001/*
002 * Licensed to the Apache Software Foundation (ASF) under one or more
003 * contributor license agreements.  See the NOTICE file distributed with
004 * this work for additional information regarding copyright ownership.
005 * The ASF licenses this file to You under the Apache License, Version 2.0
006 * (the "License"); you may not use this file except in compliance with
007 * the License.  You may obtain a copy of the License at
008 *
009 *      http://www.apache.org/licenses/LICENSE-2.0
010 *
011 * Unless required by applicable law or agreed to in writing, software
012 * distributed under the License is distributed on an "AS IS" BASIS,
013 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
014 * See the License for the specific language governing permissions and
015 * limitations under the License.
016 */
017
018package org.apache.commons.codec.digest;
019
020import static java.lang.Integer.rotateLeft;
021
022import java.util.zip.Checksum;
023
024/**
025 * Implements the xxHash32 hash algorithm.
026 *
027 * <p>
028 * Copied from Commons Compress 1.14 <a href=
029 * "https://gitbox.apache.org/repos/asf?p=commons-compress.git;a=blob;f=src/main/java/org/apache/commons/compress/compressors/lz4/XXHash32.java;h=a406ffc197449be594d46f0d2712b2d4786a1e68;hb=HEAD">https://gitbox.apache.org/repos/asf?p=commons-compress.git;a=blob;f=src/main/java/org/apache/commons/compress/compressors/lz4/XXHash32.java;h=a406ffc197449be594d46f0d2712b2d4786a1e68;hb=HEAD</a>
030 * </p>
031 * <p>
032 * NotThreadSafe
033 * </p>
034 *
035 * @see <a href="https://cyan4973.github.io/xxHash/">xxHash</a>
036 * @since 1.11
037 */
038public class XXHash32 implements Checksum {
039
040    private static final int BUF_SIZE = 16;
041    private static final int ROTATE_BITS = 13;
042
043    private static final int PRIME1 = (int) 2654435761L;
044    private static final int PRIME2 = (int) 2246822519L;
045    private static final int PRIME3 = (int) 3266489917L;
046    private static final int PRIME4 =  668265263;
047    private static final int PRIME5 =  374761393;
048
049    /**
050     * Gets the little-endian int from 4 bytes starting at the specified index.
051     *
052     * @param buffer The data
053     * @param idx The index
054     * @return The little-endian int
055     */
056    private static int getInt(final byte[] buffer, final int idx) {
057        return buffer[idx    ] & 0xff |
058               (buffer[idx + 1] & 0xff) <<  8 |
059               (buffer[idx + 2] & 0xff) << 16 |
060               (buffer[idx + 3] & 0xff) << 24;
061    }
062    private final byte[] oneByte = new byte[1];
063    private final int[] state = new int[4];
064    // Note: the code used to use ByteBuffer but the manual method is 50% faster
065    // See: https://gitbox.apache.org/repos/asf/commons-compress/diff/2f56fb5c
066    private final byte[] buffer = new byte[BUF_SIZE];
067
068    private final int seed;
069    private int totalLen;
070
071    private int pos;
072
073    /** Sets to true when the state array has been updated since the last reset. */
074    private boolean stateUpdated;
075
076    /**
077     * Creates an XXHash32 instance with a seed of 0.
078     */
079    public XXHash32() {
080        this(0);
081    }
082
083    /**
084     * Creates an XXHash32 instance.
085     * @param seed the seed to use
086     */
087    public XXHash32(final int seed) {
088        this.seed = seed;
089        initializeState();
090    }
091
092    @Override
093    public long getValue() {
094        int hash;
095        if (stateUpdated) {
096            // Hash with the state
097            hash =
098                rotateLeft(state[0],  1) +
099                rotateLeft(state[1],  7) +
100                rotateLeft(state[2], 12) +
101                rotateLeft(state[3], 18);
102        } else {
103            // Hash using the original seed from position 2
104            hash = state[2] + PRIME5;
105        }
106        hash += totalLen;
107
108        int idx = 0;
109        final int limit = pos - 4;
110        for (; idx <= limit; idx += 4) {
111            hash = rotateLeft(hash + getInt(buffer, idx) * PRIME3, 17) * PRIME4;
112        }
113        while (idx < pos) {
114            hash = rotateLeft(hash + (buffer[idx++] & 0xff) * PRIME5, 11) * PRIME1;
115        }
116
117        hash ^= hash >>> 15;
118        hash *= PRIME2;
119        hash ^= hash >>> 13;
120        hash *= PRIME3;
121        hash ^= hash >>> 16;
122        return hash & 0xffffffffL;
123    }
124
125    private void initializeState() {
126        state[0] = seed + PRIME1 + PRIME2;
127        state[1] = seed + PRIME2;
128        state[2] = seed;
129        state[3] = seed - PRIME1;
130    }
131
132    private void process(final byte[] b, final int offset) {
133        // local shadows for performance
134        int s0 = state[0];
135        int s1 = state[1];
136        int s2 = state[2];
137        int s3 = state[3];
138
139        s0 = rotateLeft(s0 + getInt(b, offset) * PRIME2, ROTATE_BITS) * PRIME1;
140        s1 = rotateLeft(s1 + getInt(b, offset + 4) * PRIME2, ROTATE_BITS) * PRIME1;
141        s2 = rotateLeft(s2 + getInt(b, offset + 8) * PRIME2, ROTATE_BITS) * PRIME1;
142        s3 = rotateLeft(s3 + getInt(b, offset + 12) * PRIME2, ROTATE_BITS) * PRIME1;
143
144        state[0] = s0;
145        state[1] = s1;
146        state[2] = s2;
147        state[3] = s3;
148
149        stateUpdated = true;
150    }
151
152    @Override
153    public void reset() {
154        initializeState();
155        totalLen = 0;
156        pos = 0;
157        stateUpdated = false;
158    }
159
160    @Override
161    public void update(final byte[] b, int off, final int len) {
162        if (len <= 0) {
163            return;
164        }
165        totalLen += len;
166
167        final int end = off + len;
168
169        // Check if the unprocessed bytes and new bytes can fill a block of 16.
170        // Make this overflow safe in the event that len is Integer.MAX_VALUE.
171        // Equivalent to: (pos + len < BUF_SIZE)
172        if (pos + len - BUF_SIZE < 0) {
173            System.arraycopy(b, off, buffer, pos, len);
174            pos += len;
175            return;
176        }
177
178        // Process left-over bytes with new bytes
179        if (pos > 0) {
180            final int size = BUF_SIZE - pos;
181            System.arraycopy(b, off, buffer, pos, size);
182            process(buffer, 0);
183            off += size;
184        }
185
186        final int limit = end - BUF_SIZE;
187        while (off <= limit) {
188            process(b, off);
189            off += BUF_SIZE;
190        }
191
192        // Handle left-over bytes
193        if (off < end) {
194            pos = end - off;
195            System.arraycopy(b, off, buffer, 0, pos);
196        } else {
197            pos = 0;
198        }
199    }
200
201    @Override
202    public void update(final int b) {
203        oneByte[0] = (byte) (b & 0xff);
204        update(oneByte, 0, 1);
205    }
206}