| 1 | /* |
| 2 | * Copyright (C) 2010 The Android Open Source Project |
| 3 | * |
| 4 | * Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 | * you may not use this file except in compliance with the License. |
| 6 | * You may obtain a copy of the License at |
| 7 | * |
| 8 | * http://www.apache.org/licenses/LICENSE-2.0 |
| 9 | * |
| 10 | * Unless required by applicable law or agreed to in writing, software |
| 11 | * distributed under the License is distributed on an "AS IS" BASIS, |
| 12 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 | * See the License for the specific language governing permissions and |
| 14 | * limitations under the License. |
| 15 | */ |
| 16 | |
| 17 | /* |
| 18 | * Changes (@author niki): |
| 19 | * - default charset -> UTF-8 |
| 20 | */ |
| 21 | |
| 22 | package be.nikiroo.utils.streams; |
| 23 | |
| 24 | import java.io.UnsupportedEncodingException; |
| 25 | |
| 26 | import be.nikiroo.utils.StringUtils; |
| 27 | |
| 28 | /** |
| 29 | * Utilities for encoding and decoding the Base64 representation of |
| 30 | * binary data. See RFCs <a |
| 31 | * href="http://www.ietf.org/rfc/rfc2045.txt">2045</a> and <a |
| 32 | * href="http://www.ietf.org/rfc/rfc3548.txt">3548</a>. |
| 33 | */ |
| 34 | class Base64 { |
| 35 | /** |
| 36 | * Default values for encoder/decoder flags. |
| 37 | */ |
| 38 | public static final int DEFAULT = 0; |
| 39 | |
| 40 | /** |
| 41 | * Encoder flag bit to omit the padding '=' characters at the end |
| 42 | * of the output (if any). |
| 43 | */ |
| 44 | public static final int NO_PADDING = 1; |
| 45 | |
| 46 | /** |
| 47 | * Encoder flag bit to omit all line terminators (i.e., the output |
| 48 | * will be on one long line). |
| 49 | */ |
| 50 | public static final int NO_WRAP = 2; |
| 51 | |
| 52 | /** |
| 53 | * Encoder flag bit to indicate lines should be terminated with a |
| 54 | * CRLF pair instead of just an LF. Has no effect if {@code |
| 55 | * NO_WRAP} is specified as well. |
| 56 | */ |
| 57 | public static final int CRLF = 4; |
| 58 | |
| 59 | /** |
| 60 | * Encoder/decoder flag bit to indicate using the "URL and |
| 61 | * filename safe" variant of Base64 (see RFC 3548 section 4) where |
| 62 | * {@code -} and {@code _} are used in place of {@code +} and |
| 63 | * {@code /}. |
| 64 | */ |
| 65 | public static final int URL_SAFE = 8; |
| 66 | |
| 67 | /** |
| 68 | * Flag to pass to {@link Base64OutputStream} to indicate that it |
| 69 | * should not close the output stream it is wrapping when it |
| 70 | * itself is closed. |
| 71 | */ |
| 72 | public static final int NO_CLOSE = 16; |
| 73 | |
| 74 | // -------------------------------------------------------- |
| 75 | // shared code |
| 76 | // -------------------------------------------------------- |
| 77 | |
| 78 | /* package */ static abstract class Coder { |
| 79 | public byte[] output; |
| 80 | public int op; |
| 81 | |
| 82 | /** |
| 83 | * Encode/decode another block of input data. this.output is |
| 84 | * provided by the caller, and must be big enough to hold all |
| 85 | * the coded data. On exit, this.opwill be set to the length |
| 86 | * of the coded data. |
| 87 | * |
| 88 | * @param finish true if this is the final call to process for |
| 89 | * this object. Will finalize the coder state and |
| 90 | * include any final bytes in the output. |
| 91 | * |
| 92 | * @return true if the input so far is good; false if some |
| 93 | * error has been detected in the input stream.. |
| 94 | */ |
| 95 | public abstract boolean process(byte[] input, int offset, int len, boolean finish); |
| 96 | |
| 97 | /** |
| 98 | * @return the maximum number of bytes a call to process() |
| 99 | * could produce for the given number of input bytes. This may |
| 100 | * be an overestimate. |
| 101 | */ |
| 102 | public abstract int maxOutputSize(int len); |
| 103 | } |
| 104 | |
| 105 | // -------------------------------------------------------- |
| 106 | // decoding |
| 107 | // -------------------------------------------------------- |
| 108 | |
| 109 | /** |
| 110 | * Decode the Base64-encoded data in input and return the data in |
| 111 | * a new byte array. |
| 112 | * |
| 113 | * <p>The padding '=' characters at the end are considered optional, but |
| 114 | * if any are present, there must be the correct number of them. |
| 115 | * |
| 116 | * @param str the input String to decode, which is converted to |
| 117 | * bytes using the default charset |
| 118 | * @param flags controls certain features of the decoded output. |
| 119 | * Pass {@code DEFAULT} to decode standard Base64. |
| 120 | * |
| 121 | * @throws IllegalArgumentException if the input contains |
| 122 | * incorrect padding |
| 123 | */ |
| 124 | public static byte[] decode(String str, int flags) { |
| 125 | return decode(StringUtils.getBytes(str), flags); |
| 126 | } |
| 127 | |
| 128 | /** |
| 129 | * Decode the Base64-encoded data in input and return the data in |
| 130 | * a new byte array. |
| 131 | * |
| 132 | * <p>The padding '=' characters at the end are considered optional, but |
| 133 | * if any are present, there must be the correct number of them. |
| 134 | * |
| 135 | * @param input the input array to decode |
| 136 | * @param flags controls certain features of the decoded output. |
| 137 | * Pass {@code DEFAULT} to decode standard Base64. |
| 138 | * |
| 139 | * @throws IllegalArgumentException if the input contains |
| 140 | * incorrect padding |
| 141 | */ |
| 142 | public static byte[] decode(byte[] input, int flags) { |
| 143 | return decode(input, 0, input.length, flags); |
| 144 | } |
| 145 | |
| 146 | /** |
| 147 | * Decode the Base64-encoded data in input and return the data in |
| 148 | * a new byte array. |
| 149 | * |
| 150 | * <p>The padding '=' characters at the end are considered optional, but |
| 151 | * if any are present, there must be the correct number of them. |
| 152 | * |
| 153 | * @param input the data to decode |
| 154 | * @param offset the position within the input array at which to start |
| 155 | * @param len the number of bytes of input to decode |
| 156 | * @param flags controls certain features of the decoded output. |
| 157 | * Pass {@code DEFAULT} to decode standard Base64. |
| 158 | * |
| 159 | * @throws IllegalArgumentException if the input contains |
| 160 | * incorrect padding |
| 161 | */ |
| 162 | public static byte[] decode(byte[] input, int offset, int len, int flags) { |
| 163 | // Allocate space for the most data the input could represent. |
| 164 | // (It could contain less if it contains whitespace, etc.) |
| 165 | Decoder decoder = new Decoder(flags, new byte[len*3/4]); |
| 166 | |
| 167 | if (!decoder.process(input, offset, len, true)) { |
| 168 | throw new IllegalArgumentException("bad base-64"); |
| 169 | } |
| 170 | |
| 171 | // Maybe we got lucky and allocated exactly enough output space. |
| 172 | if (decoder.op == decoder.output.length) { |
| 173 | return decoder.output; |
| 174 | } |
| 175 | |
| 176 | // Need to shorten the array, so allocate a new one of the |
| 177 | // right size and copy. |
| 178 | byte[] temp = new byte[decoder.op]; |
| 179 | System.arraycopy(decoder.output, 0, temp, 0, decoder.op); |
| 180 | return temp; |
| 181 | } |
| 182 | |
| 183 | /* package */ static class Decoder extends Coder { |
| 184 | /** |
| 185 | * Lookup table for turning bytes into their position in the |
| 186 | * Base64 alphabet. |
| 187 | */ |
| 188 | private static final int DECODE[] = { |
| 189 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 190 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 191 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, -1, -1, 63, |
| 192 | 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, -2, -1, -1, |
| 193 | -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, |
| 194 | 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, -1, -1, -1, |
| 195 | -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, |
| 196 | 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, -1, -1, -1, |
| 197 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 198 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 199 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 200 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 201 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 202 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 203 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 204 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 205 | }; |
| 206 | |
| 207 | /** |
| 208 | * Decode lookup table for the "web safe" variant (RFC 3548 |
| 209 | * sec. 4) where - and _ replace + and /. |
| 210 | */ |
| 211 | private static final int DECODE_WEBSAFE[] = { |
| 212 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 213 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 214 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, -1, |
| 215 | 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, -2, -1, -1, |
| 216 | -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, |
| 217 | 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, -1, -1, 63, |
| 218 | -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, |
| 219 | 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, -1, -1, -1, |
| 220 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 221 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 222 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 223 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 224 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 225 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 226 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 227 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, |
| 228 | }; |
| 229 | |
| 230 | /** Non-data values in the DECODE arrays. */ |
| 231 | private static final int SKIP = -1; |
| 232 | private static final int EQUALS = -2; |
| 233 | |
| 234 | /** |
| 235 | * States 0-3 are reading through the next input tuple. |
| 236 | * State 4 is having read one '=' and expecting exactly |
| 237 | * one more. |
| 238 | * State 5 is expecting no more data or padding characters |
| 239 | * in the input. |
| 240 | * State 6 is the error state; an error has been detected |
| 241 | * in the input and no future input can "fix" it. |
| 242 | */ |
| 243 | private int state; // state number (0 to 6) |
| 244 | private int value; |
| 245 | |
| 246 | final private int[] alphabet; |
| 247 | |
| 248 | public Decoder(int flags, byte[] output) { |
| 249 | this.output = output; |
| 250 | |
| 251 | alphabet = ((flags & URL_SAFE) == 0) ? DECODE : DECODE_WEBSAFE; |
| 252 | state = 0; |
| 253 | value = 0; |
| 254 | } |
| 255 | |
| 256 | /** |
| 257 | * @return an overestimate for the number of bytes {@code |
| 258 | * len} bytes could decode to. |
| 259 | */ |
| 260 | @Override |
| 261 | public int maxOutputSize(int len) { |
| 262 | return len * 3/4 + 10; |
| 263 | } |
| 264 | |
| 265 | /** |
| 266 | * Decode another block of input data. |
| 267 | * |
| 268 | * @return true if the state machine is still healthy. false if |
| 269 | * bad base-64 data has been detected in the input stream. |
| 270 | */ |
| 271 | @Override |
| 272 | public boolean process(byte[] input, int offset, int len, boolean finish) { |
| 273 | if (this.state == 6) return false; |
| 274 | |
| 275 | int p = offset; |
| 276 | len += offset; |
| 277 | |
| 278 | // Using local variables makes the decoder about 12% |
| 279 | // faster than if we manipulate the member variables in |
| 280 | // the loop. (Even alphabet makes a measurable |
| 281 | // difference, which is somewhat surprising to me since |
| 282 | // the member variable is final.) |
| 283 | int state = this.state; |
| 284 | int value = this.value; |
| 285 | int op = 0; |
| 286 | final byte[] output = this.output; |
| 287 | final int[] alphabet = this.alphabet; |
| 288 | |
| 289 | while (p < len) { |
| 290 | // Try the fast path: we're starting a new tuple and the |
| 291 | // next four bytes of the input stream are all data |
| 292 | // bytes. This corresponds to going through states |
| 293 | // 0-1-2-3-0. We expect to use this method for most of |
| 294 | // the data. |
| 295 | // |
| 296 | // If any of the next four bytes of input are non-data |
| 297 | // (whitespace, etc.), value will end up negative. (All |
| 298 | // the non-data values in decode are small negative |
| 299 | // numbers, so shifting any of them up and or'ing them |
| 300 | // together will result in a value with its top bit set.) |
| 301 | // |
| 302 | // You can remove this whole block and the output should |
| 303 | // be the same, just slower. |
| 304 | if (state == 0) { |
| 305 | while (p+4 <= len && |
| 306 | (value = ((alphabet[input[p] & 0xff] << 18) | |
| 307 | (alphabet[input[p+1] & 0xff] << 12) | |
| 308 | (alphabet[input[p+2] & 0xff] << 6) | |
| 309 | (alphabet[input[p+3] & 0xff]))) >= 0) { |
| 310 | output[op+2] = (byte) value; |
| 311 | output[op+1] = (byte) (value >> 8); |
| 312 | output[op] = (byte) (value >> 16); |
| 313 | op += 3; |
| 314 | p += 4; |
| 315 | } |
| 316 | if (p >= len) break; |
| 317 | } |
| 318 | |
| 319 | // The fast path isn't available -- either we've read a |
| 320 | // partial tuple, or the next four input bytes aren't all |
| 321 | // data, or whatever. Fall back to the slower state |
| 322 | // machine implementation. |
| 323 | |
| 324 | int d = alphabet[input[p++] & 0xff]; |
| 325 | |
| 326 | switch (state) { |
| 327 | case 0: |
| 328 | if (d >= 0) { |
| 329 | value = d; |
| 330 | ++state; |
| 331 | } else if (d != SKIP) { |
| 332 | this.state = 6; |
| 333 | return false; |
| 334 | } |
| 335 | break; |
| 336 | |
| 337 | case 1: |
| 338 | if (d >= 0) { |
| 339 | value = (value << 6) | d; |
| 340 | ++state; |
| 341 | } else if (d != SKIP) { |
| 342 | this.state = 6; |
| 343 | return false; |
| 344 | } |
| 345 | break; |
| 346 | |
| 347 | case 2: |
| 348 | if (d >= 0) { |
| 349 | value = (value << 6) | d; |
| 350 | ++state; |
| 351 | } else if (d == EQUALS) { |
| 352 | // Emit the last (partial) output tuple; |
| 353 | // expect exactly one more padding character. |
| 354 | output[op++] = (byte) (value >> 4); |
| 355 | state = 4; |
| 356 | } else if (d != SKIP) { |
| 357 | this.state = 6; |
| 358 | return false; |
| 359 | } |
| 360 | break; |
| 361 | |
| 362 | case 3: |
| 363 | if (d >= 0) { |
| 364 | // Emit the output triple and return to state 0. |
| 365 | value = (value << 6) | d; |
| 366 | output[op+2] = (byte) value; |
| 367 | output[op+1] = (byte) (value >> 8); |
| 368 | output[op] = (byte) (value >> 16); |
| 369 | op += 3; |
| 370 | state = 0; |
| 371 | } else if (d == EQUALS) { |
| 372 | // Emit the last (partial) output tuple; |
| 373 | // expect no further data or padding characters. |
| 374 | output[op+1] = (byte) (value >> 2); |
| 375 | output[op] = (byte) (value >> 10); |
| 376 | op += 2; |
| 377 | state = 5; |
| 378 | } else if (d != SKIP) { |
| 379 | this.state = 6; |
| 380 | return false; |
| 381 | } |
| 382 | break; |
| 383 | |
| 384 | case 4: |
| 385 | if (d == EQUALS) { |
| 386 | ++state; |
| 387 | } else if (d != SKIP) { |
| 388 | this.state = 6; |
| 389 | return false; |
| 390 | } |
| 391 | break; |
| 392 | |
| 393 | case 5: |
| 394 | if (d != SKIP) { |
| 395 | this.state = 6; |
| 396 | return false; |
| 397 | } |
| 398 | break; |
| 399 | } |
| 400 | } |
| 401 | |
| 402 | if (!finish) { |
| 403 | // We're out of input, but a future call could provide |
| 404 | // more. |
| 405 | this.state = state; |
| 406 | this.value = value; |
| 407 | this.op = op; |
| 408 | return true; |
| 409 | } |
| 410 | |
| 411 | // Done reading input. Now figure out where we are left in |
| 412 | // the state machine and finish up. |
| 413 | |
| 414 | switch (state) { |
| 415 | case 0: |
| 416 | // Output length is a multiple of three. Fine. |
| 417 | break; |
| 418 | case 1: |
| 419 | // Read one extra input byte, which isn't enough to |
| 420 | // make another output byte. Illegal. |
| 421 | this.state = 6; |
| 422 | return false; |
| 423 | case 2: |
| 424 | // Read two extra input bytes, enough to emit 1 more |
| 425 | // output byte. Fine. |
| 426 | output[op++] = (byte) (value >> 4); |
| 427 | break; |
| 428 | case 3: |
| 429 | // Read three extra input bytes, enough to emit 2 more |
| 430 | // output bytes. Fine. |
| 431 | output[op++] = (byte) (value >> 10); |
| 432 | output[op++] = (byte) (value >> 2); |
| 433 | break; |
| 434 | case 4: |
| 435 | // Read one padding '=' when we expected 2. Illegal. |
| 436 | this.state = 6; |
| 437 | return false; |
| 438 | case 5: |
| 439 | // Read all the padding '='s we expected and no more. |
| 440 | // Fine. |
| 441 | break; |
| 442 | } |
| 443 | |
| 444 | this.state = state; |
| 445 | this.op = op; |
| 446 | return true; |
| 447 | } |
| 448 | } |
| 449 | |
| 450 | // -------------------------------------------------------- |
| 451 | // encoding |
| 452 | // -------------------------------------------------------- |
| 453 | |
| 454 | /** |
| 455 | * Base64-encode the given data and return a newly allocated |
| 456 | * String with the result. |
| 457 | * |
| 458 | * @param input the data to encode |
| 459 | * @param flags controls certain features of the encoded output. |
| 460 | * Passing {@code DEFAULT} results in output that |
| 461 | * adheres to RFC 2045. |
| 462 | */ |
| 463 | public static String encodeToString(byte[] input, int flags) { |
| 464 | try { |
| 465 | return new String(encode(input, flags), "US-ASCII"); |
| 466 | } catch (UnsupportedEncodingException e) { |
| 467 | // US-ASCII is guaranteed to be available. |
| 468 | throw new AssertionError(e); |
| 469 | } |
| 470 | } |
| 471 | |
| 472 | /** |
| 473 | * Base64-encode the given data and return a newly allocated |
| 474 | * String with the result. |
| 475 | * |
| 476 | * @param input the data to encode |
| 477 | * @param offset the position within the input array at which to |
| 478 | * start |
| 479 | * @param len the number of bytes of input to encode |
| 480 | * @param flags controls certain features of the encoded output. |
| 481 | * Passing {@code DEFAULT} results in output that |
| 482 | * adheres to RFC 2045. |
| 483 | */ |
| 484 | public static String encodeToString(byte[] input, int offset, int len, int flags) { |
| 485 | try { |
| 486 | return new String(encode(input, offset, len, flags), "US-ASCII"); |
| 487 | } catch (UnsupportedEncodingException e) { |
| 488 | // US-ASCII is guaranteed to be available. |
| 489 | throw new AssertionError(e); |
| 490 | } |
| 491 | } |
| 492 | |
| 493 | /** |
| 494 | * Base64-encode the given data and return a newly allocated |
| 495 | * byte[] with the result. |
| 496 | * |
| 497 | * @param input the data to encode |
| 498 | * @param flags controls certain features of the encoded output. |
| 499 | * Passing {@code DEFAULT} results in output that |
| 500 | * adheres to RFC 2045. |
| 501 | */ |
| 502 | public static byte[] encode(byte[] input, int flags) { |
| 503 | return encode(input, 0, input.length, flags); |
| 504 | } |
| 505 | |
| 506 | /** |
| 507 | * Base64-encode the given data and return a newly allocated |
| 508 | * byte[] with the result. |
| 509 | * |
| 510 | * @param input the data to encode |
| 511 | * @param offset the position within the input array at which to |
| 512 | * start |
| 513 | * @param len the number of bytes of input to encode |
| 514 | * @param flags controls certain features of the encoded output. |
| 515 | * Passing {@code DEFAULT} results in output that |
| 516 | * adheres to RFC 2045. |
| 517 | */ |
| 518 | public static byte[] encode(byte[] input, int offset, int len, int flags) { |
| 519 | Encoder encoder = new Encoder(flags, null); |
| 520 | |
| 521 | // Compute the exact length of the array we will produce. |
| 522 | int output_len = len / 3 * 4; |
| 523 | |
| 524 | // Account for the tail of the data and the padding bytes, if any. |
| 525 | if (encoder.do_padding) { |
| 526 | if (len % 3 > 0) { |
| 527 | output_len += 4; |
| 528 | } |
| 529 | } else { |
| 530 | switch (len % 3) { |
| 531 | case 0: break; |
| 532 | case 1: output_len += 2; break; |
| 533 | case 2: output_len += 3; break; |
| 534 | } |
| 535 | } |
| 536 | |
| 537 | // Account for the newlines, if any. |
| 538 | if (encoder.do_newline && len > 0) { |
| 539 | output_len += (((len-1) / (3 * Encoder.LINE_GROUPS)) + 1) * |
| 540 | (encoder.do_cr ? 2 : 1); |
| 541 | } |
| 542 | |
| 543 | encoder.output = new byte[output_len]; |
| 544 | encoder.process(input, offset, len, true); |
| 545 | |
| 546 | assert encoder.op == output_len; |
| 547 | |
| 548 | return encoder.output; |
| 549 | } |
| 550 | |
| 551 | /* package */ static class Encoder extends Coder { |
| 552 | /** |
| 553 | * Emit a new line every this many output tuples. Corresponds to |
| 554 | * a 76-character line length (the maximum allowable according to |
| 555 | * <a href="http://www.ietf.org/rfc/rfc2045.txt">RFC 2045</a>). |
| 556 | */ |
| 557 | public static final int LINE_GROUPS = 19; |
| 558 | |
| 559 | /** |
| 560 | * Lookup table for turning Base64 alphabet positions (6 bits) |
| 561 | * into output bytes. |
| 562 | */ |
| 563 | private static final byte ENCODE[] = { |
| 564 | 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', |
| 565 | 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', |
| 566 | 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', |
| 567 | 'w', 'x', 'y', 'z', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '+', '/', |
| 568 | }; |
| 569 | |
| 570 | /** |
| 571 | * Lookup table for turning Base64 alphabet positions (6 bits) |
| 572 | * into output bytes. |
| 573 | */ |
| 574 | private static final byte ENCODE_WEBSAFE[] = { |
| 575 | 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', |
| 576 | 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', |
| 577 | 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', |
| 578 | 'w', 'x', 'y', 'z', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '-', '_', |
| 579 | }; |
| 580 | |
| 581 | final private byte[] tail; |
| 582 | /* package */ int tailLen; |
| 583 | private int count; |
| 584 | |
| 585 | final public boolean do_padding; |
| 586 | final public boolean do_newline; |
| 587 | final public boolean do_cr; |
| 588 | final private byte[] alphabet; |
| 589 | |
| 590 | public Encoder(int flags, byte[] output) { |
| 591 | this.output = output; |
| 592 | |
| 593 | do_padding = (flags & NO_PADDING) == 0; |
| 594 | do_newline = (flags & NO_WRAP) == 0; |
| 595 | do_cr = (flags & CRLF) != 0; |
| 596 | alphabet = ((flags & URL_SAFE) == 0) ? ENCODE : ENCODE_WEBSAFE; |
| 597 | |
| 598 | tail = new byte[2]; |
| 599 | tailLen = 0; |
| 600 | |
| 601 | count = do_newline ? LINE_GROUPS : -1; |
| 602 | } |
| 603 | |
| 604 | /** |
| 605 | * @return an overestimate for the number of bytes {@code |
| 606 | * len} bytes could encode to. |
| 607 | */ |
| 608 | @Override |
| 609 | public int maxOutputSize(int len) { |
| 610 | return len * 8/5 + 10; |
| 611 | } |
| 612 | |
| 613 | @Override |
| 614 | public boolean process(byte[] input, int offset, int len, boolean finish) { |
| 615 | // Using local variables makes the encoder about 9% faster. |
| 616 | final byte[] alphabet = this.alphabet; |
| 617 | final byte[] output = this.output; |
| 618 | int op = 0; |
| 619 | int count = this.count; |
| 620 | |
| 621 | int p = offset; |
| 622 | len += offset; |
| 623 | int v = -1; |
| 624 | |
| 625 | // First we need to concatenate the tail of the previous call |
| 626 | // with any input bytes available now and see if we can empty |
| 627 | // the tail. |
| 628 | |
| 629 | switch (tailLen) { |
| 630 | case 0: |
| 631 | // There was no tail. |
| 632 | break; |
| 633 | |
| 634 | case 1: |
| 635 | if (p+2 <= len) { |
| 636 | // A 1-byte tail with at least 2 bytes of |
| 637 | // input available now. |
| 638 | v = ((tail[0] & 0xff) << 16) | |
| 639 | ((input[p++] & 0xff) << 8) | |
| 640 | (input[p++] & 0xff); |
| 641 | tailLen = 0; |
| 642 | } |
| 643 | break; |
| 644 | |
| 645 | case 2: |
| 646 | if (p+1 <= len) { |
| 647 | // A 2-byte tail with at least 1 byte of input. |
| 648 | v = ((tail[0] & 0xff) << 16) | |
| 649 | ((tail[1] & 0xff) << 8) | |
| 650 | (input[p++] & 0xff); |
| 651 | tailLen = 0; |
| 652 | } |
| 653 | break; |
| 654 | } |
| 655 | |
| 656 | if (v != -1) { |
| 657 | output[op++] = alphabet[(v >> 18) & 0x3f]; |
| 658 | output[op++] = alphabet[(v >> 12) & 0x3f]; |
| 659 | output[op++] = alphabet[(v >> 6) & 0x3f]; |
| 660 | output[op++] = alphabet[v & 0x3f]; |
| 661 | if (--count == 0) { |
| 662 | if (do_cr) output[op++] = '\r'; |
| 663 | output[op++] = '\n'; |
| 664 | count = LINE_GROUPS; |
| 665 | } |
| 666 | } |
| 667 | |
| 668 | // At this point either there is no tail, or there are fewer |
| 669 | // than 3 bytes of input available. |
| 670 | |
| 671 | // The main loop, turning 3 input bytes into 4 output bytes on |
| 672 | // each iteration. |
| 673 | while (p+3 <= len) { |
| 674 | v = ((input[p] & 0xff) << 16) | |
| 675 | ((input[p+1] & 0xff) << 8) | |
| 676 | (input[p+2] & 0xff); |
| 677 | output[op] = alphabet[(v >> 18) & 0x3f]; |
| 678 | output[op+1] = alphabet[(v >> 12) & 0x3f]; |
| 679 | output[op+2] = alphabet[(v >> 6) & 0x3f]; |
| 680 | output[op+3] = alphabet[v & 0x3f]; |
| 681 | p += 3; |
| 682 | op += 4; |
| 683 | if (--count == 0) { |
| 684 | if (do_cr) output[op++] = '\r'; |
| 685 | output[op++] = '\n'; |
| 686 | count = LINE_GROUPS; |
| 687 | } |
| 688 | } |
| 689 | |
| 690 | if (finish) { |
| 691 | // Finish up the tail of the input. Note that we need to |
| 692 | // consume any bytes in tail before any bytes |
| 693 | // remaining in input; there should be at most two bytes |
| 694 | // total. |
| 695 | |
| 696 | if (p-tailLen == len-1) { |
| 697 | int t = 0; |
| 698 | v = ((tailLen > 0 ? tail[t++] : input[p++]) & 0xff) << 4; |
| 699 | tailLen -= t; |
| 700 | output[op++] = alphabet[(v >> 6) & 0x3f]; |
| 701 | output[op++] = alphabet[v & 0x3f]; |
| 702 | if (do_padding) { |
| 703 | output[op++] = '='; |
| 704 | output[op++] = '='; |
| 705 | } |
| 706 | if (do_newline) { |
| 707 | if (do_cr) output[op++] = '\r'; |
| 708 | output[op++] = '\n'; |
| 709 | } |
| 710 | } else if (p-tailLen == len-2) { |
| 711 | int t = 0; |
| 712 | v = (((tailLen > 1 ? tail[t++] : input[p++]) & 0xff) << 10) | |
| 713 | (((tailLen > 0 ? tail[t++] : input[p++]) & 0xff) << 2); |
| 714 | tailLen -= t; |
| 715 | output[op++] = alphabet[(v >> 12) & 0x3f]; |
| 716 | output[op++] = alphabet[(v >> 6) & 0x3f]; |
| 717 | output[op++] = alphabet[v & 0x3f]; |
| 718 | if (do_padding) { |
| 719 | output[op++] = '='; |
| 720 | } |
| 721 | if (do_newline) { |
| 722 | if (do_cr) output[op++] = '\r'; |
| 723 | output[op++] = '\n'; |
| 724 | } |
| 725 | } else if (do_newline && op > 0 && count != LINE_GROUPS) { |
| 726 | if (do_cr) output[op++] = '\r'; |
| 727 | output[op++] = '\n'; |
| 728 | } |
| 729 | |
| 730 | assert tailLen == 0; |
| 731 | assert p == len; |
| 732 | } else { |
| 733 | // Save the leftovers in tail to be consumed on the next |
| 734 | // call to encodeInternal. |
| 735 | |
| 736 | if (p == len-1) { |
| 737 | tail[tailLen++] = input[p]; |
| 738 | } else if (p == len-2) { |
| 739 | tail[tailLen++] = input[p]; |
| 740 | tail[tailLen++] = input[p+1]; |
| 741 | } |
| 742 | } |
| 743 | |
| 744 | this.op = op; |
| 745 | this.count = count; |
| 746 | |
| 747 | return true; |
| 748 | } |
| 749 | } |
| 750 | |
| 751 | private Base64() { } // don't instantiate |
| 752 | } |