PageRenderTime 48ms CodeModel.GetById 18ms RepoModel.GetById 1ms app.codeStats 0ms

/module/geb-core/src/main/java/geb/report/Base64.java

http://github.com/geb/geb
Java | 579 lines | 321 code | 82 blank | 176 comment | 119 complexity | 9ad1b0ccc55f354575fe35db92dde0c8 MD5 | raw file
Possible License(s): Apache-2.0
  1. package geb.report;
  2. import java.util.Arrays;
  3. /**
  4. * A very fast and memory efficient class to encode and decode to and from BASE64 in full accordance with RFC 2045.<br><br> On Windows XP sp1 with 1.4.2_04 and later ;), this encoder and decoder is
  5. * about 10 times faster on small arrays (10 - 1000 bytes) and 2-3 times as fast on larger arrays (10000 - 1000000 bytes) compared to <code>sun.misc.Encoder()/Decoder()</code>.<br><br>
  6. *
  7. * On byte arrays the encoder is about 20% faster than Jakarta Commons Base64 Codec for encode and about 50% faster for decoding large arrays. This implementation is about twice as fast on very small
  8. * arrays (&lt 30 bytes). If source/destination is a <code>String</code> this version is about three times as fast due to the fact that the Commons Codec result has to be recoded to a
  9. * <code>String</code> from <code>byte[]</code>, which is very expensive.<br><br>
  10. *
  11. * This encode/decode algorithm doesn't create any temporary arrays as many other codecs do, it only allocates the resulting array. This produces less garbage and it is possible to handle arrays twice
  12. * as large as algorithms that create a temporary array. (E.g. Jakarta Commons Codec). It is unknown whether Sun's <code>sun.misc.Encoder()/Decoder()</code> produce temporary arrays but since
  13. * performance is quite low it probably does.<br><br>
  14. *
  15. * The encoder produces the same output as the Sun one except that the Sun's encoder appends a trailing line separator if the last character isn't a pad. Unclear why but it only adds to the length and
  16. * is probably a side effect. Both are in conformance with RFC 2045 though.<br> Commons codec seem to always att a trailing line separator.<br><br>
  17. *
  18. * <b>Note!</b> The encode/decode method pairs (types) come in three versions with the <b>exact</b> same algorithm and thus a lot of code redundancy. This is to not create any temporary arrays for
  19. * transcoding to/from different format types. The methods not used can simply be commented out.<br><br>
  20. *
  21. * There is also a "fast" version of all decode methods that works the same way as the normal ones, but har a few demands on the decoded input. Normally though, these fast verions should be used if
  22. * the source if the input is known and it hasn't bee tampered with.<br><br>
  23. *
  24. * If you find the code useful or you find a bug, please send me a note at base64 @ miginfocom . com.
  25. *
  26. * Licence (BSD): ==============
  27. *
  28. * Copyright (c) 2004, Mikael Grev, MiG InfoCom AB. (base64 @ miginfocom . com) All rights reserved.
  29. *
  30. * Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: Redistributions of source code must retain the above
  31. * copyright notice, this list of conditions and the following disclaimer. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following
  32. * disclaimer in the documentation and/or other materials provided with the distribution. Neither the name of the MiG InfoCom AB nor the names of its contributors may be used to endorse or promote
  33. * products derived from this software without specific prior written permission.
  34. *
  35. * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  36. * FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
  37. * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
  38. * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  39. *
  40. * @author Mikael Grev Date: 2004-aug-02 Time: 11:31:11
  41. * @version 2.2
  42. */
  43. public class Base64 {
  44. private static final char[] CA = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/".toCharArray();
  45. private static final int[] IA = new int[256];
  46. static {
  47. Arrays.fill(IA, -1);
  48. for (int i = 0, iS = CA.length; i < iS; i++) {
  49. IA[CA[i]] = i;
  50. }
  51. IA['='] = 0;
  52. }
  53. // ****************************************************************************************
  54. // * char[] version
  55. // ****************************************************************************************
  56. /**
  57. * Encodes a raw byte array into a BASE64 <code>char[]</code> representation i accordance with RFC 2045.
  58. *
  59. * @param sArr The bytes to convert. If <code>null</code> or length 0 an empty array will be returned.
  60. * @param lineSep Optional "\r\n" after 76 characters, unless end of file.<br> No line separator will be in breach of RFC 2045 which specifies max 76 per line but will be a little faster.
  61. * @return A BASE64 encoded array. Never <code>null</code>.
  62. */
  63. public final static char[] encodeToChar(byte[] sArr, boolean lineSep) {
  64. // Check special case
  65. int sLen = sArr != null ? sArr.length : 0;
  66. if (sLen == 0) {
  67. return new char[0];
  68. }
  69. int eLen = (sLen / 3) * 3; // Length of even 24-bits.
  70. int cCnt = ((sLen - 1) / 3 + 1) << 2; // Returned character count
  71. int dLen = cCnt + (lineSep ? (cCnt - 1) / 76 << 1 : 0); // Length of returned array
  72. char[] dArr = new char[dLen];
  73. // Encode even 24-bits
  74. for (int s = 0, d = 0, cc = 0; s < eLen; ) {
  75. // Copy next three bytes into lower 24 bits of int, paying attension to sign.
  76. int i = (sArr[s++] & 0xff) << 16 | (sArr[s++] & 0xff) << 8 | (sArr[s++] & 0xff);
  77. // Encode the int into four chars
  78. dArr[d++] = CA[(i >>> 18) & 0x3f];
  79. dArr[d++] = CA[(i >>> 12) & 0x3f];
  80. dArr[d++] = CA[(i >>> 6) & 0x3f];
  81. dArr[d++] = CA[i & 0x3f];
  82. // Add optional line separator
  83. if (lineSep && ++cc == 19 && d < dLen - 2) {
  84. dArr[d++] = '\r';
  85. dArr[d++] = '\n';
  86. cc = 0;
  87. }
  88. }
  89. // Pad and encode last bits if source isn't even 24 bits.
  90. int left = sLen - eLen; // 0 - 2.
  91. if (left > 0) {
  92. // Prepare the int
  93. int i = ((sArr[eLen] & 0xff) << 10) | (left == 2 ? ((sArr[sLen - 1] & 0xff) << 2) : 0);
  94. // Set last four chars
  95. dArr[dLen - 4] = CA[i >> 12];
  96. dArr[dLen - 3] = CA[(i >>> 6) & 0x3f];
  97. dArr[dLen - 2] = left == 2 ? CA[i & 0x3f] : '=';
  98. dArr[dLen - 1] = '=';
  99. }
  100. return dArr;
  101. }
  102. /**
  103. * Decodes a BASE64 encoded char array. All illegal characters will be ignored and can handle both arrays with and without line separators.
  104. *
  105. * @param sArr The source array. <code>null</code> or length 0 will return an empty array.
  106. * @return The decoded array of bytes. May be of length 0. Will be <code>null</code> if the legal characters (including '=') isn't divideable by 4. (I.e. definitely corrupted).
  107. */
  108. public final static byte[] decode(char[] sArr) {
  109. // Check special case
  110. int sLen = sArr != null ? sArr.length : 0;
  111. if (sLen == 0) {
  112. return new byte[0];
  113. }
  114. // Count illegal characters (including '\r', '\n') to know what size the returned array will be,
  115. // so we don't have to reallocate & copy it later.
  116. int sepCnt = 0; // Number of separator characters. (Actually illegal characters, but that's a bonus[])
  117. for (int i = 0; i < sLen; i++) // If input is "pure" (I.e. no line separators or illegal chars) base64 this loop can be commented out.
  118. {
  119. if (IA[sArr[i]] < 0) {
  120. sepCnt++;
  121. }
  122. }
  123. // Check so that legal chars (including '=') are evenly divideable by 4 as specified in RFC 2045.
  124. if ((sLen - sepCnt) % 4 != 0) {
  125. return null;
  126. }
  127. int pad = 0;
  128. for (int i = sLen; i > 1 && IA[sArr[--i]] <= 0; ) {
  129. if (sArr[i] == '=') {
  130. pad++;
  131. }
  132. }
  133. int len = ((sLen - sepCnt) * 6 >> 3) - pad;
  134. byte[] dArr = new byte[len]; // Preallocate byte[] of exact length
  135. for (int s = 0, d = 0; d < len; ) {
  136. // Assemble three bytes into an int from four "valid" characters.
  137. int i = 0;
  138. for (int j = 0; j < 4; j++) { // j only increased if a valid char was found.
  139. int c = IA[sArr[s++]];
  140. if (c >= 0) {
  141. i |= c << (18 - j * 6);
  142. } else {
  143. j--;
  144. }
  145. }
  146. // Add the bytes
  147. dArr[d++] = (byte) (i >> 16);
  148. if (d < len) {
  149. dArr[d++] = (byte) (i >> 8);
  150. if (d < len) {
  151. dArr[d++] = (byte) i;
  152. }
  153. }
  154. }
  155. return dArr;
  156. }
  157. /**
  158. * Decodes a BASE64 encoded char array that is known to be resonably well formatted. The method is about twice as fast as {@link #decode(char[])}. The preconditions are:<br> + The array must have
  159. * a line length of 76 chars OR no line separators at all (one line).<br> + Line separator must be "\r\n", as specified in RFC 2045 + The array must not contain illegal characters within the
  160. * encoded string<br> + The array CAN have illegal characters at the beginning and end, those will be dealt with appropriately.<br>
  161. *
  162. * @param sArr The source array. Length 0 will return an empty array. <code>null</code> will throw an exception.
  163. * @return The decoded array of bytes. May be of length 0.
  164. */
  165. public final static byte[] decodeFast(char[] sArr) {
  166. // Check special case
  167. int sLen = sArr.length;
  168. if (sLen == 0) {
  169. return new byte[0];
  170. }
  171. int sIx = 0, eIx = sLen - 1; // Start and end index after trimming.
  172. // Trim illegal chars from start
  173. while (sIx < eIx && IA[sArr[sIx]] < 0) {
  174. sIx++;
  175. }
  176. // Trim illegal chars from end
  177. while (eIx > 0 && IA[sArr[eIx]] < 0) {
  178. eIx--;
  179. }
  180. // get the padding count (=) (0, 1 or 2)
  181. int pad = sArr[eIx] == '=' ? (sArr[eIx - 1] == '=' ? 2 : 1) : 0; // Count '=' at end.
  182. int cCnt = eIx - sIx + 1; // Content count including possible separators
  183. int sepCnt = sLen > 76 ? (sArr[76] == '\r' ? cCnt / 78 : 0) << 1 : 0;
  184. int len = ((cCnt - sepCnt) * 6 >> 3) - pad; // The number of decoded bytes
  185. byte[] dArr = new byte[len]; // Preallocate byte[] of exact length
  186. // Decode all but the last 0 - 2 bytes.
  187. int d = 0;
  188. for (int cc = 0, eLen = (len / 3) * 3; d < eLen; ) {
  189. // Assemble three bytes into an int from four "valid" characters.
  190. int i = IA[sArr[sIx++]] << 18 | IA[sArr[sIx++]] << 12 | IA[sArr[sIx++]] << 6 | IA[sArr[sIx++]];
  191. // Add the bytes
  192. dArr[d++] = (byte) (i >> 16);
  193. dArr[d++] = (byte) (i >> 8);
  194. dArr[d++] = (byte) i;
  195. // If line separator, jump over it.
  196. if (sepCnt > 0 && ++cc == 19) {
  197. sIx += 2;
  198. cc = 0;
  199. }
  200. }
  201. if (d < len) {
  202. // Decode last 1-3 bytes (incl '=') into 1-3 bytes
  203. int i = 0;
  204. for (int j = 0; sIx <= eIx - pad; j++) {
  205. i |= IA[sArr[sIx++]] << (18 - j * 6);
  206. }
  207. for (int r = 16; d < len; r -= 8) {
  208. dArr[d++] = (byte) (i >> r);
  209. }
  210. }
  211. return dArr;
  212. }
  213. // ****************************************************************************************
  214. // * byte[] version
  215. // ****************************************************************************************
  216. /**
  217. * Encodes a raw byte array into a BASE64 <code>byte[]</code> representation i accordance with RFC 2045.
  218. *
  219. * @param sArr The bytes to convert. If <code>null</code> or length 0 an empty array will be returned.
  220. * @param lineSep Optional "\r\n" after 76 characters, unless end of file.<br> No line separator will be in breach of RFC 2045 which specifies max 76 per line but will be a little faster.
  221. * @return A BASE64 encoded array. Never <code>null</code>.
  222. */
  223. public final static byte[] encodeToByte(byte[] sArr, boolean lineSep) {
  224. // Check special case
  225. int sLen = sArr != null ? sArr.length : 0;
  226. if (sLen == 0) {
  227. return new byte[0];
  228. }
  229. int eLen = (sLen / 3) * 3; // Length of even 24-bits.
  230. int cCnt = ((sLen - 1) / 3 + 1) << 2; // Returned character count
  231. int dLen = cCnt + (lineSep ? (cCnt - 1) / 76 << 1 : 0); // Length of returned array
  232. byte[] dArr = new byte[dLen];
  233. // Encode even 24-bits
  234. for (int s = 0, d = 0, cc = 0; s < eLen; ) {
  235. // Copy next three bytes into lower 24 bits of int, paying attension to sign.
  236. int i = (sArr[s++] & 0xff) << 16 | (sArr[s++] & 0xff) << 8 | (sArr[s++] & 0xff);
  237. // Encode the int into four chars
  238. dArr[d++] = (byte) CA[(i >>> 18) & 0x3f];
  239. dArr[d++] = (byte) CA[(i >>> 12) & 0x3f];
  240. dArr[d++] = (byte) CA[(i >>> 6) & 0x3f];
  241. dArr[d++] = (byte) CA[i & 0x3f];
  242. // Add optional line separator
  243. if (lineSep && ++cc == 19 && d < dLen - 2) {
  244. dArr[d++] = '\r';
  245. dArr[d++] = '\n';
  246. cc = 0;
  247. }
  248. }
  249. // Pad and encode last bits if source isn't an even 24 bits.
  250. int left = sLen - eLen; // 0 - 2.
  251. if (left > 0) {
  252. // Prepare the int
  253. int i = ((sArr[eLen] & 0xff) << 10) | (left == 2 ? ((sArr[sLen - 1] & 0xff) << 2) : 0);
  254. // Set last four chars
  255. dArr[dLen - 4] = (byte) CA[i >> 12];
  256. dArr[dLen - 3] = (byte) CA[(i >>> 6) & 0x3f];
  257. dArr[dLen - 2] = left == 2 ? (byte) CA[i & 0x3f] : (byte) '=';
  258. dArr[dLen - 1] = '=';
  259. }
  260. return dArr;
  261. }
  262. /**
  263. * Decodes a BASE64 encoded byte array. All illegal characters will be ignored and can handle both arrays with and without line separators.
  264. *
  265. * @param sArr The source array. Length 0 will return an empty array. <code>null</code> will throw an exception.
  266. * @return The decoded array of bytes. May be of length 0. Will be <code>null</code> if the legal characters (including '=') isn't divideable by 4. (I.e. definitely corrupted).
  267. */
  268. public final static byte[] decode(byte[] sArr) {
  269. // Check special case
  270. int sLen = sArr.length;
  271. // Count illegal characters (including '\r', '\n') to know what size the returned array will be,
  272. // so we don't have to reallocate & copy it later.
  273. int sepCnt = 0; // Number of separator characters. (Actually illegal characters, but that's a bonus[])
  274. for (int i = 0; i < sLen; i++) // If input is "pure" (I.e. no line separators or illegal chars) base64 this loop can be commented out.
  275. {
  276. if (IA[sArr[i] & 0xff] < 0) {
  277. sepCnt++;
  278. }
  279. }
  280. // Check so that legal chars (including '=') are evenly divideable by 4 as specified in RFC 2045.
  281. if ((sLen - sepCnt) % 4 != 0) {
  282. return null;
  283. }
  284. int pad = 0;
  285. for (int i = sLen; i > 1 && IA[sArr[--i] & 0xff] <= 0; ) {
  286. if (sArr[i] == '=') {
  287. pad++;
  288. }
  289. }
  290. int len = ((sLen - sepCnt) * 6 >> 3) - pad;
  291. byte[] dArr = new byte[len]; // Preallocate byte[] of exact length
  292. for (int s = 0, d = 0; d < len; ) {
  293. // Assemble three bytes into an int from four "valid" characters.
  294. int i = 0;
  295. for (int j = 0; j < 4; j++) { // j only increased if a valid char was found.
  296. int c = IA[sArr[s++] & 0xff];
  297. if (c >= 0) {
  298. i |= c << (18 - j * 6);
  299. } else {
  300. j--;
  301. }
  302. }
  303. // Add the bytes
  304. dArr[d++] = (byte) (i >> 16);
  305. if (d < len) {
  306. dArr[d++] = (byte) (i >> 8);
  307. if (d < len) {
  308. dArr[d++] = (byte) i;
  309. }
  310. }
  311. }
  312. return dArr;
  313. }
  314. /**
  315. * Decodes a BASE64 encoded byte array that is known to be resonably well formatted. The method is about twice as fast as {@link #decode(byte[])}. The preconditions are:<br> + The array must have
  316. * a line length of 76 chars OR no line separators at all (one line).<br> + Line separator must be "\r\n", as specified in RFC 2045 + The array must not contain illegal characters within the
  317. * encoded string<br> + The array CAN have illegal characters at the beginning and end, those will be dealt with appropriately.<br>
  318. *
  319. * @param sArr The source array. Length 0 will return an empty array. <code>null</code> will throw an exception.
  320. * @return The decoded array of bytes. May be of length 0.
  321. */
  322. public final static byte[] decodeFast(byte[] sArr) {
  323. // Check special case
  324. int sLen = sArr.length;
  325. if (sLen == 0) {
  326. return new byte[0];
  327. }
  328. int sIx = 0, eIx = sLen - 1; // Start and end index after trimming.
  329. // Trim illegal chars from start
  330. while (sIx < eIx && IA[sArr[sIx] & 0xff] < 0) {
  331. sIx++;
  332. }
  333. // Trim illegal chars from end
  334. while (eIx > 0 && IA[sArr[eIx] & 0xff] < 0) {
  335. eIx--;
  336. }
  337. // get the padding count (=) (0, 1 or 2)
  338. int pad = sArr[eIx] == '=' ? (sArr[eIx - 1] == '=' ? 2 : 1) : 0; // Count '=' at end.
  339. int cCnt = eIx - sIx + 1; // Content count including possible separators
  340. int sepCnt = sLen > 76 ? (sArr[76] == '\r' ? cCnt / 78 : 0) << 1 : 0;
  341. int len = ((cCnt - sepCnt) * 6 >> 3) - pad; // The number of decoded bytes
  342. byte[] dArr = new byte[len]; // Preallocate byte[] of exact length
  343. // Decode all but the last 0 - 2 bytes.
  344. int d = 0;
  345. for (int cc = 0, eLen = (len / 3) * 3; d < eLen; ) {
  346. // Assemble three bytes into an int from four "valid" characters.
  347. int i = IA[sArr[sIx++]] << 18 | IA[sArr[sIx++]] << 12 | IA[sArr[sIx++]] << 6 | IA[sArr[sIx++]];
  348. // Add the bytes
  349. dArr[d++] = (byte) (i >> 16);
  350. dArr[d++] = (byte) (i >> 8);
  351. dArr[d++] = (byte) i;
  352. // If line separator, jump over it.
  353. if (sepCnt > 0 && ++cc == 19) {
  354. sIx += 2;
  355. cc = 0;
  356. }
  357. }
  358. if (d < len) {
  359. // Decode last 1-3 bytes (incl '=') into 1-3 bytes
  360. int i = 0;
  361. for (int j = 0; sIx <= eIx - pad; j++) {
  362. i |= IA[sArr[sIx++]] << (18 - j * 6);
  363. }
  364. for (int r = 16; d < len; r -= 8) {
  365. dArr[d++] = (byte) (i >> r);
  366. }
  367. }
  368. return dArr;
  369. }
  370. // ****************************************************************************************
  371. // * String version
  372. // ****************************************************************************************
  373. /**
  374. * Encodes a raw byte array into a BASE64 <code>String</code> representation i accordance with RFC 2045.
  375. *
  376. * @param sArr The bytes to convert. If <code>null</code> or length 0 an empty array will be returned.
  377. * @param lineSep Optional "\r\n" after 76 characters, unless end of file.<br> No line separator will be in breach of RFC 2045 which specifies max 76 per line but will be a little faster.
  378. * @return A BASE64 encoded array. Never <code>null</code>.
  379. */
  380. public final static String encodeToString(byte[] sArr, boolean lineSep) {
  381. // Reuse char[] since we can't create a String incrementally anyway and StringBuffer/Builder would be slower.
  382. return new String(encodeToChar(sArr, lineSep));
  383. }
  384. /**
  385. * Decodes a BASE64 encoded <code>String</code>. All illegal characters will be ignored and can handle both strings with and without line separators.<br> <b>Note!</b> It can be up to about 2x the
  386. * speed to call <code>decode(str.toCharArray())</code> instead. That will create a temporary array though. This version will use <code>str.charAt(i)</code> to iterate the string.
  387. *
  388. * @param str The source string. <code>null</code> or length 0 will return an empty array.
  389. * @return The decoded array of bytes. May be of length 0. Will be <code>null</code> if the legal characters (including '=') isn't divideable by 4. (I.e. definitely corrupted).
  390. */
  391. public final static byte[] decode(String str) {
  392. // Check special case
  393. int sLen = str != null ? str.length() : 0;
  394. if (sLen == 0) {
  395. return new byte[0];
  396. }
  397. // Count illegal characters (including '\r', '\n') to know what size the returned array will be,
  398. // so we don't have to reallocate & copy it later.
  399. int sepCnt = 0; // Number of separator characters. (Actually illegal characters, but that's a bonus[])
  400. for (int i = 0; i < sLen; i++) // If input is "pure" (I.e. no line separators or illegal chars) base64 this loop can be commented out.
  401. {
  402. if (IA[str.charAt(i)] < 0) {
  403. sepCnt++;
  404. }
  405. }
  406. // Check so that legal chars (including '=') are evenly divideable by 4 as specified in RFC 2045.
  407. if ((sLen - sepCnt) % 4 != 0) {
  408. return null;
  409. }
  410. // Count '=' at end
  411. int pad = 0;
  412. for (int i = sLen; i > 1 && IA[str.charAt(--i)] <= 0; ) {
  413. if (str.charAt(i) == '=') {
  414. pad++;
  415. }
  416. }
  417. int len = ((sLen - sepCnt) * 6 >> 3) - pad;
  418. byte[] dArr = new byte[len]; // Preallocate byte[] of exact length
  419. for (int s = 0, d = 0; d < len; ) {
  420. // Assemble three bytes into an int from four "valid" characters.
  421. int i = 0;
  422. for (int j = 0; j < 4; j++) { // j only increased if a valid char was found.
  423. int c = IA[str.charAt(s++)];
  424. if (c >= 0) {
  425. i |= c << (18 - j * 6);
  426. } else {
  427. j--;
  428. }
  429. }
  430. // Add the bytes
  431. dArr[d++] = (byte) (i >> 16);
  432. if (d < len) {
  433. dArr[d++] = (byte) (i >> 8);
  434. if (d < len) {
  435. dArr[d++] = (byte) i;
  436. }
  437. }
  438. }
  439. return dArr;
  440. }
  441. /**
  442. * Decodes a BASE64 encoded string that is known to be resonably well formatted. The method is about twice as fast as {@link #decode(String)}. The preconditions are:<br> + The array must have a
  443. * line length of 76 chars OR no line separators at all (one line).<br> + Line separator must be "\r\n", as specified in RFC 2045 + The array must not contain illegal characters within the encoded
  444. * string<br> + The array CAN have illegal characters at the beginning and end, those will be dealt with appropriately.<br>
  445. *
  446. * @param s The source string. Length 0 will return an empty array. <code>null</code> will throw an exception.
  447. * @return The decoded array of bytes. May be of length 0.
  448. */
  449. public final static byte[] decodeFast(String s) {
  450. // Check special case
  451. int sLen = s.length();
  452. if (sLen == 0) {
  453. return new byte[0];
  454. }
  455. int sIx = 0, eIx = sLen - 1; // Start and end index after trimming.
  456. // Trim illegal chars from start
  457. while (sIx < eIx && IA[s.charAt(sIx) & 0xff] < 0) {
  458. sIx++;
  459. }
  460. // Trim illegal chars from end
  461. while (eIx > 0 && IA[s.charAt(eIx) & 0xff] < 0) {
  462. eIx--;
  463. }
  464. // get the padding count (=) (0, 1 or 2)
  465. int pad = s.charAt(eIx) == '=' ? (s.charAt(eIx - 1) == '=' ? 2 : 1) : 0; // Count '=' at end.
  466. int cCnt = eIx - sIx + 1; // Content count including possible separators
  467. int sepCnt = sLen > 76 ? (s.charAt(76) == '\r' ? cCnt / 78 : 0) << 1 : 0;
  468. int len = ((cCnt - sepCnt) * 6 >> 3) - pad; // The number of decoded bytes
  469. byte[] dArr = new byte[len]; // Preallocate byte[] of exact length
  470. // Decode all but the last 0 - 2 bytes.
  471. int d = 0;
  472. for (int cc = 0, eLen = (len / 3) * 3; d < eLen; ) {
  473. // Assemble three bytes into an int from four "valid" characters.
  474. int i = IA[s.charAt(sIx++)] << 18 | IA[s.charAt(sIx++)] << 12 | IA[s.charAt(sIx++)] << 6 | IA[s.charAt(sIx++)];
  475. // Add the bytes
  476. dArr[d++] = (byte) (i >> 16);
  477. dArr[d++] = (byte) (i >> 8);
  478. dArr[d++] = (byte) i;
  479. // If line separator, jump over it.
  480. if (sepCnt > 0 && ++cc == 19) {
  481. sIx += 2;
  482. cc = 0;
  483. }
  484. }
  485. if (d < len) {
  486. // Decode last 1-3 bytes (incl '=') into 1-3 bytes
  487. int i = 0;
  488. for (int j = 0; sIx <= eIx - pad; j++) {
  489. i |= IA[s.charAt(sIx++)] << (18 - j * 6);
  490. }
  491. for (int r = 16; d < len; r -= 8) {
  492. dArr[d++] = (byte) (i >> r);
  493. }
  494. }
  495. return dArr;
  496. }
  497. }