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 static com.google.common.base.Preconditions.checkArgument;
018import static com.google.common.base.Preconditions.checkNotNull;
019import static com.google.common.base.Preconditions.checkState;
020
021import com.google.common.annotations.Beta;
022import com.google.common.base.Preconditions;
023import com.google.common.primitives.Ints;
024import com.google.common.primitives.UnsignedInts;
025
026import java.io.Serializable;
027import java.security.MessageDigest;
028
029import javax.annotation.Nullable;
030
031/**
032 * An immutable hash code of arbitrary bit length.
033 *
034 * @author Dimitris Andreou
035 * @author Kurt Alfred Kluever
036 * @since 11.0
037 */
038@Beta
039public abstract class HashCode {
040  HashCode() {}
041
042  /**
043   * Returns the first four bytes of {@linkplain #asBytes() this hashcode's bytes}, converted to
044   * an {@code int} value in little-endian order.
045   *
046   * @throws IllegalStateException if {@code bits() < 32}
047   */
048  public abstract int asInt();
049
050  /**
051   * Returns the first eight bytes of {@linkplain #asBytes() this hashcode's bytes}, converted to
052   * a {@code long} value in little-endian order.
053   *
054   * @throws IllegalStateException if {@code bits() < 64}
055   */
056  public abstract long asLong();
057
058  /**
059   * If this hashcode has enough bits, returns {@code asLong()}, otherwise returns a {@code long}
060   * value with {@code asBytes()} as the least-significant bytes and {@code 0x00} as the remaining
061   * most-significant bytes.
062   *
063   * @since 14.0 (since 11.0 as {@code Hashing.padToLong(HashCode)})
064   */
065  public abstract long padToLong();
066
067  /**
068   * Returns the value of this hash code as a byte array. The caller may modify the byte array;
069   * changes to it will <i>not</i> be reflected in this {@code HashCode} object or any other arrays
070   * returned by this method.
071   */
072  // TODO(user): consider ByteString here, when that is available
073  public abstract byte[] asBytes();
074
075  /**
076   * Creates a 32-bit {@code HashCode} representation of the given int value. The underlying bytes
077   * are interpreted in little endian order.
078   *
079   * @since 15.0 (since 12.0 in HashCodes)
080   */
081  public static HashCode fromInt(int hash) {
082    return new IntHashCode(hash);
083  }
084
085  private static final class IntHashCode extends HashCode implements Serializable {
086    final int hash;
087
088    IntHashCode(int hash) {
089      this.hash = hash;
090    }
091
092    @Override
093    public int bits() {
094      return 32;
095    }
096
097    @Override
098    public byte[] asBytes() {
099      return new byte[] {
100          (byte) hash,
101          (byte) (hash >> 8),
102          (byte) (hash >> 16),
103          (byte) (hash >> 24)};
104    }
105
106    @Override
107    public int asInt() {
108      return hash;
109    }
110
111    @Override
112    public long asLong() {
113      throw new IllegalStateException("this HashCode only has 32 bits; cannot create a long");
114    }
115
116    @Override
117    public long padToLong() {
118      return UnsignedInts.toLong(hash);
119    }
120
121    private static final long serialVersionUID = 0;
122  }
123
124  /**
125   * Creates a 64-bit {@code HashCode} representation of the given long value. The underlying bytes
126   * are interpreted in little endian order.
127   *
128   * @since 15.0 (since 12.0 in HashCodes)
129   */
130  public static HashCode fromLong(long hash) {
131    return new LongHashCode(hash);
132  }
133
134  private static final class LongHashCode extends HashCode implements Serializable {
135    final long hash;
136
137    LongHashCode(long hash) {
138      this.hash = hash;
139    }
140
141    @Override
142    public int bits() {
143      return 64;
144    }
145
146    @Override
147    public byte[] asBytes() {
148      return new byte[] {
149          (byte) hash,
150          (byte) (hash >> 8),
151          (byte) (hash >> 16),
152          (byte) (hash >> 24),
153          (byte) (hash >> 32),
154          (byte) (hash >> 40),
155          (byte) (hash >> 48),
156          (byte) (hash >> 56)};
157    }
158
159    @Override
160    public int asInt() {
161      return (int) hash;
162    }
163
164    @Override
165    public long asLong() {
166      return hash;
167    }
168
169    @Override
170    public long padToLong() {
171      return hash;
172    }
173
174    private static final long serialVersionUID = 0;
175  }
176
177  /**
178   * Creates a {@code HashCode} from a byte array. The array is defensively copied to preserve
179   * the immutability contract of {@code HashCode}. The array cannot be empty.
180   *
181   * @since 15.0 (since 12.0 in HashCodes)
182   */
183  public static HashCode fromBytes(byte[] bytes) {
184    checkArgument(bytes.length >= 1, "A HashCode must contain at least 1 byte.");
185    return fromBytesNoCopy(bytes.clone());
186  }
187
188  /**
189   * Creates a {@code HashCode} from a byte array. The array is <i>not</i> copied defensively,
190   * so it must be handed-off so as to preserve the immutability contract of {@code HashCode}.
191   */
192  static HashCode fromBytesNoCopy(byte[] bytes) {
193    return new BytesHashCode(bytes);
194  }
195
196  private static final class BytesHashCode extends HashCode implements Serializable {
197    final byte[] bytes;
198
199    BytesHashCode(byte[] bytes) {
200      this.bytes = checkNotNull(bytes);
201    }
202
203    @Override
204    public int bits() {
205      return bytes.length * 8;
206    }
207
208    @Override
209    public byte[] asBytes() {
210      return bytes.clone();
211    }
212
213    @Override
214    public int asInt() {
215      checkState(bytes.length >= 4,
216          "HashCode#asInt() requires >= 4 bytes (it only has %s bytes).", bytes.length);
217      return (bytes[0] & 0xFF)
218          | ((bytes[1] & 0xFF) << 8)
219          | ((bytes[2] & 0xFF) << 16)
220          | ((bytes[3] & 0xFF) << 24);
221    }
222
223    @Override
224    public long asLong() {
225      checkState(bytes.length >= 8,
226          "HashCode#asLong() requires >= 8 bytes (it only has %s bytes).", bytes.length);
227      return padToLong();
228    }
229
230    @Override
231    public long padToLong() {
232      long retVal = (bytes[0] & 0xFF);
233      for (int i = 1; i < Math.min(bytes.length, 8); i++) {
234        retVal |= (bytes[i] & 0xFFL) << (i * 8);
235      }
236      return retVal;
237    }
238
239    private static final long serialVersionUID = 0;
240  }
241
242  /**
243   * Returns the number of bits in this hash code; a positive multiple of 8.
244   */
245  public abstract int bits();
246
247  /**
248   * Creates a {@code HashCode} from a hexadecimal ({@code base 16}) encoded string. The string must
249   * be at least 2 characters long, and contain only valid, lower-cased hexadecimal characters.
250   *
251   * <p>This method accepts the exact format generated by {@link #toString}. If you require more
252   * lenient {@code base 16} decoding, please use
253   * {@link com.google.common.io.BaseEncoding#decode} (and pass the result to {@link #fromBytes}).
254   *
255   * @since 15.0
256   */
257  public static HashCode fromString(String string) {
258    checkArgument(string.length() >= 2,
259        "input string (%s) must have at least 2 characters", string);
260    checkArgument(string.length() % 2 == 0,
261        "input string (%s) must have an even number of characters", string);
262
263    byte[] bytes = new byte[string.length() / 2];
264    for (int i = 0; i < string.length(); i += 2) {
265      int ch1 = decode(string.charAt(i)) << 4;
266      int ch2 = decode(string.charAt(i + 1));
267      bytes[i / 2] = (byte) (ch1 + ch2);
268    }
269    return fromBytesNoCopy(bytes);
270  }
271
272  private static int decode(char ch) {
273    if (ch >= '0' && ch <= '9') {
274      return ch - '0';
275    }
276    if (ch >= 'a' && ch <= 'f') {
277      return ch - 'a' + 10;
278    }
279    throw new IllegalArgumentException("Illegal hexadecimal character: " + ch);
280  }
281
282  /**
283   * Copies bytes from this hash code into {@code dest}.
284   *
285   * @param dest the byte array into which the hash code will be written
286   * @param offset the start offset in the data
287   * @param maxLength the maximum number of bytes to write
288   * @return the number of bytes written to {@code dest}
289   * @throws IndexOutOfBoundsException if there is not enough room in {@code dest}
290   */
291  public final int writeBytesTo(byte[] dest, int offset, int maxLength) {
292    maxLength = Ints.min(maxLength, bits() / 8);
293    Preconditions.checkPositionIndexes(offset, offset + maxLength, dest.length);
294    // TODO(user): Consider avoiding the array creation in asBytes() by stepping through
295    // the bytes individually.
296    byte[] hash = asBytes();
297    System.arraycopy(hash, 0, dest, offset, maxLength);
298    return maxLength;
299  }
300
301  @Override
302  public final boolean equals(@Nullable Object object) {
303    if (object instanceof HashCode) {
304      HashCode that = (HashCode) object;
305      // Undocumented: this is a non-short-circuiting equals(), in case this is a cryptographic
306      // hash code, in which case we don't want to leak timing information
307      return MessageDigest.isEqual(this.asBytes(), that.asBytes());
308    }
309    return false;
310  }
311
312  /**
313   * Returns a "Java hash code" for this {@code HashCode} instance; this is well-defined
314   * (so, for example, you can safely put {@code HashCode} instances into a {@code
315   * HashSet}) but is otherwise probably not what you want to use.
316   */
317  @Override
318  public final int hashCode() {
319    // If we have at least 4 bytes (32 bits), just take the first 4 bytes. Since this is
320    // already a (presumably) high-quality hash code, any four bytes of it will do.
321    if (bits() >= 32) {
322      return asInt();
323    }
324    // If we have less than 4 bytes, use them all.
325    byte[] bytes = asBytes();
326    int val = (bytes[0] & 0xFF);
327    for (int i = 1; i < bytes.length; i++) {
328      val |= ((bytes[i] & 0xFF) << (i * 8));
329    }
330    return val;
331  }
332
333  /**
334   * Returns a string containing each byte of {@link #asBytes}, in order, as a two-digit unsigned
335   * hexadecimal number in lower case.
336   *
337   * <p>Note that if the output is considered to be a single hexadecimal number, this hash code's
338   * bytes are the <i>big-endian</i> representation of that number. This may be surprising since
339   * everything else in the hashing API uniformly treats multibyte values as little-endian. But
340   * this format conveniently matches that of utilities such as the UNIX {@code md5sum} command.
341   *
342   * <p>To create a {@code HashCode} from its string representation, see {@link #fromString}.
343   */
344  @Override
345  public final String toString() {
346    byte[] bytes = asBytes();
347    StringBuilder sb = new StringBuilder(2 * bytes.length);
348    for (byte b : bytes) {
349      sb.append(hexDigits[(b >> 4) & 0xf]).append(hexDigits[b & 0xf]);
350    }
351    return sb.toString();
352  }
353
354  private static final char[] hexDigits = "0123456789abcdef".toCharArray();
355}