001/*
002 * Copyright (C) 2011 The Guava Authors
003 *
004 * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
005 * in compliance with the License. You may obtain a copy of the License at
006 *
007 * http://www.apache.org/licenses/LICENSE-2.0
008 *
009 * Unless required by applicable law or agreed to in writing, software distributed under the License
010 * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
011 * or implied. See the License for the specific language governing permissions and limitations under
012 * the License.
013 */
014
015package com.google.common.hash;
016
017import com.google.common.annotations.Beta;
018import com.google.common.primitives.Ints;
019import java.nio.charset.Charset;
020
021/**
022 * A hash function is a collision-averse pure function that maps an arbitrary block of data to a
023 * number called a <i>hash code</i>.
024 *
025 * <h3>Definition</h3>
026 *
027 * <p>Unpacking this definition:
028 *
029 * <ul>
030 * <li><b>block of data:</b> the input for a hash function is always, in concept, an ordered byte
031 *     array. This hashing API accepts an arbitrary sequence of byte and multibyte values (via
032 *     {@link Hasher}), but this is merely a convenience; these are always translated into raw byte
033 *     sequences under the covers.
034 *
035 * <li><b>hash code:</b> each hash function always yields hash codes of the same fixed bit length
036 *     (given by {@link #bits}). For example, {@link Hashing#sha1} produces a 160-bit number, while
037 *     {@link Hashing#murmur3_32()} yields only 32 bits. Because a {@code long} value is clearly
038 *     insufficient to hold all hash code values, this API represents a hash code as an instance of
039 *     {@link HashCode}.
040 *
041 * <li><b>pure function:</b> the value produced must depend only on the input bytes, in the order
042 *     they appear. Input data is never modified. {@link HashFunction} instances should always be
043 *     stateless, and therefore thread-safe.
044 *
045 * <li><b>collision-averse:</b> while it can't be helped that a hash function will sometimes produce
046 *     the same hash code for distinct inputs (a "collision"), every hash function strives to
047 *     <i>some</i> degree to make this unlikely. (Without this condition, a function that always
048 *     returns zero could be called a hash function. It is not.)
049 * </ul>
050 *
051 * <p>Summarizing the last two points: "equal yield equal <i>always</i>; unequal yield unequal
052 * <i>often</i>." This is the most important characteristic of all hash functions.
053 *
054 * <h3>Desirable properties</h3>
055 *
056 * <p>A high-quality hash function strives for some subset of the following virtues:
057 *
058 * <ul>
059 * <li><b>collision-resistant:</b> while the definition above requires making at least <i>some</i>
060 *     token attempt, one measure of the quality of a hash function is <i>how well</i> it succeeds
061 *     at this goal. Important note: it may be easy to achieve the theoretical minimum collision
062 *     rate when using completely <i>random</i> sample input. The true test of a hash function is
063 *     how it performs on representative real-world data, which tends to contain many hidden
064 *     patterns and clumps. The goal of a good hash function is to stamp these patterns out as
065 *     thoroughly as possible.
066 *
067 * <li><b>bit-dispersing:</b> masking out any <i>single bit</i> from a hash code should yield only
068 *     the expected <i>twofold</i> increase to all collision rates. Informally, the "information" in
069 *     the hash code should be as evenly "spread out" through the hash code's bits as possible. The
070 *     result is that, for example, when choosing a bucket in a hash table of size 2^8, <i>any</i>
071 *     eight bits could be consistently used.
072 *
073 * <li><b>cryptographic:</b> certain hash functions such as {@link Hashing#sha512} are designed to
074 *     make it as infeasible as possible to reverse-engineer the input that produced a given hash
075 *     code, or even to discover <i>any</i> two distinct inputs that yield the same result. These
076 *     are called <i>cryptographic hash functions</i>. But, whenever it is learned that either of
077 *     these feats has become computationally feasible, the function is deemed "broken" and should
078 *     no longer be used for secure purposes. (This is the likely eventual fate of <i>all</i>
079 *     cryptographic hashes.)
080 *
081 * <li><b>fast:</b> perhaps self-explanatory, but often the most important consideration. We have
082 *     published <a href="#noWeHaventYet">microbenchmark results</a> for many common hash functions.
083 * </ul>
084 *
085 * <h3>Providing input to a hash function</h3>
086 *
087 * <p>The primary way to provide the data that your hash function should act on is via a
088 * {@link Hasher}. Obtain a new hasher from the hash function using {@link #newHasher}, "push" the
089 * relevant data into it using methods like {@link Hasher#putBytes(byte[])}, and finally ask for the
090 * {@code HashCode} when finished using {@link Hasher#hash}. (See an {@linkplain #newHasher example}
091 * of this.)
092 *
093 * <p>If all you want to hash is a single byte array, string or {@code long} value, there are
094 * convenient shortcut methods defined directly on {@link HashFunction} to make this easier.
095 *
096 * <p>Hasher accepts primitive data types, but can also accept any Object of type {@code
097 * T} provided that you implement a {@link Funnel}{@code <T>} to specify how to "feed" data from
098 * that object into the function. (See {@linkplain Hasher#putObject an example} of this.)
099 *
100 * <p><b>Compatibility note:</b> Throughout this API, multibyte values are always interpreted in
101 * <i>little-endian</i> order. That is, hashing the byte array {@code {0x01, 0x02, 0x03, 0x04}} is
102 * equivalent to hashing the {@code int} value {@code 0x04030201}. If this isn't what you need,
103 * methods such as {@link Integer#reverseBytes} and {@link Ints#toByteArray} will help.
104 *
105 * <h3>Relationship to {@link Object#hashCode}</h3>
106 *
107 * <p>Java's baked-in concept of hash codes is constrained to 32 bits, and provides no separation
108 * between hash algorithms and the data they act on, so alternate hash algorithms can't be easily
109 * substituted. Also, implementations of {@code hashCode} tend to be poor-quality, in part because
110 * they end up depending on <i>other</i> existing poor-quality {@code hashCode} implementations,
111 * including those in many JDK classes.
112 *
113 * <p>{@code Object.hashCode} implementations tend to be very fast, but have weak collision
114 * prevention and <i>no</i> expectation of bit dispersion. This leaves them perfectly suitable for
115 * use in hash tables, because extra collisions cause only a slight performance hit, while poor bit
116 * dispersion is easily corrected using a secondary hash function (which all reasonable hash table
117 * implementations in Java use). For the many uses of hash functions beyond data structures,
118 * however, {@code Object.hashCode} almost always falls short -- hence this library.
119 *
120 * @author Kevin Bourrillion
121 * @since 11.0
122 */
123@Beta
124public interface HashFunction {
125  /**
126   * Begins a new hash code computation by returning an initialized, stateful {@code
127   * Hasher} instance that is ready to receive data. Example: <pre>   {@code
128   *
129   *   HashFunction hf = Hashing.md5();
130   *   HashCode hc = hf.newHasher()
131   *       .putLong(id)
132   *       .putBoolean(isActive)
133   *       .hash();}</pre>
134   */
135  Hasher newHasher();
136
137  /**
138   * Begins a new hash code computation as {@link #newHasher()}, but provides a hint of the expected
139   * size of the input (in bytes). This is only important for non-streaming hash functions (hash
140   * functions that need to buffer their whole input before processing any of it).
141   */
142  Hasher newHasher(int expectedInputSize);
143
144  /**
145   * Shortcut for {@code newHasher().putInt(input).hash()}; returns the hash code for the given
146   * {@code int} value, interpreted in little-endian byte order. The implementation <i>might</i>
147   * perform better than its longhand equivalent, but should not perform worse.
148   *
149   * @since 12.0
150   */
151  HashCode hashInt(int input);
152
153  /**
154   * Shortcut for {@code newHasher().putLong(input).hash()}; returns the hash code for the given
155   * {@code long} value, interpreted in little-endian byte order. The implementation <i>might</i>
156   * perform better than its longhand equivalent, but should not perform worse.
157   */
158  HashCode hashLong(long input);
159
160  /**
161   * Shortcut for {@code newHasher().putBytes(input).hash()}. The implementation <i>might</i>
162   * perform better than its longhand equivalent, but should not perform worse.
163   */
164  HashCode hashBytes(byte[] input);
165
166  /**
167   * Shortcut for {@code newHasher().putBytes(input, off, len).hash()}. The implementation
168   * <i>might</i> perform better than its longhand equivalent, but should not perform worse.
169   *
170   * @throws IndexOutOfBoundsException if {@code off < 0} or {@code off + len > bytes.length} or
171   *     {@code len < 0}
172   */
173  HashCode hashBytes(byte[] input, int off, int len);
174
175  /**
176   * Shortcut for {@code newHasher().putUnencodedChars(input).hash()}. The implementation
177   * <i>might</i> perform better than its longhand equivalent, but should not perform worse. Note
178   * that no character encoding is performed; the low byte and high byte of each {@code char} are
179   * hashed directly (in that order).
180   *
181   * <p><b>Warning:</b> This method will produce different output than most other languages do when
182   * running the same hash function on the equivalent input. For cross-language compatibility, use
183   * {@link #hashString}, usually with a charset of UTF-8. For other use cases, use {@code
184   * hashUnencodedChars}.
185   *
186   * @since 15.0 (since 11.0 as hashString(CharSequence)).
187   */
188  HashCode hashUnencodedChars(CharSequence input);
189
190  /**
191   * Shortcut for {@code newHasher().putString(input, charset).hash()}. Characters are encoded using
192   * the given {@link Charset}. The implementation <i>might</i> perform better than its longhand
193   * equivalent, but should not perform worse.
194   *
195   * <p><b>Warning:</b> This method, which reencodes the input before hashing it, is useful only for
196   * cross-language compatibility. For other use cases, prefer {@link #hashUnencodedChars}, which is
197   * faster, produces the same output across Java releases, and hashes every {@code char} in the
198   * input, even if some are invalid.
199   */
200  HashCode hashString(CharSequence input, Charset charset);
201
202  /**
203   * Shortcut for {@code newHasher().putObject(instance, funnel).hash()}. The implementation
204   * <i>might</i> perform better than its longhand equivalent, but should not perform worse.
205   *
206   * @since 14.0
207   */
208  <T> HashCode hashObject(T instance, Funnel<? super T> funnel);
209
210  /**
211   * Returns the number of bits (a multiple of 32) that each hash code produced by this hash
212   * function has.
213   */
214  int bits();
215}