public class Base64{
/**
* how we separate lines, e.g. \n, \r\n, \r etc.
*/
private String lineSeparator = System.getProperty("line.separator");
/**
* max chars per line, excluding lineSeparator. A multiple of 4.
*/
private int lineLength = 72;
/* constructor */
public Base64 (){
}
/**
* Encode an arbitrary array of bytes as Base64 printable ASCII.
* It will be broken into lines of 72 chars each. The last line is not
* terminated with a line separator.
* The output will always have an even multiple of data characters,
* exclusive of \n. It is padded out with =.
*/
public String encode (byte[] b){
// Each group or partial group of 3 bytes becomes four chars
// covered quotient
int outputLength = ((b.length + 2) / 3) * 4;
// account for trailing newlines, on all but the very last line
if (lineLength != 0){
int lines = (outputLength + lineLength - 1) / lineLength - 1;
if (lines > 0){
outputLength += lines * lineSeparator.length();
}
}
// must be local for recursion to work.
StringBuffer sb = new StringBuffer(outputLength);
// must be local for recursion to work.
int linePos = 0;
// first deal with even multiples of 3 bytes.
int len = (b.length / 3) * 3;
int leftover = b.length - len;
for (int i = 0;i < len;i += 3){
// Start a new line if next 4 chars won't fit on the current line
// We can't encapsulete the following code since the variable need to
// be local to this incarnation of encode.
linePos += 4;
if (linePos > lineLength){
if (lineLength != 0){
sb.append(lineSeparator);
}
linePos = 4;
}
// get next three bytes in unsigned form lined up,
// in big-endian order
int combined = b[i + 0] & 0xff;
combined <<= 8;
combined |= b[i + 1] & 0xff;
combined <<= 8;
combined |= b[i + 2] & 0xff;
// break those 24 bits into a 4 groups of 6 bits,
// working LSB to MSB.
int c3 = combined & 0x3f;
combined >>>= 6;
int c2 = combined & 0x3f;
combined >>>= 6;
int c1 = combined & 0x3f;
combined >>>= 6;
int c0 = combined & 0x3f;
// Translate into the equivalent alpha character
// emitting them in big-endian order.
sb.append(valueToChar[c0]);
sb.append(valueToChar[c1]);
sb.append(valueToChar[c2]);
sb.append(valueToChar[c3]);
}
// deal with leftover bytes
switch (leftover){
case 0:
default:
// nothing to do
break;
case 1:
// One leftover byte generates xx==
// Start a new line if next 4 chars won't fit on the current line
linePos += 4;
if (linePos > lineLength){
if (lineLength != 0){
sb.append(lineSeparator);
}
linePos = 4;
}
// Handle this recursively with a faked complete triple.
// Throw away last two chars and replace with ==
sb.append(encode(new byte[]{b[len],0,0}
).substring(0,2));
sb.append("==");
break;
case 2:
// Two leftover bytes generates xxx=
// Start a new line if next 4 chars won't fit on the current line
linePos += 4;
if (linePos > lineLength){
if (lineLength != 0){
sb.append(lineSeparator);
}
linePos = 4;
}
// Handle this recursively with a faked complete triple.
// Throw away last char and replace with =
sb.append(encode(new byte[]{b[len],b[len + 1],0}
).substring(0,3));
sb.append("=");
break;
} // end switch;
if (outputLength != sb.length()){
System.out.println("oops: minor program flaw: output length mis-estimated");
System.out.println("estimate:" + outputLength);
System.out.println("actual:" + sb.length());
}
return sb.toString();
} // end encode
/**
* decode a well-formed complete Base64 string back into an array of bytes.
* It must have an even multiple of 4 data characters (not counting \n),
* padded out with = as needed.
*/
public byte[] decode (String s){
// estimate worst case size of output array, no embedded newlines.
byte[] b = new byte[(s.length() / 4) * 3];
// tracks where we are in a cycle of 4 input chars.
int cycle = 0;
// where we combine 4 groups of 6 bits and take apart as 3 groups of 8.
int combined = 0;
// how many bytes we have prepared.
int j = 0;
// will be an even multiple of 4 chars, plus some embedded \n
int len = s.length();
int dummies = 0;
for (int i = 0;i < len;i++){
int c = s.charAt(i);
int value = (c <= 255) ? charToValue[c] : IGNORE;
// there are two magic values PAD (=) and IGNORE.
switch (value){
case IGNORE:
// e.g. \n, just ignore it.
break;
case PAD:
value = 0;
dummies++;
// fallthrough
default:
/* regular value character */
switch (cycle){
case 0:
combined = value;
cycle = 1;
break;
case 1:
combined <<= 6;
combined |= value;
cycle = 2;
break;
case 2:
combined <<= 6;
combined |= value;
cycle = 3;
break;
case 3:
combined <<= 6;
combined |= value;
// we have just completed a cycle of 4 chars.
// the four 6-bit values are in combined in big-endian order
// peel them off 8 bits at a time working lsb to msb
// to get our original 3 8-bit bytes back
b[j + 2] = (byte)combined;
combined >>>= 8;
b[j + 1] = (byte)combined;
combined >>>= 8;
b[j] = (byte)combined;
j += 3;
cycle = 0;
break;
}
break;
}
} // end for
if (cycle != 0){
throw new ArrayIndexOutOfBoundsException(
"Input to decode not an even multiple of 4 characters; pad with =.");
}
j -= dummies;
if (b.length != j){
byte[] b2 = new byte[j];
System.arraycopy(b,0,b2,0,j);
b = b2;
}
return b;
} // end decode
/**
* determines how long the lines are that are generated by encode.
* Ignored by decode.
* @param length 0 means no newlines inserted. Must be a multiple of 4.
*/
public void setLineLength (int length){
this.lineLength = (length / 4) * 4;
}
/**
* How lines are separated.
* Ignored by decode.
* @param lineSeparator may be "" but not null.
* Usually contains only a combination of chars \n and \r.
* Could be any chars not in set A-Z a-z 0-9 + /.
*/
public void setLineSeparator (String lineSeparator){
this.lineSeparator = lineSeparator;
}
/**
* letter of the alphabet used to encode binary values 0..63
*/
static final char[] valueToChar = new char[64];
/**
* binary value encoded by a given letter of the alphabet 0..63
*/
static final int[] charToValue = new int[256];
/**
* Marker value for chars we just ignore, e.g. \n \r high ascii
*/
static final int IGNORE = -1;
/**
* Marker for = trailing pad
*/
static final int PAD = -2;
static
/* initialise valueToChar and charToValue tables */{
// build translate valueToChar table only once.
// 0..25 -> 'A'..'Z'
for (int i = 0;i <= 25;i++){
valueToChar[i] = (char)('A' + i);
// 26..51 -> 'a'..'z'
}
for (int i = 0;i <= 25;i++){
valueToChar[i + 26] = (char)('a' + i);
// 52..61 -> '0'..'9'
}
for (int i = 0;i <= 9;i++){
valueToChar[i + 52] = (char)('0' + i);
}
valueToChar[62] = '+';
valueToChar[63] = '/';
// build translate charToValue table only once.
for (int i = 0;i < 256;i++){
charToValue[i] = IGNORE; // default is to ignore
}
for (int i = 0;i < 64;i++){
charToValue[valueToChar[i]] = i;
}
charToValue['='] = PAD;
}
/**
* used to disable test driver
*/
private static final boolean debug = false;
/**
* debug display array
*/
public static void show (byte[] b){
for (int i = 0;i < b.length;i++){
System.out.print(Integer.toHexString(b[i] & 0xff) + " ");
}
System.out.println();
}
/**
* debug display array
*/
public static void display (byte[] b){
for (int i = 0;i < b.length;i++){
System.out.print((char)b[i]);
}
System.out.println();
}
/**
* test driver
*/
public static void main (String[] args){
if (debug){
byte[] a = {(byte)0xfc,(byte)0x0f,(byte)0xc0};
byte[] b = {(byte)0x03,(byte)0xf0,(byte)0x3f};
byte[] c = {(byte)0x00,(byte)0x00,(byte)0x00};
byte[] d = {(byte)0xff,(byte)0xff,(byte)0xff};
byte[] e = {(byte)0xfc,(byte)0x0f,(byte)0xc0,(byte)1};
byte[] f = {(byte)0xfc,(byte)0x0f,(byte)0xc0,(byte)1,(byte)2};
byte[] g = {(byte)0xfc,(byte)0x0f,(byte)0xc0,(byte)1,(byte)2,(byte)3};
byte[] h = "AAAAAAAAAAB".getBytes();
show(a);
show(b);
show(c);
show(d);
show(e);
show(f);
show(g);
show(h);
Base64 b64 = new Base64();
show(b64.decode(b64.encode(a)));
show(b64.decode(b64.encode(b)));
show(b64.decode(b64.encode(c)));
show(b64.decode(b64.encode(d)));
show(b64.decode(b64.encode(e)));
show(b64.decode(b64.encode(f)));
show(b64.decode(b64.encode(g)));
show(b64.decode(b64.encode(h)));
b64.setLineLength(8);
show((b64.encode(h)).getBytes());
}
} // end main
} // end Base64
382
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
