io.nosqlbench.virtdata.api.murmur.Murmur3F Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of virtdata-lib-curves4 Show documentation
Show all versions of virtdata-lib-curves4 Show documentation
Statistical sampling library for use in virtdata libraries, based
on apache commons math 4
/*
* Copyright (C) 2014-2016 Markus Junginger, greenrobot (http://greenrobot.org)
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package io.nosqlbench.virtdata.api.murmur;
import java.math.BigInteger;
/** Murmur3F (MurmurHash3_x64_128) */
public class Murmur3F implements Checksum128 {
private static final long C1 = 0x87c37b91114253d5L;
private static final long C2 = 0x4cf5ad432745937fL;
private final long seed;
private long h1;
private long h2;
private int length;
private int partialPos;
private long partialK1;
private long partialK2;
private boolean finished;
private long finishedH1;
private long finishedH2;
/**
* This constructor allows you to require that an unsafe implementation of
* primitive array operations is used, for added speed on platforms that you
* know can support it. This allows callers to have an easy way to
* exclusively opt in or out of unsafe behavior at a class-loader level,
* rather than forcing an unsafe behavior before the caller gets a chance
* to intervene. The choice is either/or, not "optional, but fall back if not
* supported." Callers can instrument for this with exception handling if it
* is needed. Effectively, setting the unsafe value chooses an implementation.
*
* @param seed A seed to initialize this hash with, or the input when using it
* as a function.
* @param unsafe Whether to require that the implementation relies on an unsafe
* calls.
*/
public Murmur3F(int seed, boolean unsafe) {
this.seed = seed & 0xffffffffL;
h1 = h2 = this.seed;
}
public Murmur3F() {
this(0,false);
}
public Murmur3F(int seed) {
this(seed,false);
}
@Override
public void update(int b) {
finished = false;
switch (partialPos) {
case 0:
partialK1 = 0xff & b;
break;
case 1:
partialK1 |= (0xff & b) << 8;
break;
case 2:
partialK1 |= (0xff & b) << 16;
break;
case 3:
partialK1 |= (0xffL & b) << 24;
break;
case 4:
partialK1 |= (0xffL & b) << 32;
break;
case 5:
partialK1 |= (0xffL & b) << 40;
break;
case 6:
partialK1 |= (0xffL & b) << 48;
break;
case 7:
partialK1 |= (0xffL & b) << 56;
break;
case 8:
partialK2 = 0xff & b;
break;
case 9:
partialK2 |= (0xff & b) << 8;
break;
case 10:
partialK2 |= (0xff & b) << 16;
break;
case 11:
partialK2 |= (0xffL & b) << 24;
break;
case 12:
partialK2 |= (0xffL & b) << 32;
break;
case 13:
partialK2 |= (0xffL & b) << 40;
break;
case 14:
partialK2 |= (0xffL & b) << 48;
break;
case 15:
partialK2 |= (0xffL & b) << 56;
break;
}
partialPos++;
if (partialPos == 16) {
applyKs(partialK1, partialK2);
partialPos = 0;
}
length++;
}
/**
* Special update method to hash long values very efficiently using Java's native little endian (LE) byte order.
* Note, that you cannot mix this with other (previous) hash updates, because it only supports 8-bytes alignment.
* @param value The long value to update
*/
public void updateLongLE(long value) {
finished = false;
switch (partialPos) {
case 0:
partialK1 = value;
break;
case 8:
partialK2 = value;
break;
default:
throw new IllegalStateException("Cannot mix long with other alignments than 8: " + partialPos);
}
partialPos += 8;
if (partialPos == 16) {
applyKs(partialK1, partialK2);
partialPos = 0;
}
length += 8;
}
/**
* Consider {@link #updateLongLE(long)} for better performance if you do not rely on big endian (BE) byte order.
* @param value value to update
*/
public void updateLongBE(long value) {
updateLongLE(Long.reverseBytes(value));
}
public void update(byte[] b) {
update(b, 0, b.length);
}
@Override
public void update(byte[] b, int off, int len) {
finished = false;
while (partialPos != 0 && len > 0) {
update(b[off]);
off++;
len--;
}
int remainder = len & 0xF;
int stop = off + len - remainder;
for (int i = off; i < stop; i += 16) {
long k1 = getLongLE(b, i);
long k2 = getLongLE(b, i + 8);
applyKs(k1, k2);
}
length += stop - off;
for (int i = 0; i < remainder; i++) {
update(b[stop + i]);
}
}
private void applyKs(long k1, long k2) {
k1 *= C1;
k1 = Long.rotateLeft(k1, 31);
k1 *= C2;
h1 ^= k1;
h1 = Long.rotateLeft(h1, 27);
h1 += h2;
h1 = h1 * 5 + 0x52dce729;
k2 *= C2;
k2 = Long.rotateLeft(k2, 33);
k2 *= C1;
h2 ^= k2;
h2 = Long.rotateLeft(h2, 31);
h2 += h1;
h2 = h2 * 5 + 0x38495ab5;
}
private void checkFinished() {
if (!finished) {
finished = true;
finishedH1 = h1;
finishedH2 = h2;
if (partialPos > 0) {
if (partialPos > 8) {
long k2 = partialK2 * C2;
k2 = Long.rotateLeft(k2, 33);
k2 *= C1;
finishedH2 ^= k2;
}
long k1 = partialK1 * C1;
k1 = Long.rotateLeft(k1, 31);
k1 *= C2;
finishedH1 ^= k1;
}
finishedH1 ^= length;
finishedH2 ^= length;
finishedH1 += finishedH2;
finishedH2 += finishedH1;
finishedH1 = fmix64(finishedH1);
finishedH2 = fmix64(finishedH2);
finishedH1 += finishedH2;
finishedH2 += finishedH1;
}
}
private long fmix64(long k) {
k ^= k >>> 33;
k *= 0xff51afd7ed558ccdL;
k ^= k >>> 33;
k *= 0xc4ceb9fe1a85ec53L;
k ^= k >>> 33;
return k;
}
@Override
/** Returns the lower 64 bits of the 128 bit hash (you can use just this value this as a 64 bit hash). */
public long getValue() {
checkFinished();
return finishedH1;
}
/** Returns the higher 64 bits of the 128 bit hash. */
public long getValueHigh() {
checkFinished();
return finishedH2;
}
/** Positive value. */
public BigInteger getValueBigInteger() {
byte[] bytes = getValueBytesBigEndian();
return new BigInteger(1, bytes);
}
/** Padded with leading 0s to ensure length of 32. */
public String getValueHexString() {
checkFinished();
return getPaddedHexString(finishedH2) + getPaddedHexString(finishedH1);
}
private String getPaddedHexString(long value) {
String string = Long.toHexString(value);
while (string.length() < 16) {
string = '0' + string;
}
return string;
}
public byte[] getValueBytesBigEndian() {
checkFinished();
byte[] bytes = new byte[16];
for (int i = 0; i < 8; i++) {
bytes[i] = (byte) ((finishedH2 >>> (56 - i * 8)) & 0xff);
}
for (int i = 0; i < 8; i++) {
bytes[8 + i] = (byte) ((finishedH1 >>> (56 - i * 8)) & 0xff);
}
return bytes;
}
public byte[] getValueBytesLittleEndian() {
checkFinished();
byte[] bytes = new byte[16];
for (int i = 0; i < 8; i++) {
bytes[i] = (byte) ((finishedH1 >>> (i * 8)) & 0xff);
}
for (int i = 0; i < 8; i++) {
bytes[8 + i] = (byte) ((finishedH2 >>> (i * 8)) & 0xff);
}
return bytes;
}
@Override
public void reset() {
h1 = h2 = seed;
length = 0;
partialPos = 0;
finished = false;
// The remainder is not really necessary, but looks nicer when debugging
partialK1 = partialK2 = 0;
finishedH1 = finishedH2 = 0;
}
private static long getLongBE(byte[] bytes, int index) {
return (bytes[index + 7] & 0xff) | ((bytes[index + 6] & 0xff) << 8) |
((bytes[index + 5] & 0xff) << 16) | ((bytes[index + 4] & 0xffL) << 24) |
((bytes[index + 3] & 0xffL) << 32) | ((bytes[index + 2] & 0xffL) << 40) |
((bytes[index + 1] & 0xffL) << 48) | (((long) bytes[index]) << 56);
}
private static long getLongLE(byte[] bytes, int index) {
return (bytes[index] & 0xff) | ((bytes[index + 1] & 0xff) << 8) |
((bytes[index + 2] & 0xff) << 16) | ((bytes[index + 3] & 0xffL) << 24) |
((bytes[index + 4] & 0xffL) << 32) | ((bytes[index + 5] & 0xffL) << 40) |
((bytes[index + 6] & 0xffL) << 48) | (((long) bytes[index + 7]) << 56);
}
}