All Downloads are FREE. Search and download functionalities are using the official Maven repository.

jersey.repackaged.com.google.common.hash.AbstractStreamingHashFunction Maven / Gradle / Ivy

There is a newer version: 2.22.2
Show newest version
/*
 * Copyright (C) 2011 The Guava Authors
 *
 * 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 jersey.repackaged.com.google.common.hash;

import static jersey.repackaged.com.google.common.base.Preconditions.checkArgument;

import jersey.repackaged.com.google.common.base.Preconditions;

import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.charset.Charset;

/**
 * Skeleton implementation of {@link HashFunction}. Provides default implementations which
 * invokes the appropriate method on {@link #newHasher()}, then return the result of
 * {@link Hasher#hash}.
 *
 * 

Invocations of {@link #newHasher(int)} also delegate to {@linkplain #newHasher()}, ignoring * the expected input size parameter. * * @author Kevin Bourrillion */ abstract class AbstractStreamingHashFunction implements HashFunction { @Override public HashCode hashObject(T instance, Funnel funnel) { return newHasher().putObject(instance, funnel).hash(); } @Override public HashCode hashString(CharSequence input) { return newHasher().putString(input).hash(); } @Override public HashCode hashString(CharSequence input, Charset charset) { return newHasher().putString(input, charset).hash(); } @Override public HashCode hashInt(int input) { return newHasher().putInt(input).hash(); } @Override public HashCode hashLong(long input) { return newHasher().putLong(input).hash(); } @Override public HashCode hashBytes(byte[] input) { return newHasher().putBytes(input).hash(); } @Override public HashCode hashBytes(byte[] input, int off, int len) { return newHasher().putBytes(input, off, len).hash(); } @Override public Hasher newHasher(int expectedInputSize) { Preconditions.checkArgument(expectedInputSize >= 0); return newHasher(); } /** * A convenience base class for implementors of {@code Hasher}; handles accumulating data * until an entire "chunk" (of implementation-dependent length) is ready to be hashed. * * @author Kevin Bourrillion * @author Dimitris Andreou */ // TODO(kevinb): this class still needs some design-and-document-for-inheritance love protected static abstract class AbstractStreamingHasher extends AbstractHasher { /** Buffer via which we pass data to the hash algorithm (the implementor) */ private final ByteBuffer buffer; /** Number of bytes to be filled before process() invocation(s). */ private final int bufferSize; /** Number of bytes processed per process() invocation. */ private final int chunkSize; /** * Constructor for use by subclasses. This hasher instance will process chunks of the specified * size. * * @param chunkSize the number of bytes available per {@link #process(ByteBuffer)} invocation; * must be at least 4 */ protected AbstractStreamingHasher(int chunkSize) { this(chunkSize, chunkSize); } /** * Constructor for use by subclasses. This hasher instance will process chunks of the specified * size, using an internal buffer of {@code bufferSize} size, which must be a multiple of * {@code chunkSize}. * * @param chunkSize the number of bytes available per {@link #process(ByteBuffer)} invocation; * must be at least 4 * @param bufferSize the size of the internal buffer. Must be a multiple of chunkSize */ protected AbstractStreamingHasher(int chunkSize, int bufferSize) { // TODO(kevinb): check more preconditions (as bufferSize >= chunkSize) if this is ever public checkArgument(bufferSize % chunkSize == 0); // TODO(user): benchmark performance difference with longer buffer this.buffer = ByteBuffer .allocate(bufferSize + 7) // always space for a single primitive .order(ByteOrder.LITTLE_ENDIAN); this.bufferSize = bufferSize; this.chunkSize = chunkSize; } /** * Processes the available bytes of the buffer (at most {@code chunk} bytes). */ protected abstract void process(ByteBuffer bb); /** * This is invoked for the last bytes of the input, which are not enough to * fill a whole chunk. The passed {@code ByteBuffer} is guaranteed to be * non-empty. * *

This implementation simply pads with zeros and delegates to * {@link #process(ByteBuffer)}. */ protected void processRemaining(ByteBuffer bb) { bb.position(bb.limit()); // move at the end bb.limit(chunkSize + 7); // get ready to pad with longs while (bb.position() < chunkSize) { bb.putLong(0); } bb.limit(chunkSize); bb.flip(); process(bb); } @Override public final Hasher putBytes(byte[] bytes) { return putBytes(bytes, 0, bytes.length); } @Override public final Hasher putBytes(byte[] bytes, int off, int len) { return putBytes(ByteBuffer.wrap(bytes, off, len).order(ByteOrder.LITTLE_ENDIAN)); } private Hasher putBytes(ByteBuffer readBuffer) { // If we have room for all of it, this is easy if (readBuffer.remaining() <= buffer.remaining()) { buffer.put(readBuffer); munchIfFull(); return this; } // First add just enough to fill buffer size, and munch that int bytesToCopy = bufferSize - buffer.position(); for (int i = 0; i < bytesToCopy; i++) { buffer.put(readBuffer.get()); } munch(); // buffer becomes empty here, since chunkSize divides bufferSize // Now process directly from the rest of the input buffer while (readBuffer.remaining() >= chunkSize) { process(readBuffer); } // Finally stick the remainder back in our usual buffer buffer.put(readBuffer); return this; } @Override public final Hasher putString(CharSequence charSequence) { for (int i = 0; i < charSequence.length(); i++) { putChar(charSequence.charAt(i)); } return this; } @Override public final Hasher putByte(byte b) { buffer.put(b); munchIfFull(); return this; } @Override public final Hasher putShort(short s) { buffer.putShort(s); munchIfFull(); return this; } @Override public final Hasher putChar(char c) { buffer.putChar(c); munchIfFull(); return this; } @Override public final Hasher putInt(int i) { buffer.putInt(i); munchIfFull(); return this; } @Override public final Hasher putLong(long l) { buffer.putLong(l); munchIfFull(); return this; } @Override public final Hasher putObject(T instance, Funnel funnel) { funnel.funnel(instance, this); return this; } @Override public final HashCode hash() { munch(); buffer.flip(); if (buffer.remaining() > 0) { processRemaining(buffer); } return makeHash(); } abstract HashCode makeHash(); // Process pent-up data in chunks private void munchIfFull() { if (buffer.remaining() < 8) { // buffer is full; not enough room for a primitive. We have at least one full chunk. munch(); } } private void munch() { buffer.flip(); while (buffer.remaining() >= chunkSize) { // we could limit the buffer to ensure process() does not read more than // chunkSize number of bytes, but we trust the implementations process(buffer); } buffer.compact(); // preserve any remaining data that do not make a full chunk } } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy