/*******************************************************************************
* Copyright (c) 2006-2008, Cloudsmith Inc.
* The code, documentation and other materials contained herein have been
* licensed under the Eclipse Public License - v 1.0 by the copyright holder
* listed above, as the Initial Contributor under such license. The text of
* such license is available at www.eclipse.org.
******************************************************************************/
package org.eclipse.buckminster.p2.remote.marshall;
/**
* This class was modelled after the <code>org.eclipse.osgi.internal.signedcontent.Base64</code>
* class but it provides direct conversions from String to byte[] and byte[] to String using offsets
* and lengths for the byte[], thus minimizing the need for copying.
* @author Thomas Hallgren
*/
public abstract class Base64
{
private static final char digits[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
'N', 'O', 'P', //
'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', //
'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', //
'w', 'x', 'y', 'z', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '+', '/' };
/**
* This method decodes the base 64 encoded String into a byte array. The Base 64 encoding has to
* be according to the specification given by the RFC 1521 (5.2).
* @param data the base 64 encoded String.
* @param result The receiving byte array. Has to be big enough.
* @param offset The starting offset into the result.
* @return the decoded byte array
*/
public static int decode(CharSequence data, byte[] result, int offset)
{
if(data == null)
return -1;
int dataLen = data.length();
if(dataLen == 0)
return 0;
int lastRealDataIndex = dataLen - 1;
while(data.charAt(lastRealDataIndex) == '=')
lastRealDataIndex--;
// original data digit is 8 bits long, but base64 digit is 6 bits long
int padBytes = dataLen - 1 - lastRealDataIndex;
final int byteLength = dataLen * 6 / 8 - padBytes;
if(result.length < byteLength + offset)
throw new IllegalArgumentException("result buffer too small");
// Each 4 bytes of input (encoded) we end up with 3 bytes of output
int dataIndex = 0;
int resultIndex = offset;
int allBits = 0;
// how many result chunks we can process before getting to pad bytes
int resultChunks = (lastRealDataIndex + 1) / 4;
for(int i = 0; i < resultChunks; i++)
{
allBits = 0;
// Loop 4 times gathering input bits (4 * 6 = 24)
for(int j = 0; j < 4; j++)
allBits = (allBits << 6) | decodeDigit(data.charAt(dataIndex++));
// Loop 3 times generating output bits (3 * 8 = 24)
for(int j = resultIndex + 2; j >= resultIndex; j--)
{
result[j] = (byte)(allBits & 0xff); // Bottom 8 bits
allBits = allBits >>> 8;
}
resultIndex += 3; // processed 3 result bytes
}
// Now we do the extra bytes in case the original (non-encoded) data
// was not multiple of 3 bytes
switch(padBytes)
{
case 1:
// 1 pad byte means 3 (4-1) extra Base64 bytes of input, 18
// bits, of which only 16 are meaningful
// Or: 2 bytes of result data
allBits = 0;
// Loop 3 times gathering input bits
for(int j = 0; j < 3; j++)
allBits = (allBits << 6) | decodeDigit(data.charAt(dataIndex++));
// NOTE - The code below ends up being equivalent to allBits =
// allBits>>>2
// But we code it in a non-optimized way for clarity
// The 4th, missing 6 bits are all 0
allBits = allBits << 6;
// The 3rd, missing 8 bits are all 0
allBits = allBits >>> 8;
// Loop 2 times generating output bits
for(int j = resultIndex + 1; j >= resultIndex; j--)
{
result[j] = (byte)(allBits & 0xff); // Bottom 8
// bits
allBits = allBits >>> 8;
}
break;
case 2:
// 2 pad bytes mean 2 (4-2) extra Base64 bytes of input, 12 bits
// of data, of which only 8 are meaningful
// Or: 1 byte of result data
allBits = 0;
// Loop 2 times gathering input bits
for(int j = 0; j < 2; j++)
allBits = (allBits << 6) | decodeDigit(data.charAt(dataIndex++));
// NOTE - The code below ends up being equivalent to allBits =
// allBits>>>4
// But we code it in a non-optimized way for clarity
// The 3rd and 4th, missing 6 bits are all 0
allBits = allBits << 6;
allBits = allBits << 6;
// The 3rd and 4th, missing 8 bits are all 0
allBits = allBits >>> 8;
allBits = allBits >>> 8;
result[resultIndex] = (byte)(allBits & 0xff); // Bottom
// 8
// bits
break;
}
return byteLength;
}
/**
* Calculates the size of the byte array decoded from the base-64 encoded character sequence.
* @param data The sequence to decode.
* @return The number of bytes needed for the resulting byte array.
*/
public static int decodedSize(CharSequence data)
{
if(data == null)
return -1;
int dataLen = data.length();
if(dataLen == 0)
return 0;
int lastRealDataIndex = dataLen - 1;
while(data.charAt(lastRealDataIndex) == '=')
lastRealDataIndex--;
// original data digit is 8 bits long, but base64 digit is 6 bits long
int padChars = dataLen - 1 - lastRealDataIndex;
return dataLen * 6 / 8 - padChars;
}
/**
* This method encodes the byte array into a char array in base 64 according to the
* specification given by the RFC 1521 (5.2).
* @param data the encoded char array
* @return the byte array that needs to be encoded
*/
public static char[] encode(byte[] data, int offset, int length)
{
int sourceChunks = length / 3;
int len = ((length + 2) / 3) * 4;
char[] result = new char[len];
int extraBytes = length - (sourceChunks * 3);
// Each 4 bytes of input (encoded) we end up with 3 bytes of output
int dataIndex = offset;
int resultIndex = 0;
int allBits = 0;
for(int i = 0; i < sourceChunks; i++)
{
allBits = 0;
// Loop 3 times gathering input bits (3 * 8 = 24)
for(int j = 0; j < 3; j++)
allBits = (allBits << 8) | (data[dataIndex++] & 0xff);
// Loop 4 times generating output bits (4 * 6 = 24)
for(int j = resultIndex + 3; j >= resultIndex; j--)
{
result[j] = digits[(allBits & 0x3f)]; // Bottom
// 6
// bits
allBits = allBits >>> 6;
}
resultIndex += 4; // processed 4 result bytes
}
// Now we do the extra bytes in case the original (non-encoded) data
// is not multiple of 4 bytes
switch(extraBytes)
{
case 1:
allBits = data[dataIndex++]; // actual byte
allBits = allBits << 8; // 8 bits of zeroes
allBits = allBits << 8; // 8 bits of zeroes
// Loop 4 times generating output bits (4 * 6 = 24)
for(int j = resultIndex + 3; j >= resultIndex; j--)
{
result[j] = digits[(allBits & 0x3f)]; // Bottom
// 6
// bits
allBits = allBits >>> 6;
}
// 2 pad tags
result[result.length - 1] = (byte)'=';
result[result.length - 2] = (byte)'=';
break;
case 2:
allBits = data[dataIndex++]; // actual byte
allBits = (allBits << 8) | (data[dataIndex++] & 0xff); // actual
// byte
allBits = allBits << 8; // 8 bits of zeroes
// Loop 4 times generating output bits (4 * 6 = 24)
for(int j = resultIndex + 3; j >= resultIndex; j--)
{
result[j] = digits[(allBits & 0x3f)]; // Bottom
// 6
// bits
allBits = allBits >>> 6;
}
// 1 pad tag
result[result.length - 1] = '=';
break;
}
return result;
}
/**
* This method converts a Base 64 digit to its numeric value.
* @param data digit (character) to convert
* @return value for the digit
*/
private static int decodeDigit(char charData)
{
if(charData <= 'Z' && charData >= 'A')
return charData - 'A';
if(charData <= 'z' && charData >= 'a')
return charData - 'a' + 26;
if(charData <= '9' && charData >= '0')
return charData - '0' + 52;
switch(charData)
{
case '+':
return 62;
case '/':
return 63;
default:
throw new IllegalArgumentException("Invalid char to decode: " + charData); //$NON-NLS-1$
}
}
}