001/* 002 * Licensed to the Apache Software Foundation (ASF) under one 003 * or more contributor license agreements. See the NOTICE file 004 * distributed with this work for additional information 005 * regarding copyright ownership. The ASF licenses this file 006 * to you under the Apache License, Version 2.0 (the 007 * "License"); you may not use this file except in compliance 008 * with the License. You may obtain a copy of the License at 009 * 010 * http://www.apache.org/licenses/LICENSE-2.0 011 * 012 * Unless required by applicable law or agreed to in writing, software 013 * distributed under the License is distributed on an "AS IS" BASIS, 014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 015 * See the License for the specific language governing permissions and 016 * limitations under the License. 017 */ 018package org.apache.hadoop.hbase.util; 019 020import static org.apache.hbase.thirdparty.com.google.common.base.Preconditions.checkArgument; 021import static org.apache.hbase.thirdparty.com.google.common.base.Preconditions.checkNotNull; 022import static org.apache.hbase.thirdparty.com.google.common.base.Preconditions.checkPositionIndex; 023 024import java.io.DataInput; 025import java.io.DataOutput; 026import java.io.IOException; 027import java.io.UnsupportedEncodingException; 028import java.math.BigDecimal; 029import java.math.BigInteger; 030import java.nio.ByteBuffer; 031import java.nio.charset.StandardCharsets; 032import java.security.SecureRandom; 033import java.util.ArrayList; 034import java.util.Arrays; 035import java.util.Collection; 036import java.util.Collections; 037import java.util.Comparator; 038import java.util.Iterator; 039import java.util.List; 040import java.util.Random; 041import org.apache.hadoop.hbase.Cell; 042import org.apache.hadoop.hbase.CellComparator; 043import org.apache.hadoop.hbase.unsafe.HBasePlatformDependent; 044import org.apache.hadoop.io.RawComparator; 045import org.apache.hadoop.io.WritableComparator; 046import org.apache.hadoop.io.WritableUtils; 047import org.apache.yetus.audience.InterfaceAudience; 048import org.slf4j.Logger; 049import org.slf4j.LoggerFactory; 050 051import org.apache.hbase.thirdparty.org.apache.commons.collections4.CollectionUtils; 052 053/** 054 * Utility class that handles byte arrays, conversions to/from other types, comparisons, hash code 055 * generation, manufacturing keys for HashMaps or HashSets, and can be used as key in maps or trees. 056 */ 057@InterfaceAudience.Public 058@edu.umd.cs.findbugs.annotations.SuppressWarnings( 059 value = "EQ_CHECK_FOR_OPERAND_NOT_COMPATIBLE_WITH_THIS", 060 justification = "It has been like this forever") 061@SuppressWarnings("MixedMutabilityReturnType") 062public class Bytes implements Comparable<Bytes> { 063 064 // Using the charset canonical name for String/byte[] conversions is much 065 // more efficient due to use of cached encoders/decoders. 066 private static final String UTF8_CSN = StandardCharsets.UTF_8.name(); 067 068 // HConstants.EMPTY_BYTE_ARRAY should be updated if this changed 069 private static final byte[] EMPTY_BYTE_ARRAY = new byte[0]; 070 071 private static final Logger LOG = LoggerFactory.getLogger(Bytes.class); 072 073 /** 074 * Size of boolean in bytes 075 */ 076 public static final int SIZEOF_BOOLEAN = Byte.SIZE / Byte.SIZE; 077 078 /** 079 * Size of byte in bytes 080 */ 081 public static final int SIZEOF_BYTE = SIZEOF_BOOLEAN; 082 083 /** 084 * Size of char in bytes 085 */ 086 public static final int SIZEOF_CHAR = Character.SIZE / Byte.SIZE; 087 088 /** 089 * Size of double in bytes 090 */ 091 public static final int SIZEOF_DOUBLE = Double.SIZE / Byte.SIZE; 092 093 /** 094 * Size of float in bytes 095 */ 096 public static final int SIZEOF_FLOAT = Float.SIZE / Byte.SIZE; 097 098 /** 099 * Size of int in bytes 100 */ 101 public static final int SIZEOF_INT = Integer.SIZE / Byte.SIZE; 102 103 /** 104 * Size of long in bytes 105 */ 106 public static final int SIZEOF_LONG = Long.SIZE / Byte.SIZE; 107 108 /** 109 * Size of short in bytes 110 */ 111 public static final int SIZEOF_SHORT = Short.SIZE / Byte.SIZE; 112 113 /** 114 * Mask to apply to a long to reveal the lower int only. Use like this: int i = 115 * (int)(0xFFFFFFFF00000000L ^ some_long_value); 116 */ 117 public static final long MASK_FOR_LOWER_INT_IN_LONG = 0xFFFFFFFF00000000L; 118 119 /** 120 * Estimate of size cost to pay beyond payload in jvm for instance of byte []. Estimate based on 121 * study of jhat and jprofiler numbers. 122 */ 123 // JHat says BU is 56 bytes. 124 // SizeOf which uses java.lang.instrument says 24 bytes. (3 longs?) 125 public static final int ESTIMATED_HEAP_TAX = 16; 126 127 static final boolean UNSAFE_UNALIGNED = HBasePlatformDependent.unaligned(); 128 129 /** 130 * Returns length of the byte array, returning 0 if the array is null. Useful for calculating 131 * sizes. 132 * @param b byte array, which can be null 133 * @return 0 if b is null, otherwise returns length 134 */ 135 final public static int len(byte[] b) { 136 return b == null ? 0 : b.length; 137 } 138 139 private byte[] bytes; 140 private int offset; 141 private int length; 142 143 /** 144 * Create a zero-size sequence. 145 */ 146 public Bytes() { 147 super(); 148 } 149 150 /** 151 * Create a Bytes using the byte array as the initial value. 152 * @param bytes This array becomes the backing storage for the object. 153 */ 154 public Bytes(byte[] bytes) { 155 this(bytes, 0, bytes.length); 156 } 157 158 /** 159 * Set the new Bytes to the contents of the passed <code>ibw</code>. 160 * @param ibw the value to set this Bytes to. 161 */ 162 public Bytes(final Bytes ibw) { 163 this(ibw.get(), ibw.getOffset(), ibw.getLength()); 164 } 165 166 /** 167 * Set the value to a given byte range 168 * @param bytes the new byte range to set to 169 * @param offset the offset in newData to start at 170 * @param length the number of bytes in the range 171 */ 172 public Bytes(final byte[] bytes, final int offset, final int length) { 173 this.bytes = bytes; 174 this.offset = offset; 175 this.length = length; 176 } 177 178 /** 179 * Get the data from the Bytes. 180 * @return The data is only valid between offset and offset+length. 181 */ 182 public byte[] get() { 183 if (this.bytes == null) { 184 throw new IllegalStateException( 185 "Uninitialiized. Null constructor " + "called w/o accompaying readFields invocation"); 186 } 187 return this.bytes; 188 } 189 190 /** Use passed bytes as backing array for this instance. */ 191 public void set(final byte[] b) { 192 set(b, 0, b.length); 193 } 194 195 /** Use passed bytes as backing array for this instance. */ 196 public void set(final byte[] b, final int offset, final int length) { 197 this.bytes = b; 198 this.offset = offset; 199 this.length = length; 200 } 201 202 /** Returns the number of valid bytes in the buffer */ 203 public int getLength() { 204 if (this.bytes == null) { 205 throw new IllegalStateException( 206 "Uninitialiized. Null constructor " + "called w/o accompaying readFields invocation"); 207 } 208 return this.length; 209 } 210 211 /** Return the offset into the buffer. */ 212 public int getOffset() { 213 return this.offset; 214 } 215 216 @Override 217 public int hashCode() { 218 return Bytes.hashCode(bytes, offset, length); 219 } 220 221 /** 222 * Define the sort order of the Bytes. 223 * @param that The other bytes writable 224 * @return Positive if left is bigger than right, 0 if they are equal, and negative if left is 225 * smaller than right. 226 */ 227 @Override 228 public int compareTo(Bytes that) { 229 return BYTES_RAWCOMPARATOR.compare(this.bytes, this.offset, this.length, that.bytes, 230 that.offset, that.length); 231 } 232 233 /** 234 * Compares the bytes in this object to the specified byte array 235 * @return Positive if left is bigger than right, 0 if they are equal, and negative if left is 236 * smaller than right. 237 */ 238 public int compareTo(final byte[] that) { 239 return BYTES_RAWCOMPARATOR.compare(this.bytes, this.offset, this.length, that, 0, that.length); 240 } 241 242 @Override 243 public boolean equals(Object right_obj) { 244 if (right_obj instanceof byte[]) { 245 return compareTo((byte[]) right_obj) == 0; 246 } 247 if (right_obj instanceof Bytes) { 248 return compareTo((Bytes) right_obj) == 0; 249 } 250 return false; 251 } 252 253 @Override 254 public String toString() { 255 return Bytes.toString(bytes, offset, length); 256 } 257 258 /** 259 * Convert a list of byte[] to an array 260 * @param array List of byte []. 261 * @return Array of byte []. 262 */ 263 public static byte[][] toArray(final List<byte[]> array) { 264 // List#toArray doesn't work on lists of byte []. 265 byte[][] results = new byte[array.size()][]; 266 for (int i = 0; i < array.size(); i++) { 267 results[i] = array.get(i); 268 } 269 return results; 270 } 271 272 /** Returns a copy of the bytes referred to by this writable */ 273 public byte[] copyBytes() { 274 return Arrays.copyOfRange(bytes, offset, offset + length); 275 } 276 277 /** Byte array comparator class. */ 278 @InterfaceAudience.Public 279 public static class ByteArrayComparator implements RawComparator<byte[]> { 280 281 public ByteArrayComparator() { 282 super(); 283 } 284 285 @Override 286 public int compare(byte[] left, byte[] right) { 287 return compareTo(left, right); 288 } 289 290 @Override 291 public int compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2) { 292 return LexicographicalComparerHolder.BEST_COMPARER.compareTo(b1, s1, l1, b2, s2, l2); 293 } 294 } 295 296 /** 297 * A {@link ByteArrayComparator} that treats the empty array as the largest value. This is useful 298 * for comparing row end keys for regions. 299 */ 300 // TODO: unfortunately, HBase uses byte[0] as both start and end keys for region 301 // boundaries. Thus semantically, we should treat empty byte array as the smallest value 302 // while comparing row keys, start keys etc; but as the largest value for comparing 303 // region boundaries for endKeys. 304 @InterfaceAudience.Public 305 public static class RowEndKeyComparator extends ByteArrayComparator { 306 @Override 307 public int compare(byte[] left, byte[] right) { 308 return compare(left, 0, left.length, right, 0, right.length); 309 } 310 311 @Override 312 public int compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2) { 313 if (b1 == b2 && s1 == s2 && l1 == l2) { 314 return 0; 315 } 316 if (l1 == 0) { 317 return l2; // 0 or positive 318 } 319 if (l2 == 0) { 320 return -1; 321 } 322 return super.compare(b1, s1, l1, b2, s2, l2); 323 } 324 } 325 326 /** Pass this to TreeMaps where byte [] are keys. */ 327 public final static Comparator<byte[]> BYTES_COMPARATOR = new ByteArrayComparator(); 328 329 /** Use comparing byte arrays, byte-by-byte */ 330 public final static RawComparator<byte[]> BYTES_RAWCOMPARATOR = new ByteArrayComparator(); 331 332 /** 333 * Read byte-array written with a WritableableUtils.vint prefix. 334 * @param in Input to read from. 335 * @return byte array read off <code>in</code> 336 * @throws IOException e 337 */ 338 public static byte[] readByteArray(final DataInput in) throws IOException { 339 int len = WritableUtils.readVInt(in); 340 if (len < 0) { 341 throw new NegativeArraySizeException(Integer.toString(len)); 342 } 343 byte[] result = new byte[len]; 344 in.readFully(result, 0, len); 345 return result; 346 } 347 348 /** 349 * Read byte-array written with a WritableableUtils.vint prefix. IOException is converted to a 350 * RuntimeException. 351 * @param in Input to read from. 352 * @return byte array read off <code>in</code> 353 */ 354 public static byte[] readByteArrayThrowsRuntime(final DataInput in) { 355 try { 356 return readByteArray(in); 357 } catch (Exception e) { 358 throw new RuntimeException(e); 359 } 360 } 361 362 /** 363 * Write byte-array with a WritableableUtils.vint prefix. 364 * @param out output stream to be written to 365 * @param b array to write 366 * @throws IOException e 367 */ 368 public static void writeByteArray(final DataOutput out, final byte[] b) throws IOException { 369 if (b == null) { 370 WritableUtils.writeVInt(out, 0); 371 } else { 372 writeByteArray(out, b, 0, b.length); 373 } 374 } 375 376 /** 377 * Write byte-array to out with a vint length prefix. 378 * @param out output stream 379 * @param b array 380 * @param offset offset into array 381 * @param length length past offset 382 * @throws IOException e 383 */ 384 public static void writeByteArray(final DataOutput out, final byte[] b, final int offset, 385 final int length) throws IOException { 386 WritableUtils.writeVInt(out, length); 387 out.write(b, offset, length); 388 } 389 390 /** 391 * Write byte-array from src to tgt with a vint length prefix. 392 * @param tgt target array 393 * @param tgtOffset offset into target array 394 * @param src source array 395 * @param srcOffset source offset 396 * @param srcLength source length 397 * @return New offset in src array. 398 */ 399 public static int writeByteArray(final byte[] tgt, final int tgtOffset, final byte[] src, 400 final int srcOffset, final int srcLength) { 401 byte[] vint = vintToBytes(srcLength); 402 System.arraycopy(vint, 0, tgt, tgtOffset, vint.length); 403 int offset = tgtOffset + vint.length; 404 System.arraycopy(src, srcOffset, tgt, offset, srcLength); 405 return offset + srcLength; 406 } 407 408 /** 409 * Put bytes at the specified byte array position. 410 * @param tgtBytes the byte array 411 * @param tgtOffset position in the array 412 * @param srcBytes array to write out 413 * @param srcOffset source offset 414 * @param srcLength source length 415 * @return incremented offset 416 */ 417 public static int putBytes(byte[] tgtBytes, int tgtOffset, byte[] srcBytes, int srcOffset, 418 int srcLength) { 419 System.arraycopy(srcBytes, srcOffset, tgtBytes, tgtOffset, srcLength); 420 return tgtOffset + srcLength; 421 } 422 423 /** 424 * Write a single byte out to the specified byte array position. 425 * @param bytes the byte array 426 * @param offset position in the array 427 * @param b byte to write out 428 * @return incremented offset 429 */ 430 public static int putByte(byte[] bytes, int offset, byte b) { 431 bytes[offset] = b; 432 return offset + 1; 433 } 434 435 /** 436 * Add the whole content of the ByteBuffer to the bytes arrays. The ByteBuffer is modified. 437 * @param bytes the byte array 438 * @param offset position in the array 439 * @param buf ByteBuffer to write out 440 * @return incremented offset 441 */ 442 public static int putByteBuffer(byte[] bytes, int offset, ByteBuffer buf) { 443 int len = buf.remaining(); 444 buf.get(bytes, offset, len); 445 return offset + len; 446 } 447 448 /** 449 * Returns a new byte array, copied from the given {@code buf}, from the index 0 (inclusive) to 450 * the limit (exclusive), regardless of the current position. The position and the other index 451 * parameters are not changed. 452 * @param buf a byte buffer 453 * @return the byte array 454 * @see #getBytes(ByteBuffer) 455 */ 456 public static byte[] toBytes(ByteBuffer buf) { 457 ByteBuffer dup = buf.duplicate(); 458 dup.position(0); 459 return readBytes(dup); 460 } 461 462 private static byte[] readBytes(ByteBuffer buf) { 463 byte[] result = new byte[buf.remaining()]; 464 buf.get(result); 465 return result; 466 } 467 468 /** 469 * Convert a byte[] into a string. Charset is assumed to be UTF-8. 470 * @param b Presumed UTF-8 encoded byte array. 471 * @return String made from <code>b</code> 472 */ 473 public static String toString(final byte[] b) { 474 if (b == null) { 475 return null; 476 } 477 return toString(b, 0, b.length); 478 } 479 480 /** 481 * Joins two byte arrays together using a separator. 482 * @param b1 The first byte array. 483 * @param sep The separator to use. 484 * @param b2 The second byte array. 485 */ 486 public static String toString(final byte[] b1, String sep, final byte[] b2) { 487 return toString(b1, 0, b1.length) + sep + toString(b2, 0, b2.length); 488 } 489 490 /** 491 * This method will convert utf8 encoded bytes into a string. If the given byte array is null, 492 * this method will return null. 493 * @param b Presumed UTF-8 encoded byte array. 494 * @param off offset into array 495 * @return String made from <code>b</code> or null 496 */ 497 public static String toString(final byte[] b, int off) { 498 if (b == null) { 499 return null; 500 } 501 int len = b.length - off; 502 if (len <= 0) { 503 return ""; 504 } 505 try { 506 return new String(b, off, len, UTF8_CSN); 507 } catch (UnsupportedEncodingException e) { 508 // should never happen! 509 throw new IllegalArgumentException("UTF8 encoding is not supported", e); 510 } 511 } 512 513 /** 514 * This method will convert utf8 encoded bytes into a string. If the given byte array is null, 515 * this method will return null. 516 * @param b Presumed UTF-8 encoded byte array. 517 * @param off offset into array 518 * @param len length of utf-8 sequence 519 * @return String made from <code>b</code> or null 520 */ 521 public static String toString(final byte[] b, int off, int len) { 522 if (b == null) { 523 return null; 524 } 525 if (len == 0) { 526 return ""; 527 } 528 try { 529 return new String(b, off, len, UTF8_CSN); 530 } catch (UnsupportedEncodingException e) { 531 // should never happen! 532 throw new IllegalArgumentException("UTF8 encoding is not supported", e); 533 } 534 } 535 536 /** 537 * Write a printable representation of a byte array. 538 * @param b byte array 539 * @see #toStringBinary(byte[], int, int) 540 */ 541 public static String toStringBinary(final byte[] b) { 542 if (b == null) return "null"; 543 return toStringBinary(b, 0, b.length); 544 } 545 546 /** 547 * Converts the given byte buffer to a printable representation, from the index 0 (inclusive) to 548 * the limit (exclusive), regardless of the current position. The position and the other index 549 * parameters are not changed. 550 * @param buf a byte buffer 551 * @return a string representation of the buffer's binary contents 552 * @see #toBytes(ByteBuffer) 553 * @see #getBytes(ByteBuffer) 554 */ 555 public static String toStringBinary(ByteBuffer buf) { 556 if (buf == null) return "null"; 557 if (buf.hasArray()) { 558 return toStringBinary(buf.array(), buf.arrayOffset(), buf.limit()); 559 } 560 return toStringBinary(toBytes(buf)); 561 } 562 563 private static final char[] HEX_CHARS_UPPER = 564 { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; 565 566 /** 567 * Write a printable representation of a byte array. Non-printable characters are hex escaped in 568 * the format \\x%02X, eg: \x00 \x05 etc 569 * @param b array to write out 570 * @param off offset to start at 571 * @param len length to write 572 * @return string output 573 */ 574 public static String toStringBinary(final byte[] b, int off, int len) { 575 StringBuilder result = new StringBuilder(); 576 // Just in case we are passed a 'len' that is > buffer length... 577 if (off >= b.length) return result.toString(); 578 if (off + len > b.length) len = b.length - off; 579 for (int i = off; i < off + len; ++i) { 580 int ch = b[i] & 0xFF; 581 if (ch >= ' ' && ch <= '~' && ch != '\\') { 582 result.append((char) ch); 583 } else { 584 result.append("\\x"); 585 result.append(HEX_CHARS_UPPER[ch / 0x10]); 586 result.append(HEX_CHARS_UPPER[ch % 0x10]); 587 } 588 } 589 return result.toString(); 590 } 591 592 private static boolean isHexDigit(char c) { 593 return (c >= 'A' && c <= 'F') || (c >= '0' && c <= '9'); 594 } 595 596 /** 597 * Takes a ASCII digit in the range A-F0-9 and returns the corresponding integer/ordinal value. 598 * @param ch The hex digit. 599 * @return The converted hex value as a byte. 600 */ 601 public static byte toBinaryFromHex(byte ch) { 602 if (ch >= 'A' && ch <= 'F') return (byte) ((byte) 10 + (byte) (ch - 'A')); 603 // else 604 return (byte) (ch - '0'); 605 } 606 607 public static byte[] toBytesBinary(String in) { 608 // this may be bigger than we need, but let's be safe. 609 byte[] b = new byte[in.length()]; 610 int size = 0; 611 for (int i = 0; i < in.length(); ++i) { 612 char ch = in.charAt(i); 613 if (ch == '\\' && in.length() > i + 1 && in.charAt(i + 1) == 'x') { 614 // ok, take next 2 hex digits. 615 char hd1 = in.charAt(i + 2); 616 char hd2 = in.charAt(i + 3); 617 618 // they need to be A-F0-9: 619 if (!isHexDigit(hd1) || !isHexDigit(hd2)) { 620 // bogus escape code, ignore: 621 continue; 622 } 623 // turn hex ASCII digit -> number 624 byte d = (byte) ((toBinaryFromHex((byte) hd1) << 4) + toBinaryFromHex((byte) hd2)); 625 626 b[size++] = d; 627 i += 3; // skip 3 628 } else { 629 b[size++] = (byte) ch; 630 } 631 } 632 // resize: 633 byte[] b2 = new byte[size]; 634 System.arraycopy(b, 0, b2, 0, size); 635 return b2; 636 } 637 638 /** 639 * Converts a string to a UTF-8 byte array. 640 * @param s string 641 * @return the byte array 642 */ 643 public static byte[] toBytes(String s) { 644 try { 645 return s.getBytes(UTF8_CSN); 646 } catch (UnsupportedEncodingException e) { 647 // should never happen! 648 throw new IllegalArgumentException("UTF8 decoding is not supported", e); 649 } 650 } 651 652 /** 653 * Convert a boolean to a byte array. True becomes -1 and false becomes 0. 654 * @param b value 655 * @return <code>b</code> encoded in a byte array. 656 */ 657 public static byte[] toBytes(final boolean b) { 658 return new byte[] { b ? (byte) -1 : (byte) 0 }; 659 } 660 661 /** 662 * Reverses {@link #toBytes(boolean)} 663 * @param b array 664 * @return True or false. 665 */ 666 public static boolean toBoolean(final byte[] b) { 667 if (b.length != 1) { 668 throw new IllegalArgumentException("Array has wrong size: " + b.length); 669 } 670 return b[0] != (byte) 0; 671 } 672 673 /** 674 * Convert a long value to a byte array using big-endian. 675 * @param val value to convert 676 * @return the byte array 677 */ 678 public static byte[] toBytes(long val) { 679 byte[] b = new byte[8]; 680 for (int i = 7; i > 0; i--) { 681 b[i] = (byte) val; 682 val >>>= 8; 683 } 684 b[0] = (byte) val; 685 return b; 686 } 687 688 /** 689 * Converts a byte array to a long value. Reverses {@link #toBytes(long)} 690 * @param bytes array 691 * @return the long value 692 */ 693 public static long toLong(byte[] bytes) { 694 return toLong(bytes, 0, SIZEOF_LONG); 695 } 696 697 /** 698 * Converts a byte array to a long value. Assumes there will be {@link #SIZEOF_LONG} bytes 699 * available. 700 * @param bytes bytes 701 * @param offset offset 702 * @return the long value 703 */ 704 public static long toLong(byte[] bytes, int offset) { 705 return toLong(bytes, offset, SIZEOF_LONG); 706 } 707 708 /** 709 * Converts a byte array to a long value. 710 * @param bytes array of bytes 711 * @param offset offset into array 712 * @param length length of data (must be {@link #SIZEOF_LONG}) 713 * @return the long value 714 * @throws IllegalArgumentException if length is not {@link #SIZEOF_LONG} or if there's not enough 715 * room in the array at the offset indicated. 716 */ 717 public static long toLong(byte[] bytes, int offset, final int length) { 718 if (length != SIZEOF_LONG || offset + length > bytes.length) { 719 throw explainWrongLengthOrOffset(bytes, offset, length, SIZEOF_LONG); 720 } 721 return ConverterHolder.BEST_CONVERTER.toLong(bytes, offset, length); 722 } 723 724 private static IllegalArgumentException explainWrongLengthOrOffset(final byte[] bytes, 725 final int offset, final int length, final int expectedLength) { 726 String reason; 727 if (length != expectedLength) { 728 reason = "Wrong length: " + length + ", expected " + expectedLength; 729 } else { 730 reason = "offset (" + offset + ") + length (" + length + ") exceed the" 731 + " capacity of the array: " + bytes.length; 732 } 733 return new IllegalArgumentException(reason); 734 } 735 736 /** 737 * Put a long value out to the specified byte array position. 738 * @param bytes the byte array 739 * @param offset position in the array 740 * @param val long to write out 741 * @return incremented offset 742 * @throws IllegalArgumentException if the byte array given doesn't have enough room at the offset 743 * specified. 744 */ 745 public static int putLong(byte[] bytes, int offset, long val) { 746 if (bytes.length - offset < SIZEOF_LONG) { 747 throw new IllegalArgumentException("Not enough room to put a long at" + " offset " + offset 748 + " in a " + bytes.length + " byte array"); 749 } 750 return ConverterHolder.BEST_CONVERTER.putLong(bytes, offset, val); 751 } 752 753 /** 754 * Put a float value out to the specified byte array position. Presumes float encoded as IEEE 754 755 * floating-point "single format" 756 * @param bytes byte array 757 * @return Float made from passed byte array. 758 */ 759 public static float toFloat(byte[] bytes) { 760 return toFloat(bytes, 0); 761 } 762 763 /** 764 * Put a float value out to the specified byte array position. Presumes float encoded as IEEE 754 765 * floating-point "single format" 766 * @param bytes array to convert 767 * @param offset offset into array 768 * @return Float made from passed byte array. 769 */ 770 public static float toFloat(byte[] bytes, int offset) { 771 return Float.intBitsToFloat(toInt(bytes, offset, SIZEOF_INT)); 772 } 773 774 /** 775 * Put a float value out to the specified byte array position. 776 * @param bytes byte array 777 * @param offset offset to write to 778 * @param f float value 779 * @return New offset in <code>bytes</code> 780 */ 781 public static int putFloat(byte[] bytes, int offset, float f) { 782 return putInt(bytes, offset, Float.floatToRawIntBits(f)); 783 } 784 785 /** Return the float represented as byte[] */ 786 public static byte[] toBytes(final float f) { 787 // Encode it as int 788 return Bytes.toBytes(Float.floatToRawIntBits(f)); 789 } 790 791 /** Return double made from passed bytes. */ 792 public static double toDouble(final byte[] bytes) { 793 return toDouble(bytes, 0); 794 } 795 796 /** Return double made from passed bytes. */ 797 public static double toDouble(final byte[] bytes, final int offset) { 798 return Double.longBitsToDouble(toLong(bytes, offset, SIZEOF_LONG)); 799 } 800 801 /** 802 * Put a double value out to the specified byte array position as the IEEE 754 double format. 803 * @param bytes byte array 804 * @param offset offset to write to 805 * @param d value 806 * @return New offset into array <code>bytes</code> 807 */ 808 public static int putDouble(byte[] bytes, int offset, double d) { 809 return putLong(bytes, offset, Double.doubleToLongBits(d)); 810 } 811 812 /** 813 * Serialize a double as the IEEE 754 double format output. The resultant array will be 8 bytes 814 * long. 815 * @param d value 816 * @return the double represented as byte [] 817 */ 818 public static byte[] toBytes(final double d) { 819 // Encode it as a long 820 return Bytes.toBytes(Double.doubleToRawLongBits(d)); 821 } 822 823 /** 824 * Convert an int value to a byte array. Big-endian. Same as what DataOutputStream.writeInt does. 825 * @param val value 826 * @return the byte array 827 */ 828 public static byte[] toBytes(int val) { 829 byte[] b = new byte[4]; 830 for (int i = 3; i > 0; i--) { 831 b[i] = (byte) val; 832 val >>>= 8; 833 } 834 b[0] = (byte) val; 835 return b; 836 } 837 838 /** 839 * Converts a byte array to an int value 840 * @param bytes byte array 841 * @return the int value 842 */ 843 public static int toInt(byte[] bytes) { 844 return toInt(bytes, 0, SIZEOF_INT); 845 } 846 847 /** 848 * Converts a byte array to an int value 849 * @param bytes byte array 850 * @param offset offset into array 851 * @return the int value 852 */ 853 public static int toInt(byte[] bytes, int offset) { 854 return toInt(bytes, offset, SIZEOF_INT); 855 } 856 857 /** 858 * Converts a byte array to an int value 859 * @param bytes byte array 860 * @param offset offset into array 861 * @param length length of int (has to be {@link #SIZEOF_INT}) 862 * @return the int value 863 * @throws IllegalArgumentException if length is not {@link #SIZEOF_INT} or if there's not enough 864 * room in the array at the offset indicated. 865 */ 866 public static int toInt(byte[] bytes, int offset, final int length) { 867 if (length != SIZEOF_INT || offset + length > bytes.length) { 868 throw explainWrongLengthOrOffset(bytes, offset, length, SIZEOF_INT); 869 } 870 return ConverterHolder.BEST_CONVERTER.toInt(bytes, offset, length); 871 } 872 873 /** 874 * Converts a byte array to an int value 875 * @param bytes byte array 876 * @param offset offset into array 877 * @param length how many bytes should be considered for creating int 878 * @return the int value 879 * @throws IllegalArgumentException if there's not enough room in the array at the offset 880 * indicated. 881 */ 882 public static int readAsInt(byte[] bytes, int offset, final int length) { 883 if (offset + length > bytes.length) { 884 throw new IllegalArgumentException("offset (" + offset + ") + length (" + length 885 + ") exceed the" + " capacity of the array: " + bytes.length); 886 } 887 int n = 0; 888 for (int i = offset; i < (offset + length); i++) { 889 n <<= 8; 890 n ^= bytes[i] & 0xFF; 891 } 892 return n; 893 } 894 895 /** 896 * Put an int value out to the specified byte array position. 897 * @param bytes the byte array 898 * @param offset position in the array 899 * @param val int to write out 900 * @return incremented offset 901 * @throws IllegalArgumentException if the byte array given doesn't have enough room at the offset 902 * specified. 903 */ 904 public static int putInt(byte[] bytes, int offset, int val) { 905 if (bytes.length - offset < SIZEOF_INT) { 906 throw new IllegalArgumentException("Not enough room to put an int at" + " offset " + offset 907 + " in a " + bytes.length + " byte array"); 908 } 909 return ConverterHolder.BEST_CONVERTER.putInt(bytes, offset, val); 910 } 911 912 /** 913 * Convert a short value to a byte array of {@link #SIZEOF_SHORT} bytes long. 914 * @param val value 915 * @return the byte array 916 */ 917 public static byte[] toBytes(short val) { 918 byte[] b = new byte[SIZEOF_SHORT]; 919 b[1] = (byte) val; 920 val >>= 8; 921 b[0] = (byte) val; 922 return b; 923 } 924 925 /** 926 * Converts a byte array to a short value 927 * @param bytes byte array 928 * @return the short value 929 */ 930 public static short toShort(byte[] bytes) { 931 return toShort(bytes, 0, SIZEOF_SHORT); 932 } 933 934 /** 935 * Converts a byte array to a short value 936 * @param bytes byte array 937 * @param offset offset into array 938 * @return the short value 939 */ 940 public static short toShort(byte[] bytes, int offset) { 941 return toShort(bytes, offset, SIZEOF_SHORT); 942 } 943 944 /** 945 * Converts a byte array to a short value 946 * @param bytes byte array 947 * @param offset offset into array 948 * @param length length, has to be {@link #SIZEOF_SHORT} 949 * @return the short value 950 * @throws IllegalArgumentException if length is not {@link #SIZEOF_SHORT} or if there's not 951 * enough room in the array at the offset indicated. 952 */ 953 public static short toShort(byte[] bytes, int offset, final int length) { 954 if (length != SIZEOF_SHORT || offset + length > bytes.length) { 955 throw explainWrongLengthOrOffset(bytes, offset, length, SIZEOF_SHORT); 956 } 957 return ConverterHolder.BEST_CONVERTER.toShort(bytes, offset, length); 958 } 959 960 /** 961 * Returns a new byte array, copied from the given {@code buf}, from the position (inclusive) to 962 * the limit (exclusive). The position and the other index parameters are not changed. 963 * @param buf a byte buffer 964 * @return the byte array 965 * @see #toBytes(ByteBuffer) 966 */ 967 public static byte[] getBytes(ByteBuffer buf) { 968 return readBytes(buf.duplicate()); 969 } 970 971 /** 972 * Put a short value out to the specified byte array position. 973 * @param bytes the byte array 974 * @param offset position in the array 975 * @param val short to write out 976 * @return incremented offset 977 * @throws IllegalArgumentException if the byte array given doesn't have enough room at the offset 978 * specified. 979 */ 980 public static int putShort(byte[] bytes, int offset, short val) { 981 if (bytes.length - offset < SIZEOF_SHORT) { 982 throw new IllegalArgumentException("Not enough room to put a short at" + " offset " + offset 983 + " in a " + bytes.length + " byte array"); 984 } 985 return ConverterHolder.BEST_CONVERTER.putShort(bytes, offset, val); 986 } 987 988 /** 989 * Put an int value as short out to the specified byte array position. Only the lower 2 bytes of 990 * the short will be put into the array. The caller of the API need to make sure they will not 991 * loose the value by doing so. This is useful to store an unsigned short which is represented as 992 * int in other parts. 993 * @param bytes the byte array 994 * @param offset position in the array 995 * @param val value to write out 996 * @return incremented offset 997 * @throws IllegalArgumentException if the byte array given doesn't have enough room at the offset 998 * specified. 999 */ 1000 public static int putAsShort(byte[] bytes, int offset, int val) { 1001 if (bytes.length - offset < SIZEOF_SHORT) { 1002 throw new IllegalArgumentException("Not enough room to put a short at" + " offset " + offset 1003 + " in a " + bytes.length + " byte array"); 1004 } 1005 bytes[offset + 1] = (byte) val; 1006 val >>= 8; 1007 bytes[offset] = (byte) val; 1008 return offset + SIZEOF_SHORT; 1009 } 1010 1011 /** Convert a BigDecimal value to a byte array */ 1012 public static byte[] toBytes(BigDecimal val) { 1013 byte[] valueBytes = val.unscaledValue().toByteArray(); 1014 byte[] result = new byte[valueBytes.length + SIZEOF_INT]; 1015 int offset = putInt(result, 0, val.scale()); 1016 putBytes(result, offset, valueBytes, 0, valueBytes.length); 1017 return result; 1018 } 1019 1020 /** Converts a byte array to a BigDecimal */ 1021 public static BigDecimal toBigDecimal(byte[] bytes) { 1022 return toBigDecimal(bytes, 0, bytes.length); 1023 } 1024 1025 /** Converts a byte array to a BigDecimal value */ 1026 public static BigDecimal toBigDecimal(byte[] bytes, int offset, final int length) { 1027 if (bytes == null || length < SIZEOF_INT + 1 || (offset + length > bytes.length)) { 1028 return null; 1029 } 1030 1031 int scale = toInt(bytes, offset); 1032 byte[] tcBytes = new byte[length - SIZEOF_INT]; 1033 System.arraycopy(bytes, offset + SIZEOF_INT, tcBytes, 0, length - SIZEOF_INT); 1034 return new BigDecimal(new BigInteger(tcBytes), scale); 1035 } 1036 1037 /** 1038 * Put a BigDecimal value out to the specified byte array position. 1039 * @param bytes the byte array 1040 * @param offset position in the array 1041 * @param val BigDecimal to write out 1042 * @return incremented offset 1043 */ 1044 public static int putBigDecimal(byte[] bytes, int offset, BigDecimal val) { 1045 if (bytes == null) { 1046 return offset; 1047 } 1048 1049 byte[] valueBytes = val.unscaledValue().toByteArray(); 1050 byte[] result = new byte[valueBytes.length + SIZEOF_INT]; 1051 offset = putInt(result, offset, val.scale()); 1052 return putBytes(result, offset, valueBytes, 0, valueBytes.length); 1053 } 1054 1055 /** 1056 * Encode a long value as a variable length integer. 1057 * @param vint Integer to make a vint of. 1058 * @return Vint as bytes array. 1059 */ 1060 public static byte[] vintToBytes(final long vint) { 1061 long i = vint; 1062 int size = WritableUtils.getVIntSize(i); 1063 byte[] result = new byte[size]; 1064 int offset = 0; 1065 if (i >= -112 && i <= 127) { 1066 result[offset] = (byte) i; 1067 return result; 1068 } 1069 1070 int len = -112; 1071 if (i < 0) { 1072 i ^= -1L; // take one's complement' 1073 len = -120; 1074 } 1075 1076 long tmp = i; 1077 while (tmp != 0) { 1078 tmp = tmp >> 8; 1079 len--; 1080 } 1081 1082 result[offset++] = (byte) len; 1083 1084 len = (len < -120) ? -(len + 120) : -(len + 112); 1085 1086 for (int idx = len; idx != 0; idx--) { 1087 int shiftbits = (idx - 1) * 8; 1088 long mask = 0xFFL << shiftbits; 1089 result[offset++] = (byte) ((i & mask) >> shiftbits); 1090 } 1091 return result; 1092 } 1093 1094 /** 1095 * Reads a zero-compressed encoded long from input buffer and returns it. 1096 * @param buffer buffer to convert 1097 * @return vint bytes as an integer. 1098 */ 1099 public static long bytesToVint(final byte[] buffer) { 1100 int offset = 0; 1101 byte firstByte = buffer[offset++]; 1102 int len = WritableUtils.decodeVIntSize(firstByte); 1103 if (len == 1) { 1104 return firstByte; 1105 } 1106 long i = 0; 1107 for (int idx = 0; idx < len - 1; idx++) { 1108 byte b = buffer[offset++]; 1109 i = i << 8; 1110 i = i | (b & 0xFF); 1111 } 1112 return (WritableUtils.isNegativeVInt(firstByte) ? ~i : i); 1113 } 1114 1115 /** 1116 * Reads a zero-compressed encoded long from input buffer and returns it. 1117 * @param buffer Binary array 1118 * @param offset Offset into array at which vint begins. 1119 * @return deserialized long from buffer. 1120 */ 1121 public static long readAsVLong(final byte[] buffer, final int offset) { 1122 byte firstByte = buffer[offset]; 1123 int len = WritableUtils.decodeVIntSize(firstByte); 1124 if (len == 1) { 1125 return firstByte; 1126 } 1127 long i = 0; 1128 for (int idx = 0; idx < len - 1; idx++) { 1129 byte b = buffer[offset + 1 + idx]; 1130 i = i << 8; 1131 i = i | (b & 0xFF); 1132 } 1133 return (WritableUtils.isNegativeVInt(firstByte) ? ~i : i); 1134 } 1135 1136 /** 1137 * Lexicographically compare two arrays. 1138 * @param left left operand 1139 * @param right right operand 1140 * @return 0 if equal, < 0 if left is less than right, etc. 1141 */ 1142 public static int compareTo(final byte[] left, final byte[] right) { 1143 return LexicographicalComparerHolder.BEST_COMPARER.compareTo(left, 0, 1144 left == null ? 0 : left.length, right, 0, right == null ? 0 : right.length); 1145 } 1146 1147 /** 1148 * Lexicographically compare two arrays. 1149 * @param buffer1 left operand 1150 * @param buffer2 right operand 1151 * @param offset1 Where to start comparing in the left buffer 1152 * @param offset2 Where to start comparing in the right buffer 1153 * @param length1 How much to compare from the left buffer 1154 * @param length2 How much to compare from the right buffer 1155 * @return 0 if equal, < 0 if left is less than right, etc. 1156 */ 1157 public static int compareTo(byte[] buffer1, int offset1, int length1, byte[] buffer2, int offset2, 1158 int length2) { 1159 return LexicographicalComparerHolder.BEST_COMPARER.compareTo(buffer1, offset1, length1, buffer2, 1160 offset2, length2); 1161 } 1162 1163 interface Comparer<T> { 1164 int compareTo(T buffer1, int offset1, int length1, T buffer2, int offset2, int length2); 1165 } 1166 1167 static abstract class Converter { 1168 abstract long toLong(byte[] bytes, int offset, int length); 1169 1170 abstract int putLong(byte[] bytes, int offset, long val); 1171 1172 abstract int toInt(byte[] bytes, int offset, final int length); 1173 1174 abstract int putInt(byte[] bytes, int offset, int val); 1175 1176 abstract short toShort(byte[] bytes, int offset, final int length); 1177 1178 abstract int putShort(byte[] bytes, int offset, short val); 1179 1180 } 1181 1182 static abstract class CommonPrefixer { 1183 abstract int findCommonPrefix(byte[] left, int leftOffset, int leftLength, byte[] right, 1184 int rightOffset, int rightLength); 1185 } 1186 1187 static Comparer<byte[]> lexicographicalComparerJavaImpl() { 1188 return LexicographicalComparerHolder.PureJavaComparer.INSTANCE; 1189 } 1190 1191 static class ConverterHolder { 1192 static final String UNSAFE_CONVERTER_NAME = 1193 ConverterHolder.class.getName() + "$UnsafeConverter"; 1194 1195 static final Converter BEST_CONVERTER = getBestConverter(); 1196 1197 /** 1198 * Returns the Unsafe-using Converter, or falls back to the pure-Java implementation if unable 1199 * to do so. 1200 */ 1201 static Converter getBestConverter() { 1202 try { 1203 Class<?> theClass = Class.forName(UNSAFE_CONVERTER_NAME); 1204 1205 // yes, UnsafeComparer does implement Comparer<byte[]> 1206 @SuppressWarnings("unchecked") 1207 Converter converter = (Converter) theClass.getConstructor().newInstance(); 1208 return converter; 1209 } catch (Throwable t) { // ensure we really catch *everything* 1210 return PureJavaConverter.INSTANCE; 1211 } 1212 } 1213 1214 protected static final class PureJavaConverter extends Converter { 1215 static final PureJavaConverter INSTANCE = new PureJavaConverter(); 1216 1217 private PureJavaConverter() { 1218 } 1219 1220 @Override 1221 long toLong(byte[] bytes, int offset, int length) { 1222 long l = 0; 1223 for (int i = offset; i < offset + length; i++) { 1224 l <<= 8; 1225 l ^= bytes[i] & 0xFF; 1226 } 1227 return l; 1228 } 1229 1230 @Override 1231 int putLong(byte[] bytes, int offset, long val) { 1232 for (int i = offset + 7; i > offset; i--) { 1233 bytes[i] = (byte) val; 1234 val >>>= 8; 1235 } 1236 bytes[offset] = (byte) val; 1237 return offset + SIZEOF_LONG; 1238 } 1239 1240 @Override 1241 int toInt(byte[] bytes, int offset, int length) { 1242 int n = 0; 1243 for (int i = offset; i < (offset + length); i++) { 1244 n <<= 8; 1245 n ^= bytes[i] & 0xFF; 1246 } 1247 return n; 1248 } 1249 1250 @Override 1251 int putInt(byte[] bytes, int offset, int val) { 1252 for (int i = offset + 3; i > offset; i--) { 1253 bytes[i] = (byte) val; 1254 val >>>= 8; 1255 } 1256 bytes[offset] = (byte) val; 1257 return offset + SIZEOF_INT; 1258 } 1259 1260 @Override 1261 short toShort(byte[] bytes, int offset, int length) { 1262 short n = 0; 1263 n = (short) ((n ^ bytes[offset]) & 0xFF); 1264 n = (short) (n << 8); 1265 n ^= (short) (bytes[offset + 1] & 0xFF); 1266 return n; 1267 } 1268 1269 @Override 1270 int putShort(byte[] bytes, int offset, short val) { 1271 bytes[offset + 1] = (byte) val; 1272 val >>= 8; 1273 bytes[offset] = (byte) val; 1274 return offset + SIZEOF_SHORT; 1275 } 1276 } 1277 1278 protected static final class UnsafeConverter extends Converter { 1279 1280 public UnsafeConverter() { 1281 } 1282 1283 static { 1284 if (!UNSAFE_UNALIGNED) { 1285 // It doesn't matter what we throw; 1286 // it's swallowed in getBestComparer(). 1287 throw new Error(); 1288 } 1289 1290 // sanity check - this should never fail 1291 if (HBasePlatformDependent.arrayIndexScale(byte[].class) != 1) { 1292 throw new AssertionError(); 1293 } 1294 } 1295 1296 @Override 1297 long toLong(byte[] bytes, int offset, int length) { 1298 return UnsafeAccess.toLong(bytes, offset); 1299 } 1300 1301 @Override 1302 int putLong(byte[] bytes, int offset, long val) { 1303 return UnsafeAccess.putLong(bytes, offset, val); 1304 } 1305 1306 @Override 1307 int toInt(byte[] bytes, int offset, int length) { 1308 return UnsafeAccess.toInt(bytes, offset); 1309 } 1310 1311 @Override 1312 int putInt(byte[] bytes, int offset, int val) { 1313 return UnsafeAccess.putInt(bytes, offset, val); 1314 } 1315 1316 @Override 1317 short toShort(byte[] bytes, int offset, int length) { 1318 return UnsafeAccess.toShort(bytes, offset); 1319 } 1320 1321 @Override 1322 int putShort(byte[] bytes, int offset, short val) { 1323 return UnsafeAccess.putShort(bytes, offset, val); 1324 } 1325 } 1326 } 1327 1328 /** 1329 * Provides a lexicographical comparer implementation; either a Java implementation or a faster 1330 * implementation based on {@code Unsafe}. 1331 * <p> 1332 * Uses reflection to gracefully fall back to the Java implementation if {@code Unsafe} isn't 1333 * available. 1334 */ 1335 static class LexicographicalComparerHolder { 1336 static final String UNSAFE_COMPARER_NAME = 1337 LexicographicalComparerHolder.class.getName() + "$UnsafeComparer"; 1338 1339 static final Comparer<byte[]> BEST_COMPARER = getBestComparer(); 1340 1341 /** 1342 * Returns the Unsafe-using Comparer, or falls back to the pure-Java implementation if unable to 1343 * do so. 1344 */ 1345 static Comparer<byte[]> getBestComparer() { 1346 try { 1347 Class<?> theClass = Class.forName(UNSAFE_COMPARER_NAME); 1348 1349 // yes, UnsafeComparer does implement Comparer<byte[]> 1350 @SuppressWarnings("unchecked") 1351 Comparer<byte[]> comparer = (Comparer<byte[]>) theClass.getEnumConstants()[0]; 1352 return comparer; 1353 } catch (Throwable t) { // ensure we really catch *everything* 1354 return lexicographicalComparerJavaImpl(); 1355 } 1356 } 1357 1358 enum PureJavaComparer implements Comparer<byte[]> { 1359 INSTANCE; 1360 1361 @Override 1362 public int compareTo(byte[] buffer1, int offset1, int length1, byte[] buffer2, int offset2, 1363 int length2) { 1364 // Short circuit equal case 1365 if (buffer1 == buffer2 && offset1 == offset2 && length1 == length2) { 1366 return 0; 1367 } 1368 // Bring WritableComparator code local 1369 int end1 = offset1 + length1; 1370 int end2 = offset2 + length2; 1371 for (int i = offset1, j = offset2; i < end1 && j < end2; i++, j++) { 1372 int a = (buffer1[i] & 0xff); 1373 int b = (buffer2[j] & 0xff); 1374 if (a != b) { 1375 return a - b; 1376 } 1377 } 1378 return length1 - length2; 1379 } 1380 } 1381 1382 enum UnsafeComparer implements Comparer<byte[]> { 1383 INSTANCE; 1384 1385 static { 1386 if (!UNSAFE_UNALIGNED) { 1387 // It doesn't matter what we throw; 1388 // it's swallowed in getBestComparer(). 1389 throw new Error(); 1390 } 1391 1392 // sanity check - this should never fail 1393 if (HBasePlatformDependent.arrayIndexScale(byte[].class) != 1) { 1394 throw new AssertionError(); 1395 } 1396 } 1397 1398 /** 1399 * Lexicographically compare two arrays. 1400 * @param buffer1 left operand 1401 * @param buffer2 right operand 1402 * @param offset1 Where to start comparing in the left buffer 1403 * @param offset2 Where to start comparing in the right buffer 1404 * @param length1 How much to compare from the left buffer 1405 * @param length2 How much to compare from the right buffer 1406 * @return 0 if equal, < 0 if left is less than right, etc. 1407 */ 1408 @Override 1409 public int compareTo(byte[] buffer1, int offset1, int length1, byte[] buffer2, int offset2, 1410 int length2) { 1411 1412 // Short circuit equal case 1413 if (buffer1 == buffer2 && offset1 == offset2 && length1 == length2) { 1414 return 0; 1415 } 1416 final int stride = 8; 1417 final int minLength = Math.min(length1, length2); 1418 int strideLimit = minLength & ~(stride - 1); 1419 final long offset1Adj = offset1 + UnsafeAccess.BYTE_ARRAY_BASE_OFFSET; 1420 final long offset2Adj = offset2 + UnsafeAccess.BYTE_ARRAY_BASE_OFFSET; 1421 int i; 1422 1423 /* 1424 * Compare 8 bytes at a time. Benchmarking on x86 shows a stride of 8 bytes is no slower 1425 * than 4 bytes even on 32-bit. On the other hand, it is substantially faster on 64-bit. 1426 */ 1427 for (i = 0; i < strideLimit; i += stride) { 1428 long lw = HBasePlatformDependent.getLong(buffer1, offset1Adj + i); 1429 long rw = HBasePlatformDependent.getLong(buffer2, offset2Adj + i); 1430 if (lw != rw) { 1431 if (!UnsafeAccess.LITTLE_ENDIAN) { 1432 return ((lw + Long.MIN_VALUE) < (rw + Long.MIN_VALUE)) ? -1 : 1; 1433 } 1434 1435 /* 1436 * We want to compare only the first index where left[index] != right[index]. This 1437 * corresponds to the least significant nonzero byte in lw ^ rw, since lw and rw are 1438 * little-endian. Long.numberOfTrailingZeros(diff) tells us the least significant 1439 * nonzero bit, and zeroing out the first three bits of L.nTZ gives us the shift to get 1440 * that least significant nonzero byte. This comparison logic is based on UnsignedBytes 1441 * comparator from guava v21 1442 */ 1443 int n = Long.numberOfTrailingZeros(lw ^ rw) & ~0x7; 1444 return ((int) ((lw >>> n) & 0xFF)) - ((int) ((rw >>> n) & 0xFF)); 1445 } 1446 } 1447 1448 // The epilogue to cover the last (minLength % stride) elements. 1449 for (; i < minLength; i++) { 1450 int a = (buffer1[offset1 + i] & 0xFF); 1451 int b = (buffer2[offset2 + i] & 0xFF); 1452 if (a != b) { 1453 return a - b; 1454 } 1455 } 1456 return length1 - length2; 1457 } 1458 } 1459 } 1460 1461 static class CommonPrefixerHolder { 1462 static final String UNSAFE_COMMON_PREFIXER_NAME = 1463 CommonPrefixerHolder.class.getName() + "$UnsafeCommonPrefixer"; 1464 1465 static final CommonPrefixer BEST_COMMON_PREFIXER = getBestCommonPrefixer(); 1466 1467 static CommonPrefixer getBestCommonPrefixer() { 1468 try { 1469 Class<? extends CommonPrefixer> theClass = 1470 Class.forName(UNSAFE_COMMON_PREFIXER_NAME).asSubclass(CommonPrefixer.class); 1471 1472 return theClass.getConstructor().newInstance(); 1473 } catch (Throwable t) { // ensure we really catch *everything* 1474 return CommonPrefixerHolder.PureJavaCommonPrefixer.INSTANCE; 1475 } 1476 } 1477 1478 static final class PureJavaCommonPrefixer extends CommonPrefixer { 1479 static final PureJavaCommonPrefixer INSTANCE = new PureJavaCommonPrefixer(); 1480 1481 private PureJavaCommonPrefixer() { 1482 } 1483 1484 @Override 1485 public int findCommonPrefix(byte[] left, int leftOffset, int leftLength, byte[] right, 1486 int rightOffset, int rightLength) { 1487 int length = Math.min(leftLength, rightLength); 1488 int result = 0; 1489 1490 while (result < length && left[leftOffset + result] == right[rightOffset + result]) { 1491 result++; 1492 } 1493 return result; 1494 } 1495 } 1496 1497 static final class UnsafeCommonPrefixer extends CommonPrefixer { 1498 1499 static { 1500 if (!UNSAFE_UNALIGNED) { 1501 throw new Error(); 1502 } 1503 1504 // sanity check - this should never fail 1505 if (HBasePlatformDependent.arrayIndexScale(byte[].class) != 1) { 1506 throw new AssertionError(); 1507 } 1508 } 1509 1510 public UnsafeCommonPrefixer() { 1511 } 1512 1513 @Override 1514 public int findCommonPrefix(byte[] left, int leftOffset, int leftLength, byte[] right, 1515 int rightOffset, int rightLength) { 1516 final int stride = 8; 1517 final int minLength = Math.min(leftLength, rightLength); 1518 int strideLimit = minLength & ~(stride - 1); 1519 final long leftOffsetAdj = leftOffset + UnsafeAccess.BYTE_ARRAY_BASE_OFFSET; 1520 final long rightOffsetAdj = rightOffset + UnsafeAccess.BYTE_ARRAY_BASE_OFFSET; 1521 int result = 0; 1522 int i; 1523 1524 for (i = 0; i < strideLimit; i += stride) { 1525 long lw = HBasePlatformDependent.getLong(left, leftOffsetAdj + i); 1526 long rw = HBasePlatformDependent.getLong(right, rightOffsetAdj + i); 1527 if (lw != rw) { 1528 if (!UnsafeAccess.LITTLE_ENDIAN) { 1529 return result + (Long.numberOfLeadingZeros(lw ^ rw) / Bytes.SIZEOF_LONG); 1530 } else { 1531 return result + (Long.numberOfTrailingZeros(lw ^ rw) / Bytes.SIZEOF_LONG); 1532 } 1533 } else { 1534 result += Bytes.SIZEOF_LONG; 1535 } 1536 } 1537 1538 // The epilogue to cover the last (minLength % stride) elements. 1539 for (; i < minLength; i++) { 1540 int il = (left[leftOffset + i]); 1541 int ir = (right[rightOffset + i]); 1542 if (il != ir) { 1543 return result; 1544 } else { 1545 result++; 1546 } 1547 } 1548 1549 return result; 1550 } 1551 } 1552 } 1553 1554 /** 1555 * Lexicographically determine the equality of two arrays. 1556 * @param left left operand 1557 * @param right right operand 1558 * @return True if equal 1559 */ 1560 public static boolean equals(final byte[] left, final byte[] right) { 1561 // Could use Arrays.equals? 1562 // noinspection SimplifiableConditionalExpression 1563 if (left == right) return true; 1564 if (left == null || right == null) return false; 1565 if (left.length != right.length) return false; 1566 if (left.length == 0) return true; 1567 1568 // Since we're often comparing adjacent sorted data, 1569 // it's usual to have equal arrays except for the very last byte 1570 // so check that first 1571 if (left[left.length - 1] != right[right.length - 1]) return false; 1572 1573 return compareTo(left, right) == 0; 1574 } 1575 1576 /** 1577 * Lexicographically determine the equality of two arrays. 1578 * @param left left operand 1579 * @param leftOffset offset into left operand 1580 * @param leftLen length of left operand 1581 * @param right right operand 1582 * @param rightOffset offset into right operand 1583 * @param rightLen length of right operand 1584 * @return True if equal 1585 */ 1586 public static boolean equals(final byte[] left, int leftOffset, int leftLen, final byte[] right, 1587 int rightOffset, int rightLen) { 1588 // short circuit case 1589 if (left == right && leftOffset == rightOffset && leftLen == rightLen) { 1590 return true; 1591 } 1592 // different lengths fast check 1593 if (leftLen != rightLen) { 1594 return false; 1595 } 1596 if (leftLen == 0) { 1597 return true; 1598 } 1599 1600 // Since we're often comparing adjacent sorted data, 1601 // it's usual to have equal arrays except for the very last byte 1602 // so check that first 1603 if (left[leftOffset + leftLen - 1] != right[rightOffset + rightLen - 1]) return false; 1604 1605 return LexicographicalComparerHolder.BEST_COMPARER.compareTo(left, leftOffset, leftLen, right, 1606 rightOffset, rightLen) == 0; 1607 } 1608 1609 /** 1610 * Lexicographically determine the equality of two byte[], one as ByteBuffer. 1611 * @param a left operand 1612 * @param buf right operand 1613 * @return True if equal 1614 */ 1615 public static boolean equals(byte[] a, ByteBuffer buf) { 1616 if (a == null) return buf == null; 1617 if (buf == null) return false; 1618 if (a.length != buf.remaining()) return false; 1619 1620 // Thou shalt not modify the original byte buffer in what should be read only operations. 1621 ByteBuffer b = buf.duplicate(); 1622 for (byte anA : a) { 1623 if (anA != b.get()) { 1624 return false; 1625 } 1626 } 1627 return true; 1628 } 1629 1630 /** 1631 * Return true if the byte array on the right is a prefix of the byte array on the left. 1632 */ 1633 public static boolean startsWith(byte[] bytes, byte[] prefix) { 1634 return bytes != null && prefix != null && bytes.length >= prefix.length 1635 && LexicographicalComparerHolder.BEST_COMPARER.compareTo(bytes, 0, prefix.length, prefix, 0, 1636 prefix.length) == 0; 1637 } 1638 1639 /** 1640 * Calculate a hash code from a given byte array. 1641 * @param b bytes to hash 1642 * @return Runs {@link WritableComparator#hashBytes(byte[], int)} on the passed in array. This 1643 * method is what {@link org.apache.hadoop.io.Text} use calculating hash code. 1644 */ 1645 public static int hashCode(final byte[] b) { 1646 return hashCode(b, b.length); 1647 } 1648 1649 /** 1650 * Calculate a hash code from a given byte array. 1651 * @param b value 1652 * @param length length of the value 1653 * @return Runs {@link WritableComparator#hashBytes(byte[], int)} on the passed in array. This 1654 * method is what {@link org.apache.hadoop.io.Text} use calculating hash code. 1655 */ 1656 public static int hashCode(final byte[] b, final int length) { 1657 return WritableComparator.hashBytes(b, length); 1658 } 1659 1660 /** 1661 * Calculate a hash code from a given byte array suitable for use as a key in maps. 1662 * @param b bytes to hash 1663 * @return A hash of <code>b</code> as an Integer that can be used as key in Maps. 1664 */ 1665 public static Integer mapKey(final byte[] b) { 1666 return hashCode(b); 1667 } 1668 1669 /** 1670 * Calculate a hash code from a given byte array suitable for use as a key in maps. 1671 * @param b bytes to hash 1672 * @param length length to hash 1673 * @return A hash of <code>b</code> as an Integer that can be used as key in Maps. 1674 */ 1675 public static Integer mapKey(final byte[] b, final int length) { 1676 return hashCode(b, length); 1677 } 1678 1679 /** 1680 * Concatenate byte arrays. 1681 * @param a lower half 1682 * @param b upper half 1683 * @return New array that has a in lower half and b in upper half. 1684 */ 1685 public static byte[] add(final byte[] a, final byte[] b) { 1686 return add(a, b, EMPTY_BYTE_ARRAY); 1687 } 1688 1689 /** 1690 * Concatenate byte arrays. 1691 * @param a first third 1692 * @param b second third 1693 * @param c third third 1694 * @return New array made from a, b and c 1695 */ 1696 public static byte[] add(final byte[] a, final byte[] b, final byte[] c) { 1697 byte[] result = new byte[a.length + b.length + c.length]; 1698 System.arraycopy(a, 0, result, 0, a.length); 1699 System.arraycopy(b, 0, result, a.length, b.length); 1700 System.arraycopy(c, 0, result, a.length + b.length, c.length); 1701 return result; 1702 } 1703 1704 /** 1705 * Concatenate byte arrays. 1706 * @param arrays all the arrays to concatenate together. 1707 * @return New array made from the concatenation of the given arrays. 1708 */ 1709 public static byte[] add(final byte[][] arrays) { 1710 int length = 0; 1711 for (int i = 0; i < arrays.length; i++) { 1712 length += arrays[i].length; 1713 } 1714 byte[] result = new byte[length]; 1715 int index = 0; 1716 for (int i = 0; i < arrays.length; i++) { 1717 System.arraycopy(arrays[i], 0, result, index, arrays[i].length); 1718 index += arrays[i].length; 1719 } 1720 return result; 1721 } 1722 1723 /** 1724 * Make a new byte array from a subset of bytes at the head of another. 1725 * @param a array 1726 * @param length amount of bytes to grab 1727 * @return First <code>length</code> bytes from <code>a</code> 1728 */ 1729 public static byte[] head(final byte[] a, final int length) { 1730 if (a.length < length) { 1731 return null; 1732 } 1733 byte[] result = new byte[length]; 1734 System.arraycopy(a, 0, result, 0, length); 1735 return result; 1736 } 1737 1738 /** 1739 * Make a new byte array from a subset of bytes at the tail of another. 1740 * @param a array 1741 * @param length amount of bytes to snarf 1742 * @return Last <code>length</code> bytes from <code>a</code> 1743 */ 1744 public static byte[] tail(final byte[] a, final int length) { 1745 if (a.length < length) { 1746 return null; 1747 } 1748 byte[] result = new byte[length]; 1749 System.arraycopy(a, a.length - length, result, 0, length); 1750 return result; 1751 } 1752 1753 /** 1754 * Make a new byte array from a subset of bytes at the head of another, zero padded as desired. 1755 * @param a array 1756 * @param length new array size 1757 * @return Value in <code>a</code> plus <code>length</code> prepended 0 bytes 1758 */ 1759 public static byte[] padHead(final byte[] a, final int length) { 1760 byte[] padding = new byte[length]; 1761 for (int i = 0; i < length; i++) { 1762 padding[i] = 0; 1763 } 1764 return add(padding, a); 1765 } 1766 1767 /** 1768 * Make a new byte array from a subset of bytes at the tail of another, zero padded as desired. 1769 * @param a array 1770 * @param length new array size 1771 * @return Value in <code>a</code> plus <code>length</code> appended 0 bytes 1772 */ 1773 public static byte[] padTail(final byte[] a, final int length) { 1774 byte[] padding = new byte[length]; 1775 for (int i = 0; i < length; i++) { 1776 padding[i] = 0; 1777 } 1778 return add(a, padding); 1779 } 1780 1781 /** 1782 * Split passed range. Expensive operation relatively. Uses BigInteger math. Useful splitting 1783 * ranges for MapReduce jobs. 1784 * @param a Beginning of range 1785 * @param b End of range 1786 * @param num Number of times to split range. Pass 1 if you want to split the range in two; i.e. 1787 * one split. 1788 * @return Array of dividing values 1789 */ 1790 public static byte[][] split(final byte[] a, final byte[] b, final int num) { 1791 return split(a, b, false, num); 1792 } 1793 1794 /** 1795 * Split passed range. Expensive operation relatively. Uses BigInteger math. Useful splitting 1796 * ranges for MapReduce jobs. 1797 * @param a Beginning of range 1798 * @param b End of range 1799 * @param inclusive Whether the end of range is prefix-inclusive or is considered an exclusive 1800 * boundary. Automatic splits are generally exclusive and manual splits with an 1801 * explicit range utilize an inclusive end of range. 1802 * @param num Number of times to split range. Pass 1 if you want to split the range in two; 1803 * i.e. one split. 1804 * @return Array of dividing values 1805 */ 1806 public static byte[][] split(final byte[] a, final byte[] b, boolean inclusive, final int num) { 1807 byte[][] ret = new byte[num + 2][]; 1808 int i = 0; 1809 Iterable<byte[]> iter = iterateOnSplits(a, b, inclusive, num); 1810 if (iter == null) return null; 1811 for (byte[] elem : iter) { 1812 ret[i++] = elem; 1813 } 1814 return ret; 1815 } 1816 1817 /** 1818 * Iterate over keys within the passed range, splitting at an [a,b) boundary. 1819 */ 1820 public static Iterable<byte[]> iterateOnSplits(final byte[] a, final byte[] b, final int num) { 1821 return iterateOnSplits(a, b, false, num); 1822 } 1823 1824 /** 1825 * Iterate over keys within the passed range. 1826 */ 1827 public static Iterable<byte[]> iterateOnSplits(final byte[] a, final byte[] b, boolean inclusive, 1828 final int num) { 1829 byte[] aPadded; 1830 byte[] bPadded; 1831 if (a.length < b.length) { 1832 aPadded = padTail(a, b.length - a.length); 1833 bPadded = b; 1834 } else if (b.length < a.length) { 1835 aPadded = a; 1836 bPadded = padTail(b, a.length - b.length); 1837 } else { 1838 aPadded = a; 1839 bPadded = b; 1840 } 1841 if (compareTo(aPadded, bPadded) >= 0) { 1842 throw new IllegalArgumentException("b <= a"); 1843 } 1844 if (num <= 0) { 1845 throw new IllegalArgumentException("num cannot be <= 0"); 1846 } 1847 byte[] prependHeader = { 1, 0 }; 1848 final BigInteger startBI = new BigInteger(add(prependHeader, aPadded)); 1849 final BigInteger stopBI = new BigInteger(add(prependHeader, bPadded)); 1850 BigInteger diffBI = stopBI.subtract(startBI); 1851 if (inclusive) { 1852 diffBI = diffBI.add(BigInteger.ONE); 1853 } 1854 final BigInteger splitsBI = BigInteger.valueOf(num + 1); 1855 // when diffBI < splitBI, use an additional byte to increase diffBI 1856 if (diffBI.compareTo(splitsBI) < 0) { 1857 byte[] aPaddedAdditional = new byte[aPadded.length + 1]; 1858 byte[] bPaddedAdditional = new byte[bPadded.length + 1]; 1859 for (int i = 0; i < aPadded.length; i++) { 1860 aPaddedAdditional[i] = aPadded[i]; 1861 } 1862 for (int j = 0; j < bPadded.length; j++) { 1863 bPaddedAdditional[j] = bPadded[j]; 1864 } 1865 aPaddedAdditional[aPadded.length] = 0; 1866 bPaddedAdditional[bPadded.length] = 0; 1867 return iterateOnSplits(aPaddedAdditional, bPaddedAdditional, inclusive, num); 1868 } 1869 final BigInteger intervalBI; 1870 try { 1871 intervalBI = diffBI.divide(splitsBI); 1872 } catch (Exception e) { 1873 LOG.error("Exception caught during division", e); 1874 return null; 1875 } 1876 1877 final Iterator<byte[]> iterator = new Iterator<byte[]>() { 1878 private int i = -1; 1879 1880 @Override 1881 public boolean hasNext() { 1882 return i < num + 1; 1883 } 1884 1885 @Override 1886 public byte[] next() { 1887 i++; 1888 if (i == 0) return a; 1889 if (i == num + 1) return b; 1890 1891 BigInteger curBI = startBI.add(intervalBI.multiply(BigInteger.valueOf(i))); 1892 byte[] padded = curBI.toByteArray(); 1893 if (padded[1] == 0) padded = tail(padded, padded.length - 2); 1894 else padded = tail(padded, padded.length - 1); 1895 return padded; 1896 } 1897 1898 @Override 1899 public void remove() { 1900 throw new UnsupportedOperationException(); 1901 } 1902 1903 }; 1904 1905 return new Iterable<byte[]>() { 1906 @Override 1907 public Iterator<byte[]> iterator() { 1908 return iterator; 1909 } 1910 }; 1911 } 1912 1913 /** 1914 * Calculate the hash code for a given range of bytes. 1915 * @param bytes array to hash 1916 * @param offset offset to start from 1917 * @param length length to hash 1918 */ 1919 public static int hashCode(byte[] bytes, int offset, int length) { 1920 int hash = 1; 1921 for (int i = offset; i < offset + length; i++) 1922 hash = (31 * hash) + bytes[i]; 1923 return hash; 1924 } 1925 1926 /** 1927 * Create an array of byte[] given an array of String. 1928 * @param t operands 1929 * @return Array of byte arrays made from passed array of Text 1930 */ 1931 public static byte[][] toByteArrays(final String[] t) { 1932 byte[][] result = new byte[t.length][]; 1933 for (int i = 0; i < t.length; i++) { 1934 result[i] = Bytes.toBytes(t[i]); 1935 } 1936 return result; 1937 } 1938 1939 /** 1940 * Create an array of byte[] given an array of String. 1941 * @param t operands 1942 * @return Array of binary byte arrays made from passed array of binary strings 1943 */ 1944 public static byte[][] toBinaryByteArrays(final String[] t) { 1945 byte[][] result = new byte[t.length][]; 1946 for (int i = 0; i < t.length; i++) { 1947 result[i] = Bytes.toBytesBinary(t[i]); 1948 } 1949 return result; 1950 } 1951 1952 /** 1953 * Create a byte[][] where first and only entry is <code>column</code> 1954 * @param column operand 1955 * @return A byte array of a byte array where first and only entry is <code>column</code> 1956 */ 1957 public static byte[][] toByteArrays(final String column) { 1958 return toByteArrays(toBytes(column)); 1959 } 1960 1961 /** 1962 * Create a byte[][] where first and only entry is <code>column</code> 1963 * @param column operand 1964 * @return A byte array of a byte array where first and only entry is <code>column</code> 1965 */ 1966 public static byte[][] toByteArrays(final byte[] column) { 1967 byte[][] result = new byte[1][]; 1968 result[0] = column; 1969 return result; 1970 } 1971 1972 /** 1973 * Binary search for keys in indexes using Bytes.BYTES_RAWCOMPARATOR. 1974 * @param arr array of byte arrays to search for 1975 * @param key the key you want to find 1976 * @param offset the offset in the key you want to find 1977 * @param length the length of the key 1978 * @return zero-based index of the key, if the key is present in the array. Otherwise, a value -(i 1979 * + 1) such that the key is between arr[i - 1] and arr[i] non-inclusively, where i is in 1980 * [0, i], if we define arr[-1] = -Inf and arr[N] = Inf for an N-element array. The above 1981 * means that this function can return 2N + 1 different values ranging from -(N + 1) to N 1982 * - 1. 1983 */ 1984 public static int binarySearch(byte[][] arr, byte[] key, int offset, int length) { 1985 int low = 0; 1986 int high = arr.length - 1; 1987 1988 while (low <= high) { 1989 int mid = low + ((high - low) >> 1); 1990 // we have to compare in this order, because the comparator order 1991 // has special logic when the 'left side' is a special key. 1992 int cmp = 1993 Bytes.BYTES_RAWCOMPARATOR.compare(key, offset, length, arr[mid], 0, arr[mid].length); 1994 // key lives above the midpoint 1995 if (cmp > 0) low = mid + 1; 1996 // key lives below the midpoint 1997 else if (cmp < 0) high = mid - 1; 1998 // BAM. how often does this really happen? 1999 else return mid; 2000 } 2001 return -(low + 1); 2002 } 2003 2004 /** 2005 * Binary search for keys in indexes. 2006 * @param arr array of byte arrays to search for 2007 * @param key the key you want to find 2008 * @param comparator a comparator to compare. 2009 * @return zero-based index of the key, if the key is present in the array. Otherwise, a value -(i 2010 * + 1) such that the key is between arr[i - 1] and arr[i] non-inclusively, where i is in 2011 * [0, i], if we define arr[-1] = -Inf and arr[N] = Inf for an N-element array. The above 2012 * means that this function can return 2N + 1 different values ranging from -(N + 1) to N 2013 * - 1. 2014 * @return the index of the block 2015 */ 2016 public static int binarySearch(Cell[] arr, Cell key, CellComparator comparator) { 2017 int low = 0; 2018 int high = arr.length - 1; 2019 while (low <= high) { 2020 int mid = low + ((high - low) >> 1); 2021 // we have to compare in this order, because the comparator order 2022 // has special logic when the 'left side' is a special key. 2023 int cmp = comparator.compare(key, arr[mid]); 2024 // key lives above the midpoint 2025 if (cmp > 0) low = mid + 1; 2026 // key lives below the midpoint 2027 else if (cmp < 0) high = mid - 1; 2028 // BAM. how often does this really happen? 2029 else return mid; 2030 } 2031 return -(low + 1); 2032 } 2033 2034 /** 2035 * Bytewise binary increment/deincrement of long contained in byte array on given amount. 2036 * @param value - array of bytes containing long (length <= SIZEOF_LONG) 2037 * @param amount value will be incremented on (deincremented if negative) 2038 * @return array of bytes containing incremented long (length == SIZEOF_LONG) 2039 */ 2040 public static byte[] incrementBytes(byte[] value, long amount) { 2041 byte[] val = value; 2042 if (val.length < SIZEOF_LONG) { 2043 // Hopefully this doesn't happen too often. 2044 byte[] newvalue; 2045 if (val[0] < 0) { 2046 newvalue = new byte[] { -1, -1, -1, -1, -1, -1, -1, -1 }; 2047 } else { 2048 newvalue = new byte[SIZEOF_LONG]; 2049 } 2050 System.arraycopy(val, 0, newvalue, newvalue.length - val.length, val.length); 2051 val = newvalue; 2052 } else if (val.length > SIZEOF_LONG) { 2053 throw new IllegalArgumentException("Increment Bytes - value too big: " + val.length); 2054 } 2055 if (amount == 0) return val; 2056 if (val[0] < 0) { 2057 return binaryIncrementNeg(val, amount); 2058 } 2059 return binaryIncrementPos(val, amount); 2060 } 2061 2062 /* increment/deincrement for positive value */ 2063 private static byte[] binaryIncrementPos(byte[] value, long amount) { 2064 long amo = amount; 2065 int sign = 1; 2066 if (amount < 0) { 2067 amo = -amount; 2068 sign = -1; 2069 } 2070 for (int i = 0; i < value.length; i++) { 2071 int cur = ((int) amo % 256) * sign; 2072 amo = (amo >> 8); 2073 int val = value[value.length - i - 1] & 0x0ff; 2074 int total = val + cur; 2075 if (total > 255) { 2076 amo += sign; 2077 total %= 256; 2078 } else if (total < 0) { 2079 amo -= sign; 2080 } 2081 value[value.length - i - 1] = (byte) total; 2082 if (amo == 0) return value; 2083 } 2084 return value; 2085 } 2086 2087 /* increment/deincrement for negative value */ 2088 private static byte[] binaryIncrementNeg(byte[] value, long amount) { 2089 long amo = amount; 2090 int sign = 1; 2091 if (amount < 0) { 2092 amo = -amount; 2093 sign = -1; 2094 } 2095 for (int i = 0; i < value.length; i++) { 2096 int cur = ((int) amo % 256) * sign; 2097 amo = (amo >> 8); 2098 int val = (~value[value.length - i - 1] & 0x0ff) + 1; 2099 int total = cur - val; 2100 if (total >= 0) { 2101 amo += sign; 2102 } else if (total < -256) { 2103 amo -= sign; 2104 total %= 256; 2105 } 2106 value[value.length - i - 1] = (byte) total; 2107 if (amo == 0) return value; 2108 } 2109 return value; 2110 } 2111 2112 /** 2113 * Writes a string as a fixed-size field, padded with zeros. 2114 */ 2115 public static void writeStringFixedSize(final DataOutput out, String s, int size) 2116 throws IOException { 2117 byte[] b = toBytes(s); 2118 if (b.length > size) { 2119 throw new IOException("Trying to write " + b.length + " bytes (" + toStringBinary(b) 2120 + ") into a field of length " + size); 2121 } 2122 2123 out.writeBytes(s); 2124 for (int i = 0; i < size - s.length(); ++i) 2125 out.writeByte(0); 2126 } 2127 2128 /** 2129 * Reads a fixed-size field and interprets it as a string padded with zeros. 2130 */ 2131 public static String readStringFixedSize(final DataInput in, int size) throws IOException { 2132 byte[] b = new byte[size]; 2133 in.readFully(b); 2134 int n = b.length; 2135 while (n > 0 && b[n - 1] == 0) 2136 --n; 2137 2138 return toString(b, 0, n); 2139 } 2140 2141 /** 2142 * Copy the byte array given in parameter and return an instance of a new byte array with the same 2143 * length and the same content. 2144 * @param bytes the byte array to duplicate 2145 * @return a copy of the given byte array 2146 */ 2147 public static byte[] copy(byte[] bytes) { 2148 if (bytes == null) return null; 2149 byte[] result = new byte[bytes.length]; 2150 System.arraycopy(bytes, 0, result, 0, bytes.length); 2151 return result; 2152 } 2153 2154 /** 2155 * Copy the byte array given in parameter and return an instance of a new byte array with the same 2156 * length and the same content. 2157 * @param bytes the byte array to copy from 2158 * @return a copy of the given designated byte array 2159 */ 2160 public static byte[] copy(byte[] bytes, final int offset, final int length) { 2161 if (bytes == null) return null; 2162 byte[] result = new byte[length]; 2163 System.arraycopy(bytes, offset, result, 0, length); 2164 return result; 2165 } 2166 2167 /** 2168 * Search sorted array "a" for byte "key". I can't remember if I wrote this or copied it from 2169 * somewhere. (mcorgan) 2170 * @param a Array to search. Entries must be sorted and unique. 2171 * @param fromIndex First index inclusive of "a" to include in the search. 2172 * @param toIndex Last index exclusive of "a" to include in the search. 2173 * @param key The byte to search for. 2174 * @return The index of key if found. If not found, return -(index + 1), where negative indicates 2175 * "not found" and the "index + 1" handles the "-0" case. 2176 */ 2177 public static int unsignedBinarySearch(byte[] a, int fromIndex, int toIndex, byte key) { 2178 int unsignedKey = key & 0xff; 2179 int low = fromIndex; 2180 int high = toIndex - 1; 2181 2182 while (low <= high) { 2183 int mid = low + ((high - low) >> 1); 2184 int midVal = a[mid] & 0xff; 2185 2186 if (midVal < unsignedKey) { 2187 low = mid + 1; 2188 } else if (midVal > unsignedKey) { 2189 high = mid - 1; 2190 } else { 2191 return mid; // key found 2192 } 2193 } 2194 return -(low + 1); // key not found. 2195 } 2196 2197 /** 2198 * Treat the byte[] as an unsigned series of bytes, most significant bits first. Start by adding 1 2199 * to the rightmost bit/byte and carry over all overflows to the more significant bits/bytes. 2200 * @param input The byte[] to increment. 2201 * @return The incremented copy of "in". May be same length or 1 byte longer. 2202 */ 2203 public static byte[] unsignedCopyAndIncrement(final byte[] input) { 2204 byte[] copy = copy(input); 2205 if (copy == null) { 2206 throw new IllegalArgumentException("cannot increment null array"); 2207 } 2208 for (int i = copy.length - 1; i >= 0; --i) { 2209 if (copy[i] == -1) {// -1 is all 1-bits, which is the unsigned maximum 2210 copy[i] = 0; 2211 } else { 2212 ++copy[i]; 2213 return copy; 2214 } 2215 } 2216 // we maxed out the array 2217 byte[] out = new byte[copy.length + 1]; 2218 out[0] = 1; 2219 System.arraycopy(copy, 0, out, 1, copy.length); 2220 return out; 2221 } 2222 2223 public static boolean equals(List<byte[]> a, List<byte[]> b) { 2224 if (a == null) { 2225 if (b == null) { 2226 return true; 2227 } 2228 return false; 2229 } 2230 if (b == null) { 2231 return false; 2232 } 2233 if (a.size() != b.size()) { 2234 return false; 2235 } 2236 for (int i = 0; i < a.size(); ++i) { 2237 if (!Bytes.equals(a.get(i), b.get(i))) { 2238 return false; 2239 } 2240 } 2241 return true; 2242 } 2243 2244 public static boolean isSorted(Collection<byte[]> arrays) { 2245 if (!CollectionUtils.isEmpty(arrays)) { 2246 byte[] previous = new byte[0]; 2247 for (byte[] array : arrays) { 2248 if (Bytes.compareTo(previous, array) > 0) { 2249 return false; 2250 } 2251 previous = array; 2252 } 2253 } 2254 return true; 2255 } 2256 2257 public static List<byte[]> getUtf8ByteArrays(List<String> strings) { 2258 if (CollectionUtils.isEmpty(strings)) { 2259 return Collections.emptyList(); 2260 } 2261 List<byte[]> byteArrays = new ArrayList<>(strings.size()); 2262 strings.forEach(s -> byteArrays.add(Bytes.toBytes(s))); 2263 return byteArrays; 2264 } 2265 2266 /** 2267 * Returns the index of the first appearance of the value {@code target} in {@code array}. 2268 * @param array an array of {@code byte} values, possibly empty 2269 * @param target a primitive {@code byte} value 2270 * @return the least index {@code i} for which {@code array[i] == target}, or {@code -1} if no 2271 * such index exists. 2272 */ 2273 public static int indexOf(byte[] array, byte target) { 2274 for (int i = 0; i < array.length; i++) { 2275 if (array[i] == target) { 2276 return i; 2277 } 2278 } 2279 return -1; 2280 } 2281 2282 /** 2283 * Returns the start position of the first occurrence of the specified {@code 2284 * target} within {@code array}, or {@code -1} if there is no such occurrence. 2285 * <p> 2286 * More formally, returns the lowest index {@code i} such that {@code 2287 * java.util.Arrays.copyOfRange(array, i, i + target.length)} contains exactly the same elements 2288 * as {@code target}. 2289 * @param array the array to search for the sequence {@code target} 2290 * @param target the array to search for as a sub-sequence of {@code array} 2291 */ 2292 public static int indexOf(byte[] array, byte[] target) { 2293 checkNotNull(array, "array"); 2294 checkNotNull(target, "target"); 2295 if (target.length == 0) { 2296 return 0; 2297 } 2298 2299 outer: for (int i = 0; i < array.length - target.length + 1; i++) { 2300 for (int j = 0; j < target.length; j++) { 2301 if (array[i + j] != target[j]) { 2302 continue outer; 2303 } 2304 } 2305 return i; 2306 } 2307 return -1; 2308 } 2309 2310 /** 2311 * Return true if target is present as an element anywhere in the given array. 2312 * @param array an array of {@code byte} values, possibly empty 2313 * @param target a primitive {@code byte} value 2314 * @return {@code true} if {@code target} is present as an element anywhere in {@code array}. 2315 */ 2316 public static boolean contains(byte[] array, byte target) { 2317 return indexOf(array, target) > -1; 2318 } 2319 2320 /** 2321 * Return true if target is present as an element anywhere in the given array. 2322 * @param array an array of {@code byte} values, possibly empty 2323 * @param target an array of {@code byte} 2324 * @return {@code true} if {@code target} is present anywhere in {@code array} 2325 */ 2326 public static boolean contains(byte[] array, byte[] target) { 2327 return indexOf(array, target) > -1; 2328 } 2329 2330 /** 2331 * Fill given array with zeros. 2332 * @param b array which needs to be filled with zeros 2333 */ 2334 public static void zero(byte[] b) { 2335 zero(b, 0, b.length); 2336 } 2337 2338 /** 2339 * Fill given array with zeros at the specified position. 2340 */ 2341 public static void zero(byte[] b, int offset, int length) { 2342 checkPositionIndex(offset, b.length, "offset"); 2343 checkArgument(length > 0, "length must be greater than 0"); 2344 checkPositionIndex(offset + length, b.length, "offset + length"); 2345 Arrays.fill(b, offset, offset + length, (byte) 0); 2346 } 2347 2348 // Pseudorandom random number generator, do not use SecureRandom here 2349 private static final Random RNG = new Random(); 2350 2351 /** 2352 * Fill given array with random bytes. 2353 * @param b array which needs to be filled with random bytes 2354 * <p> 2355 * If you want random bytes generated by a strong source of randomness use 2356 * {@link Bytes#secureRandom(byte[])}. 2357 * @param b array which needs to be filled with random bytes 2358 */ 2359 public static void random(byte[] b) { 2360 RNG.nextBytes(b); 2361 } 2362 2363 /** 2364 * Fill given array with random bytes at the specified position. 2365 * <p> 2366 * If you want random bytes generated by a strong source of randomness use 2367 * {@link Bytes#secureRandom(byte[], int, int)}. 2368 * @param b array which needs to be filled with random bytes 2369 * @param offset staring offset in array 2370 * @param length number of bytes to fill 2371 */ 2372 public static void random(byte[] b, int offset, int length) { 2373 checkPositionIndex(offset, b.length, "offset"); 2374 checkArgument(length > 0, "length must be greater than 0"); 2375 checkPositionIndex(offset + length, b.length, "offset + length"); 2376 byte[] buf = new byte[length]; 2377 RNG.nextBytes(buf); 2378 System.arraycopy(buf, 0, b, offset, length); 2379 } 2380 2381 // Bytes.secureRandom may be used to create key material. 2382 private static final SecureRandom SECURE_RNG = new SecureRandom(); 2383 2384 /** 2385 * Fill given array with random bytes using a strong random number generator. 2386 * @param b array which needs to be filled with random bytes 2387 */ 2388 public static void secureRandom(byte[] b) { 2389 SECURE_RNG.nextBytes(b); 2390 } 2391 2392 /** 2393 * Fill given array with random bytes at the specified position using a strong random number 2394 * generator. 2395 * @param b array which needs to be filled with random bytes 2396 * @param offset staring offset in array 2397 * @param length number of bytes to fill 2398 */ 2399 public static void secureRandom(byte[] b, int offset, int length) { 2400 checkPositionIndex(offset, b.length, "offset"); 2401 checkArgument(length > 0, "length must be greater than 0"); 2402 checkPositionIndex(offset + length, b.length, "offset + length"); 2403 byte[] buf = new byte[length]; 2404 SECURE_RNG.nextBytes(buf); 2405 System.arraycopy(buf, 0, b, offset, length); 2406 } 2407 2408 /** 2409 * Create a max byte array with the specified max byte count 2410 * @param maxByteCount the length of returned byte array 2411 * @return the created max byte array 2412 */ 2413 public static byte[] createMaxByteArray(int maxByteCount) { 2414 byte[] maxByteArray = new byte[maxByteCount]; 2415 for (int i = 0; i < maxByteArray.length; i++) { 2416 maxByteArray[i] = (byte) 0xff; 2417 } 2418 return maxByteArray; 2419 } 2420 2421 /** 2422 * Create a byte array which is multiple given bytes 2423 * @return byte array 2424 */ 2425 public static byte[] multiple(byte[] srcBytes, int multiNum) { 2426 if (multiNum <= 0) { 2427 return new byte[0]; 2428 } 2429 byte[] result = new byte[srcBytes.length * multiNum]; 2430 for (int i = 0; i < multiNum; i++) { 2431 System.arraycopy(srcBytes, 0, result, i * srcBytes.length, srcBytes.length); 2432 } 2433 return result; 2434 } 2435 2436 private static final char[] HEX_CHARS = 2437 { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'a', 'b', 'c', 'd', 'e', 'f' }; 2438 2439 /** 2440 * Convert a byte range into a hex string 2441 */ 2442 public static String toHex(byte[] b, int offset, int length) { 2443 checkArgument(length <= Integer.MAX_VALUE / 2); 2444 int numChars = length * 2; 2445 char[] ch = new char[numChars]; 2446 for (int i = 0; i < numChars; i += 2) { 2447 byte d = b[offset + i / 2]; 2448 ch[i] = HEX_CHARS[(d >> 4) & 0x0F]; 2449 ch[i + 1] = HEX_CHARS[d & 0x0F]; 2450 } 2451 return new String(ch); 2452 } 2453 2454 /** 2455 * Convert a byte array into a hex string 2456 */ 2457 public static String toHex(byte[] b) { 2458 return toHex(b, 0, b.length); 2459 } 2460 2461 private static int hexCharToNibble(char ch) { 2462 if (ch <= '9' && ch >= '0') { 2463 return ch - '0'; 2464 } else if (ch >= 'a' && ch <= 'f') { 2465 return ch - 'a' + 10; 2466 } else if (ch >= 'A' && ch <= 'F') { 2467 return ch - 'A' + 10; 2468 } 2469 throw new IllegalArgumentException("Invalid hex char: " + ch); 2470 } 2471 2472 private static byte hexCharsToByte(char c1, char c2) { 2473 return (byte) ((hexCharToNibble(c1) << 4) | hexCharToNibble(c2)); 2474 } 2475 2476 /** 2477 * Create a byte array from a string of hash digits. The length of the string must be a multiple 2478 * of 2 2479 */ 2480 public static byte[] fromHex(String hex) { 2481 checkArgument(hex.length() % 2 == 0, "length must be a multiple of 2"); 2482 int len = hex.length(); 2483 byte[] b = new byte[len / 2]; 2484 for (int i = 0; i < len; i += 2) { 2485 b[i / 2] = hexCharsToByte(hex.charAt(i), hex.charAt(i + 1)); 2486 } 2487 return b; 2488 } 2489 2490 /** 2491 * Find index of passed delimiter. 2492 * @return Index of delimiter having started from start of <code>b</code> moving rightward. 2493 */ 2494 public static int searchDelimiterIndex(final byte[] b, int offset, final int length, 2495 final int delimiter) { 2496 if (b == null) { 2497 throw new IllegalArgumentException("Passed buffer is null"); 2498 } 2499 int result = -1; 2500 for (int i = offset; i < length + offset; i++) { 2501 if (b[i] == delimiter) { 2502 result = i; 2503 break; 2504 } 2505 } 2506 return result; 2507 } 2508 2509 /** 2510 * Find index of passed delimiter walking from end of buffer backwards. 2511 * @return Index of delimiter 2512 */ 2513 public static int searchDelimiterIndexInReverse(final byte[] b, final int offset, 2514 final int length, final int delimiter) { 2515 if (b == null) { 2516 throw new IllegalArgumentException("Passed buffer is null"); 2517 } 2518 int result = -1; 2519 for (int i = (offset + length) - 1; i >= offset; i--) { 2520 if (b[i] == delimiter) { 2521 result = i; 2522 break; 2523 } 2524 } 2525 return result; 2526 } 2527 2528 public static int findCommonPrefix(byte[] left, byte[] right, int leftLength, int rightLength, 2529 int leftOffset, int rightOffset) { 2530 return CommonPrefixerHolder.BEST_COMMON_PREFIXER.findCommonPrefix(left, leftOffset, leftLength, 2531 right, rightOffset, rightLength); 2532 } 2533}