characters in the line. Note that all the
* code below has to handle the <SPACE> character that indicates
* end of encoded stream.
*/
count = (count - ' ') & 0x3f;
if (count == 0) {
line = lin.readLine();
if (line == null || !line.equals("end")) {
if (!ignoreMissingBeginEnd) {
throw new DecodingException(
"UUDecoder: Missing End after count 0 line");
}
}
gotEnd = true;
return false;
}
int need = ((count * 8)+5)/6;
//System.out.println("count " + count + ", need " + need + ", len " + line.length());
if (line.length() < need + 1) {
if (!ignoreErrors) {
throw new DecodingException(
"UUDecoder: Short buffer error");
}
continue;
}
// got a line we're committed to, break out and decode it
break;
}
int i = 1;
byte a, b;
/*
* A correct uuencoder always encodes 3 characters at a time, even
* if there aren't 3 characters left. But since some people out
* there have broken uuencoders we handle the case where they
* don't include these "unnecessary" characters.
*/
while (bufsize < count) {
// continue decoding until we get 'count' decoded chars
a = (byte)((line.charAt(i++) - ' ') & 0x3f);
b = (byte)((line.charAt(i++) - ' ') & 0x3f);
buffer[bufsize++] = (byte)(((a << 2) & 0xfc) | ((b >>> 4) & 3));
if (bufsize < count) {
a = b;
b = (byte)((line.charAt(i++) - ' ') & 0x3f);
buffer[bufsize++] =
(byte)(((a << 4) & 0xf0) | ((b >>> 2) & 0xf));
}
if (bufsize < count) {
a = b;
b = (byte)((line.charAt(i++) - ' ') & 0x3f);
buffer[bufsize++] = (byte)(((a << 6) & 0xc0) | (b & 0x3f));
}
}
return true;
}
/*** begin TEST program *****
public static void main(String argv[]) throws Exception {
FileInputStream infile = new FileInputStream(argv[0]);
UUDecoderStream decoder = new UUDecoderStream(infile);
int c;
try {
while ((c = decoder.read()) != -1)
System.out.write(c);
System.out.flush();
} catch (Exception e) {
e.printStackTrace();
}
}
**** end TEST program ****/
}