The padding '=' characters at the end are considered optional, but * if any are present, there must be the correct number of them. * * @param str the input String to decode, which is converted to @@ -136,7 +122,7 @@ public class Base64 { * Decode the Base64-encoded data in input and return the data in * a new byte array. * - * The padding '=' characters at the end are considered optional, but + *
The padding '=' characters at the end are considered optional, but * if any are present, there must be the correct number of them. * * @param input the input array to decode @@ -154,7 +140,7 @@ public class Base64 { * Decode the Base64-encoded data in input and return the data in * a new byte array. * - * The padding '=' characters at the end are considered optional, but + *
The padding '=' characters at the end are considered optional, but * if any are present, there must be the correct number of them. * * @param input the data to decode @@ -169,121 +155,172 @@ public class Base64 { public static byte[] decode(byte[] input, int offset, int len, int flags) { // Allocate space for the most data the input could represent. // (It could contain less if it contains whitespace, etc.) - DecoderState state = new DecoderState(flags, new byte[len*3/4]); + Decoder decoder = new Decoder(flags, new byte[len*3/4]); - if (!decodeInternal(input, offset, len, state, true)) { + if (!decoder.process(input, offset, len, true)) { throw new IllegalArgumentException("bad base-64"); } // Maybe we got lucky and allocated exactly enough output space. - if (state.op == state.output.length) { - return state.output; + if (decoder.op == decoder.output.length) { + return decoder.output; } // Need to shorten the array, so allocate a new one of the // right size and copy. - byte[] temp = new byte[state.op]; - System.arraycopy(state.output, 0, temp, 0, state.op); + byte[] temp = new byte[decoder.op]; + System.arraycopy(decoder.output, 0, temp, 0, decoder.op); return temp; } - /* package */ static class DecoderState { - public byte[] output; - public int op; + /* package */ static class Decoder extends Coder { + /** + * Lookup table for turning bytes into their position in the + * Base64 alphabet. + */ + private static final int DECODE[] = { + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, -1, -1, 63, + 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, -2, -1, -1, + -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, -1, -1, -1, + -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, + 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + }; - public int state; // state number (0 to 6) - public int value; + /** + * Decode lookup table for the "web safe" variant (RFC 3548 + * sec. 4) where - and _ replace + and /. + */ + private static final int DECODE_WEBSAFE[] = { + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, -1, + 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, -2, -1, -1, + -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, -1, -1, 63, + -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, + 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + }; - final public int[] alphabet; + /** Non-data values in the DECODE arrays. */ + private static final int SKIP = -1; + private static final int EQUALS = -2; - public DecoderState(int flags, byte[] output) { + /** + * States 0-3 are reading through the next input tuple. + * State 4 is having read one '=' and expecting exactly + * one more. + * State 5 is expecting no more data or padding characters + * in the input. + * State 6 is the error state; an error has been detected + * in the input and no future input can "fix" it. + */ + private int state; // state number (0 to 6) + private int value; + + final private int[] alphabet; + + public Decoder(int flags, byte[] output) { this.output = output; - alphabet = ((flags & WEB_SAFE) == 0) ? DECODE : DECODE_WEBSAFE; + alphabet = ((flags & URL_SAFE) == 0) ? DECODE : DECODE_WEBSAFE; state = 0; value = 0; } - } - /** - * Decode another block of input data. - * - * @param dstate a DecoderState object whose (caller-provided) - * output array is big enough to hold all the decoded data. - * On return, dstate.op will be set to the length of the - * decoded data. - * @param finish true if this is the final call to decodeInternal - * with the given DecoderState object. Will finalize the - * decoder state and include any final bytes in the output. - * - * @return true if the state machine is still healthy. false if - * bad base-64 data has been detected in the input stream. - */ + /** + * @return an overestimate for the number of bytes {@code + * len} bytes could decode to. + */ + public int maxOutputSize(int len) { + return len * 3/4 + 10; + } - /* package */ static boolean decodeInternal( - byte[] input, int offset, int len, final DecoderState dstate, boolean finish) { - if (dstate.state == 6) return false; + /** + * Decode another block of input data. + * + * @return true if the state machine is still healthy. false if + * bad base-64 data has been detected in the input stream. + */ + public boolean process(byte[] input, int offset, int len, boolean finish) { + if (this.state == 6) return false; - int state = dstate.state; - int value = dstate.value; - final int[] decode = dstate.alphabet; - final byte[] output = dstate.output; - int op = 0; + int p = offset; + len += offset; - int p = offset; - len += offset; + // Using local variables makes the decoder about 12% + // faster than if we manipulate the member variables in + // the loop. (Even alphabet makes a measurable + // difference, which is somewhat surprising to me since + // the member variable is final.) + int state = this.state; + int value = this.value; + int op = 0; + final byte[] output = this.output; + final int[] alphabet = this.alphabet; - while (p < len) { + while (p < len) { + // Try the fast path: we're starting a new tuple and the + // next four bytes of the input stream are all data + // bytes. This corresponds to going through states + // 0-1-2-3-0. We expect to use this method for most of + // the data. + // + // If any of the next four bytes of input are non-data + // (whitespace, etc.), value will end up negative. (All + // the non-data values in decode are small negative + // numbers, so shifting any of them up and or'ing them + // together will result in a value with its top bit set.) + // + // You can remove this whole block and the output should + // be the same, just slower. + if (state == 0) { + while (p+4 <= len && + (value = ((alphabet[input[p] & 0xff] << 18) | + (alphabet[input[p+1] & 0xff] << 12) | + (alphabet[input[p+2] & 0xff] << 6) | + (alphabet[input[p+3] & 0xff]))) >= 0) { + output[op+2] = (byte) value; + output[op+1] = (byte) (value >> 8); + output[op] = (byte) (value >> 16); + op += 3; + p += 4; + } + if (p >= len) break; + } - // Try the fast path: we're starting a new tuple and the - // next four bytes of the input stream are all data - // bytes. This corresponds to going through states - // 0-1-2-3-0. We expect to use this method for most of - // the data. - // - // If any of the next four bytes of input are non-data - // (whitespace, etc.), value will end up negative. (All - // the non-data values in decode are small negative - // numbers, so shifting any of them up and or'ing them - // together will result in a value with its top bit set.) - // - // You can remove this whole block and the output should - // be the same, just slower. - if (state == 0 && p+4 <= len && - (value = ((decode[input[p] & 0xff] << 18) | - (decode[input[p+1] & 0xff] << 12) | - (decode[input[p+2] & 0xff] << 6) | - (decode[input[p+3] & 0xff]))) >= 0) { - output[op+2] = (byte) value; - output[op+1] = (byte) (value >> 8); - output[op] = (byte) (value >> 16); - op += 3; - p += 4; - continue; - } + // The fast path isn't available -- either we've read a + // partial tuple, or the next four input bytes aren't all + // data, or whatever. Fall back to the slower state + // machine implementation. - // The fast path isn't available -- either we've read a - // partial tuple, or the next four input bytes aren't all - // data, or whatever. Fall back to the slower state - // machine implementation. - // - // States 0-3 are reading through the next input tuple. - // State 4 is having read one '=' and expecting exactly - // one more. - // State 5 is expecting no more data or padding characters - // in the input. - // State 6 is the error state; an error has been detected - // in the input and no future input can "fix" it. + int d = alphabet[input[p++] & 0xff]; - int d = decode[input[p++] & 0xff]; - - switch (state) { + switch (state) { case 0: if (d >= 0) { value = d; ++state; } else if (d != SKIP) { - dstate.state = 6; + this.state = 6; return false; } break; @@ -293,7 +330,7 @@ public class Base64 { value = (value << 6) | d; ++state; } else if (d != SKIP) { - dstate.state = 6; + this.state = 6; return false; } break; @@ -308,7 +345,7 @@ public class Base64 { output[op++] = (byte) (value >> 4); state = 4; } else if (d != SKIP) { - dstate.state = 6; + this.state = 6; return false; } break; @@ -330,7 +367,7 @@ public class Base64 { op += 2; state = 5; } else if (d != SKIP) { - dstate.state = 6; + this.state = 6; return false; } break; @@ -339,41 +376,40 @@ public class Base64 { if (d == EQUALS) { ++state; } else if (d != SKIP) { - dstate.state = 6; + this.state = 6; return false; } break; case 5: if (d != SKIP) { - dstate.state = 6; + this.state = 6; return false; } break; + } } - } - if (!finish) { - // We're out of input, but a future call could provide - // more. Return the output we've produced on this call - // and save the current state of the state machine. - dstate.state = state; - dstate.value = value; - dstate.op = op; - return true; - } + if (!finish) { + // We're out of input, but a future call could provide + // more. + this.state = state; + this.value = value; + this.op = op; + return true; + } - // Done reading input. Now figure out where we are left in - // the state machine and finish up. + // Done reading input. Now figure out where we are left in + // the state machine and finish up. - switch (state) { + switch (state) { case 0: // Output length is a multiple of three. Fine. break; case 1: // Read one extra input byte, which isn't enough to // make another output byte. Illegal. - dstate.state = 6; + this.state = 6; return false; case 2: // Read two extra input bytes, enough to emit 1 more @@ -383,65 +419,29 @@ public class Base64 { case 3: // Read three extra input bytes, enough to emit 2 more // output bytes. Fine. - output[op+1] = (byte) (value >> 2); - output[op] = (byte) (value >> 10); - op += 2; + output[op++] = (byte) (value >> 10); + output[op++] = (byte) (value >> 2); break; case 4: // Read one padding '=' when we expected 2. Illegal. - dstate.state = 6; + this.state = 6; return false; case 5: // Read all the padding '='s we expected and no more. // Fine. break; - } + } - dstate.op = op; - return true; + this.state = state; + this.op = op; + return true; + } } // -------------------------------------------------------- // encoding // -------------------------------------------------------- - /** - * Emit a new line every this many output tuples. Corresponds to - * a 76-character line length (the maximum allowable according to - * RFC 2045). - */ - private static final int LINE_GROUPS = 19; - - /** - * Lookup table for turning Base64 alphabet positions (6 bits) - * into output bytes. - */ - private static final byte ENCODE[] = { - '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', '+', '/', - }; - - /** - * Lookup table for turning Base64 alphabet positions (6 bits) - * into output bytes. - */ - private static final byte ENCODE_WEBSAFE[] = { - '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', '-', '_', - }; - /** * Base64-encode the given data and return a newly allocated * String with the result. @@ -452,7 +452,12 @@ public class Base64 { * adheres to RFC 2045. */ public static String encodeToString(byte[] input, int flags) { - return new String(encode(input, flags)); + try { + return new String(encode(input, flags), "US-ASCII"); + } catch (UnsupportedEncodingException e) { + // US-ASCII is guaranteed to be available. + throw new AssertionError(e); + } } /** @@ -468,7 +473,12 @@ public class Base64 { * adheres to RFC 2045. */ public static String encodeToString(byte[] input, int offset, int len, int flags) { - return new String(encode(input, offset, len, flags)); + try { + return new String(encode(input, offset, len, flags), "US-ASCII"); + } catch (UnsupportedEncodingException e) { + // US-ASCII is guaranteed to be available. + throw new AssertionError(e); + } } /** @@ -497,13 +507,13 @@ public class Base64 { * adheres to RFC 2045. */ public static byte[] encode(byte[] input, int offset, int len, int flags) { - EncoderState state = new EncoderState(flags, null); + Encoder encoder = new Encoder(flags, null); // Compute the exact length of the array we will produce. int output_len = len / 3 * 4; // Account for the tail of the data and the padding bytes, if any. - if (state.do_padding) { + if (encoder.do_padding) { if (len % 3 > 0) { output_len += 4; } @@ -516,190 +526,215 @@ public class Base64 { } // Account for the newlines, if any. - if (state.do_newline && len > 0) { - output_len += (((len-1) / (3 * LINE_GROUPS)) + 1) * (state.do_cr ? 2 : 1); + if (encoder.do_newline && len > 0) { + output_len += (((len-1) / (3 * Encoder.LINE_GROUPS)) + 1) * + (encoder.do_cr ? 2 : 1); } - state.output = new byte[output_len]; - encodeInternal(input, offset, len, state, true); + encoder.output = new byte[output_len]; + encoder.process(input, offset, len, true); - assert state.op == output_len; + assert encoder.op == output_len; - return state.output; + return encoder.output; } - /* package */ static class EncoderState { - public byte[] output; - public int op; + /* package */ static class Encoder extends Coder { + /** + * Emit a new line every this many output tuples. Corresponds to + * a 76-character line length (the maximum allowable according to + * RFC 2045). + */ + public static final int LINE_GROUPS = 19; - final public byte[] tail; - public int tailLen; - public int count; + /** + * Lookup table for turning Base64 alphabet positions (6 bits) + * into output bytes. + */ + private static final byte ENCODE[] = { + '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', '+', '/', + }; + + /** + * Lookup table for turning Base64 alphabet positions (6 bits) + * into output bytes. + */ + private static final byte ENCODE_WEBSAFE[] = { + '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', '-', '_', + }; + + final private byte[] tail; + /* package */ int tailLen; + private int count; final public boolean do_padding; final public boolean do_newline; final public boolean do_cr; - final public byte[] alphabet; + final private byte[] alphabet; - public EncoderState(int flags, byte[] output) { + public Encoder(int flags, byte[] output) { this.output = output; do_padding = (flags & NO_PADDING) == 0; do_newline = (flags & NO_WRAP) == 0; do_cr = (flags & CRLF) != 0; - alphabet = ((flags & WEB_SAFE) == 0) ? ENCODE : ENCODE_WEBSAFE; + alphabet = ((flags & URL_SAFE) == 0) ? ENCODE : ENCODE_WEBSAFE; tail = new byte[2]; tailLen = 0; count = do_newline ? LINE_GROUPS : -1; } - } - /** - * Encode another block of input data. - * - * @param estate an EncoderState object whose (caller-provided) - * output array is big enough to hold all the encoded data. - * On return, estate.op will be set to the length of the - * encoded data. - * @param finish true if this is the final call to encodeInternal - * with the given EncoderState object. Will finalize the - * encoder state and include any final bytes in the output. - */ - static void encodeInternal(byte[] input, int offset, int len, - final EncoderState estate, boolean finish) { - final boolean do_cr = estate.do_cr; - final boolean do_newline = estate.do_newline; - final boolean do_padding = estate.do_padding; - final byte[] output = estate.output; - - int op = 0; - - int p = offset; - len += offset; - int v = -1; - int count = estate.count; - - // First we need to concatenate the tail of the previous call - // with any input bytes available now and see if we can empty - // the tail. - - switch (estate.tailLen) { - case 0: - // There was no tail. - break; - - case 1: - if (p+2 <= len) { - // A 1-byte tail with at least 2 bytes of - // input available now. - v = ((estate.tail[0] & 0xff) << 16) | - ((input[p++] & 0xff) << 8) | - (input[p++] & 0xff); - estate.tailLen = 0; - }; - break; - - case 2: - if (p+1 <= len) { - // A 2-byte tail with at least 1 byte of input. - v = ((estate.tail[0] & 0xff) << 16) | - ((estate.tail[1] & 0xff) << 8) | - (input[p++] & 0xff); - estate.tailLen = 0; - } - break; + /** + * @return an overestimate for the number of bytes {@code + * len} bytes could encode to. + */ + public int maxOutputSize(int len) { + return len * 8/5 + 10; } - if (v != -1) { - output[op++] = estate.alphabet[(v >> 18) & 0x3f]; - output[op++] = estate.alphabet[(v >> 12) & 0x3f]; - output[op++] = estate.alphabet[(v >> 6) & 0x3f]; - output[op++] = estate.alphabet[v & 0x3f]; - if (--count == 0) { - if (do_cr) output[op++] = '\r'; - output[op++] = '\n'; - count = LINE_GROUPS; + public boolean process(byte[] input, int offset, int len, boolean finish) { + // Using local variables makes the encoder about 9% faster. + final byte[] alphabet = this.alphabet; + final byte[] output = this.output; + int op = 0; + int count = this.count; + + int p = offset; + len += offset; + int v = -1; + + // First we need to concatenate the tail of the previous call + // with any input bytes available now and see if we can empty + // the tail. + + switch (tailLen) { + case 0: + // There was no tail. + break; + + case 1: + if (p+2 <= len) { + // A 1-byte tail with at least 2 bytes of + // input available now. + v = ((tail[0] & 0xff) << 16) | + ((input[p++] & 0xff) << 8) | + (input[p++] & 0xff); + tailLen = 0; + }; + break; + + case 2: + if (p+1 <= len) { + // A 2-byte tail with at least 1 byte of input. + v = ((tail[0] & 0xff) << 16) | + ((tail[1] & 0xff) << 8) | + (input[p++] & 0xff); + tailLen = 0; + } + break; } - } - // At this point either there is no tail, or there are fewer - // than 3 bytes of input available. - - // The main loop, turning 3 input bytes into 4 output bytes on - // each iteration. - while (p+3 <= len) { - v = ((input[p++] & 0xff) << 16) | - ((input[p++] & 0xff) << 8) | - (input[p++] & 0xff); - output[op++] = estate.alphabet[(v >> 18) & 0x3f]; - output[op++] = estate.alphabet[(v >> 12) & 0x3f]; - output[op++] = estate.alphabet[(v >> 6) & 0x3f]; - output[op++] = estate.alphabet[v & 0x3f]; - if (--count == 0) { - if (do_cr) output[op++] = '\r'; - output[op++] = '\n'; - count = LINE_GROUPS; - } - } - - if (finish) { - // Finish up the tail of the input. Note that we need to - // consume any bytes in estate.tail before any bytes - // remaining in input; there should be at most two bytes - // total. - - if (p-estate.tailLen == len-1) { - int t = 0; - v = ((estate.tailLen > 0 ? estate.tail[t++] : input[p++]) & 0xff) << 4; - estate.tailLen -= t; - output[op++] = estate.alphabet[(v >> 6) & 0x3f]; - output[op++] = estate.alphabet[v & 0x3f]; - if (do_padding) { - output[op++] = '='; - output[op++] = '='; + if (v != -1) { + output[op++] = alphabet[(v >> 18) & 0x3f]; + output[op++] = alphabet[(v >> 12) & 0x3f]; + output[op++] = alphabet[(v >> 6) & 0x3f]; + output[op++] = alphabet[v & 0x3f]; + if (--count == 0) { + if (do_cr) output[op++] = '\r'; + output[op++] = '\n'; + count = LINE_GROUPS; } - if (do_newline) { + } + + // At this point either there is no tail, or there are fewer + // than 3 bytes of input available. + + // The main loop, turning 3 input bytes into 4 output bytes on + // each iteration. + while (p+3 <= len) { + v = ((input[p] & 0xff) << 16) | + ((input[p+1] & 0xff) << 8) | + (input[p+2] & 0xff); + output[op] = alphabet[(v >> 18) & 0x3f]; + output[op+1] = alphabet[(v >> 12) & 0x3f]; + output[op+2] = alphabet[(v >> 6) & 0x3f]; + output[op+3] = alphabet[v & 0x3f]; + p += 3; + op += 4; + if (--count == 0) { + if (do_cr) output[op++] = '\r'; + output[op++] = '\n'; + count = LINE_GROUPS; + } + } + + if (finish) { + // Finish up the tail of the input. Note that we need to + // consume any bytes in tail before any bytes + // remaining in input; there should be at most two bytes + // total. + + if (p-tailLen == len-1) { + int t = 0; + v = ((tailLen > 0 ? tail[t++] : input[p++]) & 0xff) << 4; + tailLen -= t; + output[op++] = alphabet[(v >> 6) & 0x3f]; + output[op++] = alphabet[v & 0x3f]; + if (do_padding) { + output[op++] = '='; + output[op++] = '='; + } + if (do_newline) { + if (do_cr) output[op++] = '\r'; + output[op++] = '\n'; + } + } else if (p-tailLen == len-2) { + int t = 0; + v = (((tailLen > 1 ? tail[t++] : input[p++]) & 0xff) << 10) | + (((tailLen > 0 ? tail[t++] : input[p++]) & 0xff) << 2); + tailLen -= t; + output[op++] = alphabet[(v >> 12) & 0x3f]; + output[op++] = alphabet[(v >> 6) & 0x3f]; + output[op++] = alphabet[v & 0x3f]; + if (do_padding) { + output[op++] = '='; + } + if (do_newline) { + if (do_cr) output[op++] = '\r'; + output[op++] = '\n'; + } + } else if (do_newline && op > 0 && count != LINE_GROUPS) { if (do_cr) output[op++] = '\r'; output[op++] = '\n'; } - } else if (p-estate.tailLen == len-2) { - int t = 0; - v = (((estate.tailLen > 1 ? estate.tail[t++] : input[p++]) & 0xff) << 10) | - (((estate.tailLen > 0 ? estate.tail[t++] : input[p++]) & 0xff) << 2); - estate.tailLen -= t; - output[op++] = estate.alphabet[(v >> 12) & 0x3f]; - output[op++] = estate.alphabet[(v >> 6) & 0x3f]; - output[op++] = estate.alphabet[v & 0x3f]; - if (do_padding) { - output[op++] = '='; + + assert tailLen == 0; + assert p == len; + } else { + // Save the leftovers in tail to be consumed on the next + // call to encodeInternal. + + if (p == len-1) { + tail[tailLen++] = input[p]; + } else if (p == len-2) { + tail[tailLen++] = input[p]; + tail[tailLen++] = input[p+1]; } - if (do_newline) { - if (do_cr) output[op++] = '\r'; - output[op++] = '\n'; - } - } else if (do_newline && op > 0 && count != LINE_GROUPS) { - if (do_cr) output[op++] = '\r'; - output[op++] = '\n'; } - assert estate.tailLen == 0; - assert p == len; - } else { - // Save the leftovers in tail to be consumed on the next - // call to encodeInternal. + this.op = op; + this.count = count; - if (p == len-1) { - estate.tail[estate.tailLen++] = input[p]; - } else if (p == len-2) { - estate.tail[estate.tailLen++] = input[p]; - estate.tail[estate.tailLen++] = input[p+1]; - } + return true; } - - estate.op = op; - estate.count = count; } private Base64() { } // don't instantiate diff --git a/core/java/android/util/base64/Base64InputStream.java b/core/java/android/util/base64/Base64InputStream.java index 646bbdb84e8fd..935939e097c50 100644 --- a/core/java/android/util/base64/Base64InputStream.java +++ b/core/java/android/util/base64/Base64InputStream.java @@ -25,16 +25,13 @@ import java.io.InputStream; * it. */ public class Base64InputStream extends FilterInputStream { - private final boolean encode; - private final Base64.EncoderState estate; - private final Base64.DecoderState dstate; + private final Base64.Coder coder; private static byte[] EMPTY = new byte[0]; private static final int BUFFER_SIZE = 2048; private boolean eof; private byte[] inputBuffer; - private byte[] outputBuffer; private int outputStart; private int outputEnd; @@ -63,22 +60,14 @@ public class Base64InputStream extends FilterInputStream { */ public Base64InputStream(InputStream in, int flags, boolean encode) { super(in); - this.encode = encode; eof = false; inputBuffer = new byte[BUFFER_SIZE]; if (encode) { - // len*8/5+10 is an overestimate of the most bytes the - // encoder can produce for len bytes of input. - outputBuffer = new byte[BUFFER_SIZE * 8/5 + 10]; - estate = new Base64.EncoderState(flags, outputBuffer); - dstate = null; + coder = new Base64.Encoder(flags, null); } else { - // len*3/4+10 is an overestimate of the most bytes the - // decoder can produce for len bytes of input. - outputBuffer = new byte[BUFFER_SIZE * 3/4 + 10]; - estate = null; - dstate = new Base64.DecoderState(flags, outputBuffer); + coder = new Base64.Decoder(flags, null); } + coder.output = new byte[coder.maxOutputSize(BUFFER_SIZE)]; outputStart = 0; outputEnd = 0; } @@ -123,7 +112,7 @@ public class Base64InputStream extends FilterInputStream { if (outputStart >= outputEnd) { return -1; } else { - return outputBuffer[outputStart++]; + return coder.output[outputStart++]; } } @@ -135,36 +124,30 @@ public class Base64InputStream extends FilterInputStream { return -1; } int bytes = Math.min(len, outputEnd-outputStart); - System.arraycopy(outputBuffer, outputStart, b, off, bytes); + System.arraycopy(coder.output, outputStart, b, off, bytes); outputStart += bytes; return bytes; } /** * Read data from the input stream into inputBuffer, then - * decode/encode it into the empty outputBuffer, and reset the + * decode/encode it into the empty coder.output, and reset the * outputStart and outputEnd pointers. */ private void refill() throws IOException { if (eof) return; int bytesRead = in.read(inputBuffer); - if (encode) { - if (bytesRead == -1) { - eof = true; - Base64.encodeInternal(EMPTY, 0, 0, estate, true); - } else { - Base64.encodeInternal(inputBuffer, 0, bytesRead, estate, false); - } - outputEnd = estate.op; + boolean success; + if (bytesRead == -1) { + eof = true; + success = coder.process(EMPTY, 0, 0, true); } else { - if (bytesRead == -1) { - eof = true; - Base64.decodeInternal(EMPTY, 0, 0, dstate, true); - } else { - Base64.decodeInternal(inputBuffer, 0, bytesRead, dstate, false); - } - outputEnd = dstate.op; + success = coder.process(inputBuffer, 0, bytesRead, false); } + if (!success) { + throw new IOException("bad base-64"); + } + outputEnd = coder.op; outputStart = 0; } } diff --git a/core/java/android/util/base64/Base64OutputStream.java b/core/java/android/util/base64/Base64OutputStream.java index 8edb511de80eb..35e9a2b8cf847 100644 --- a/core/java/android/util/base64/Base64OutputStream.java +++ b/core/java/android/util/base64/Base64OutputStream.java @@ -25,9 +25,7 @@ import java.io.OutputStream; * it, writing the resulting data to another OutputStream. */ public class Base64OutputStream extends FilterOutputStream { - private final boolean encode; - private final Base64.EncoderState estate; - private final Base64.DecoderState dstate; + private final Base64.Coder coder; private final int flags; private byte[] buffer = null; @@ -62,13 +60,10 @@ public class Base64OutputStream extends FilterOutputStream { public Base64OutputStream(OutputStream out, int flags, boolean encode) { super(out); this.flags = flags; - this.encode = encode; if (encode) { - estate = new Base64.EncoderState(flags, null); - dstate = null; + coder = new Base64.Encoder(flags, null); } else { - estate = null; - dstate = new Base64.DecoderState(flags, null); + coder = new Base64.Decoder(flags, null); } } @@ -107,12 +102,28 @@ public class Base64OutputStream extends FilterOutputStream { } public void close() throws IOException { - flushBuffer(); - internalWrite(EMPTY, 0, 0, true); - if ((flags & Base64.NO_CLOSE) == 0) { - out.close(); - } else { - out.flush(); + IOException thrown = null; + try { + flushBuffer(); + internalWrite(EMPTY, 0, 0, true); + } catch (IOException e) { + thrown = e; + } + + try { + if ((flags & Base64.NO_CLOSE) == 0) { + out.close(); + } else { + out.flush(); + } + } catch (IOException e) { + if (thrown != null) { + thrown = e; + } + } + + if (thrown != null) { + throw thrown; } } @@ -123,21 +134,11 @@ public class Base64OutputStream extends FilterOutputStream { * encoder/decoder state to be finalized. */ private void internalWrite(byte[] b, int off, int len, boolean finish) throws IOException { - if (encode) { - // len*8/5+10 is an overestimate of the most bytes the - // encoder can produce for len bytes of input. - estate.output = embiggen(estate.output, len*8/5+10); - Base64.encodeInternal(b, off, len, estate, finish); - out.write(estate.output, 0, estate.op); - } else { - // len*3/4+10 is an overestimate of the most bytes the - // decoder can produce for len bytes of input. - dstate.output = embiggen(dstate.output, len*3/4+10); - if (!Base64.decodeInternal(b, off, len, dstate, finish)) { - throw new IOException("bad base-64"); - } - out.write(dstate.output, 0, dstate.op); + coder.output = embiggen(coder.output, coder.maxOutputSize(len)); + if (!coder.process(b, off, len, finish)) { + throw new IOException("bad base-64"); } + out.write(coder.output, 0, coder.op); } /** diff --git a/core/tests/coretests/src/android/util/base64/Base64Test.java b/core/tests/coretests/src/android/util/base64/Base64Test.java index 8f98b08dcf6d7..48192bb4fa08a 100644 --- a/core/tests/coretests/src/android/util/base64/Base64Test.java +++ b/core/tests/coretests/src/android/util/base64/Base64Test.java @@ -134,25 +134,25 @@ public class Base64Test extends TestCase { } public void testWebSafe() throws Exception { - assertEquals(BYTES, 0, Base64.decode("", Base64.WEB_SAFE)); - assertEquals(BYTES, 1, Base64.decode("_w==", Base64.WEB_SAFE)); - assertEquals(BYTES, 2, Base64.decode("_-4=", Base64.WEB_SAFE)); - assertEquals(BYTES, 3, Base64.decode("_-7d", Base64.WEB_SAFE)); - assertEquals(BYTES, 4, Base64.decode("_-7dzA==", Base64.WEB_SAFE)); - assertEquals(BYTES, 5, Base64.decode("_-7dzLs=", Base64.WEB_SAFE)); - assertEquals(BYTES, 6, Base64.decode("_-7dzLuq", Base64.WEB_SAFE)); - assertEquals(BYTES, 7, Base64.decode("_-7dzLuqmQ==", Base64.WEB_SAFE)); - assertEquals(BYTES, 8, Base64.decode("_-7dzLuqmYg=", Base64.WEB_SAFE)); + assertEquals(BYTES, 0, Base64.decode("", Base64.URL_SAFE)); + assertEquals(BYTES, 1, Base64.decode("_w==", Base64.URL_SAFE)); + assertEquals(BYTES, 2, Base64.decode("_-4=", Base64.URL_SAFE)); + assertEquals(BYTES, 3, Base64.decode("_-7d", Base64.URL_SAFE)); + assertEquals(BYTES, 4, Base64.decode("_-7dzA==", Base64.URL_SAFE)); + assertEquals(BYTES, 5, Base64.decode("_-7dzLs=", Base64.URL_SAFE)); + assertEquals(BYTES, 6, Base64.decode("_-7dzLuq", Base64.URL_SAFE)); + assertEquals(BYTES, 7, Base64.decode("_-7dzLuqmQ==", Base64.URL_SAFE)); + assertEquals(BYTES, 8, Base64.decode("_-7dzLuqmYg=", Base64.URL_SAFE)); - assertEquals("", Base64.encodeToString(BYTES, 0, 0, Base64.WEB_SAFE)); - assertEquals("_w==\n", Base64.encodeToString(BYTES, 0, 1, Base64.WEB_SAFE)); - assertEquals("_-4=\n", Base64.encodeToString(BYTES, 0, 2, Base64.WEB_SAFE)); - assertEquals("_-7d\n", Base64.encodeToString(BYTES, 0, 3, Base64.WEB_SAFE)); - assertEquals("_-7dzA==\n", Base64.encodeToString(BYTES, 0, 4, Base64.WEB_SAFE)); - assertEquals("_-7dzLs=\n", Base64.encodeToString(BYTES, 0, 5, Base64.WEB_SAFE)); - assertEquals("_-7dzLuq\n", Base64.encodeToString(BYTES, 0, 6, Base64.WEB_SAFE)); - assertEquals("_-7dzLuqmQ==\n", Base64.encodeToString(BYTES, 0, 7, Base64.WEB_SAFE)); - assertEquals("_-7dzLuqmYg=\n", Base64.encodeToString(BYTES, 0, 8, Base64.WEB_SAFE)); + assertEquals("", Base64.encodeToString(BYTES, 0, 0, Base64.URL_SAFE)); + assertEquals("_w==\n", Base64.encodeToString(BYTES, 0, 1, Base64.URL_SAFE)); + assertEquals("_-4=\n", Base64.encodeToString(BYTES, 0, 2, Base64.URL_SAFE)); + assertEquals("_-7d\n", Base64.encodeToString(BYTES, 0, 3, Base64.URL_SAFE)); + assertEquals("_-7dzA==\n", Base64.encodeToString(BYTES, 0, 4, Base64.URL_SAFE)); + assertEquals("_-7dzLs=\n", Base64.encodeToString(BYTES, 0, 5, Base64.URL_SAFE)); + assertEquals("_-7dzLuq\n", Base64.encodeToString(BYTES, 0, 6, Base64.URL_SAFE)); + assertEquals("_-7dzLuqmQ==\n", Base64.encodeToString(BYTES, 0, 7, Base64.URL_SAFE)); + assertEquals("_-7dzLuqmYg=\n", Base64.encodeToString(BYTES, 0, 8, Base64.URL_SAFE)); } public void testFlags() throws Exception { @@ -227,55 +227,55 @@ public class Base64Test extends TestCase { } /** - * Tests that Base64.encodeInternal does correct handling of the + * Tests that Base64.Encoder.encode() does correct handling of the * tail for each call. * * This test is disabled because while it passes if you can get it * to run, android's test infrastructure currently doesn't allow - * us to get at package-private members (Base64.EncoderState in + * us to get at package-private members (Base64.Encoder in * this case). */ public void XXXtestEncodeInternal() throws Exception { byte[] input = { (byte) 0x61, (byte) 0x62, (byte) 0x63 }; byte[] output = new byte[100]; - Base64.EncoderState state = new Base64.EncoderState(Base64.NO_PADDING | Base64.NO_WRAP, - output); + Base64.Encoder encoder = new Base64.Encoder(Base64.NO_PADDING | Base64.NO_WRAP, + output); - Base64.encodeInternal(input, 0, 3, state, false); - assertEquals("YWJj".getBytes(), 4, state.output, state.op); - assertEquals(0, state.tailLen); + encoder.process(input, 0, 3, false); + assertEquals("YWJj".getBytes(), 4, encoder.output, encoder.op); + assertEquals(0, encoder.tailLen); - Base64.encodeInternal(input, 0, 3, state, false); - assertEquals("YWJj".getBytes(), 4, state.output, state.op); - assertEquals(0, state.tailLen); + encoder.process(input, 0, 3, false); + assertEquals("YWJj".getBytes(), 4, encoder.output, encoder.op); + assertEquals(0, encoder.tailLen); - Base64.encodeInternal(input, 0, 1, state, false); - assertEquals(0, state.op); - assertEquals(1, state.tailLen); + encoder.process(input, 0, 1, false); + assertEquals(0, encoder.op); + assertEquals(1, encoder.tailLen); - Base64.encodeInternal(input, 0, 1, state, false); - assertEquals(0, state.op); - assertEquals(2, state.tailLen); + encoder.process(input, 0, 1, false); + assertEquals(0, encoder.op); + assertEquals(2, encoder.tailLen); - Base64.encodeInternal(input, 0, 1, state, false); - assertEquals("YWFh".getBytes(), 4, state.output, state.op); - assertEquals(0, state.tailLen); + encoder.process(input, 0, 1, false); + assertEquals("YWFh".getBytes(), 4, encoder.output, encoder.op); + assertEquals(0, encoder.tailLen); - Base64.encodeInternal(input, 0, 2, state, false); - assertEquals(0, state.op); - assertEquals(2, state.tailLen); + encoder.process(input, 0, 2, false); + assertEquals(0, encoder.op); + assertEquals(2, encoder.tailLen); - Base64.encodeInternal(input, 0, 2, state, false); - assertEquals("YWJh".getBytes(), 4, state.output, state.op); - assertEquals(1, state.tailLen); + encoder.process(input, 0, 2, false); + assertEquals("YWJh".getBytes(), 4, encoder.output, encoder.op); + assertEquals(1, encoder.tailLen); - Base64.encodeInternal(input, 0, 2, state, false); - assertEquals("YmFi".getBytes(), 4, state.output, state.op); - assertEquals(0, state.tailLen); + encoder.process(input, 0, 2, false); + assertEquals("YmFi".getBytes(), 4, encoder.output, encoder.op); + assertEquals(0, encoder.tailLen); - Base64.encodeInternal(input, 0, 1, state, true); - assertEquals("YQ".getBytes(), 2, state.output, state.op); + encoder.process(input, 0, 1, true); + assertEquals("YQ".getBytes(), 2, encoder.output, encoder.op); } private static final String lipsum = @@ -307,7 +307,7 @@ public class Base64Test extends TestCase { Base64.NO_WRAP, Base64.NO_PADDING | Base64.NO_WRAP, Base64.CRLF, - Base64.WEB_SAFE }; + Base64.URL_SAFE }; int[] writeLengths = { -10, -5, -1, 0, 1, 1, 2, 2, 3, 10, 100 }; Random rng = new Random(32176L); @@ -414,7 +414,7 @@ public class Base64Test extends TestCase { Base64.NO_WRAP, Base64.NO_PADDING | Base64.NO_WRAP, Base64.CRLF, - Base64.WEB_SAFE }; + Base64.URL_SAFE }; int[] writeLengths = { -10, -5, -1, 0, 1, 1, 2, 2, 3, 10, 100 }; Random rng = new Random(32176L);