BloomFilter32bit


1   /* BloomFilter32bit
2   *
3   * $Id: BloomFilter32bit.java,v 1.2.16.1 2007/01/13 01:31:39 stack-sf Exp $
4   *
5   * Created on Jun 21, 2005
6   *
7   * Copyright (C) 2005 Internet Archive; a slight adaptation of
8   * LGPL work (C) Sebastiano Vigna
9   *
10  * This file is part of the Heritrix web crawler (crawler.archive.org).
11  *
12  * Heritrix is free software; you can redistribute it and/or modify
13  * it under the terms of the GNU Lesser Public License as published by
14  * the Free Software Foundation; either version 2.1 of the License, or
15  * any later version.
16  *
17  * Heritrix is distributed in the hope that it will be useful,
18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
20  * GNU Lesser Public License for more details.
21  *
22  * You should have received a copy of the GNU Lesser Public License
23  * along with Heritrix; if not, write to the Free Software
24  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
25  */
26  
27  package org.archive.util;
28  
29  import java.io.Serializable  ;
30  import java.security.SecureRandom  ;
31  
32  /** A Bloom filter.
33   *
34   * SLIGHTLY ADAPTED VERSION OF MG4J it.unimi.dsi.mg4j.util.BloomFilter
35   * 
36   * <p>KEY CHANGES:
37   *
38   * <ul>
39   * <li>Adapted to use 32bit ops as much as possible... may be slightly
40   * faster on 32bit hardware/OS</li>
41   * <li>NUMBER_OF_WEIGHTS is 2083, to better avoid collisions between 
42   * similar strings</li>
43   * <li>Removed dependence on cern.colt MersenneTwister (replaced with
44   * SecureRandom) and QuickBitVector (replaced with local methods).</li>
45   * </ul>
46   * 
47   * <hr>
48   * 
49   * <P>Instances of this class represent a set of character sequences (with false positives)
50   * using a Bloom filter. Because of the way Bloom filters work,
51   * you cannot remove elements.
52   *
53   * <P>Bloom filters have an expected error rate, depending on the number
54   * of hash functions used, on the filter size and on the number of elements in the filter. This implementation
55   * uses a variable optimal number of hash functions, depending on the expected
56   * number of elements. More precisely, a Bloom
57   * filter for <var>n</var> character sequences with <var>d</var> hash functions will use
58   * ln 2 <var>d</var><var>n</var> &#8776; 1.44 <var>d</var><var>n</var> bits;
59   * false positives will happen with probability 2<sup>-<var>d</var></sup>.
60   *
61   * <P>Hash functions are generated at creation time using universal hashing. Each hash function
62   * uses {@link #NUMBER_OF_WEIGHTS} random integers, which are cyclically multiplied by
63   * the character codes in a character sequence. The resulting integers are XOR-ed together.
64   *
65   * <P>This class exports access methods that are very similar to those of {@link java.util.Set},
66   * but it does not implement that interface, as too many non-optional methods
67   * would be unimplementable (e.g., iterators).
68   *
69   * @author Sebastiano Vigna
70   */
71  public class BloomFilter32bit implements Serializable  , BloomFilter {
72      
73      private static final long serialVersionUID = -1567837798979475689L;
74      
75      /** The number of weights used to create hash functions. */
76      final public static int NUMBER_OF_WEIGHTS = 2083; // CHANGED FROM 16
77      /** The number of bits in this filter. */
78      final public long m; 
79      /** The number of hash functions used by this filter. */
80      final public int d;
81      /** The underlying bit vectorS. */
82      final private int[] bits;
83      /** The random integers used to generate the hash functions. */
84      final private int[][] weight;
85  
86      /** The number of elements currently in the filter. It may be
87       * smaller than the actual number of additions of distinct character
88       * sequences because of false positives.
89       */
90      private int size;
91  
92      /** The natural logarithm of 2, used in the computation of the number of bits. */
93      private final static double NATURAL_LOG_OF_2 = Math.log( 2 );
94  
95      private final static boolean DEBUG = false;
96  
97      /** Creates a new Bloom filter with given number of hash functions and expected number of elements.
98       *
99       * @param n the expected number of elements.
100      * @param d the number of hash functions; if the filter add not more than <code>n</code> elements,
101      * false positives will happen with probability 2<sup>-<var>d</var></sup>.
102      */
103     public BloomFilter32bit( final int n, final int d ) {
104         this.d = d;
105         int len =
106             (int)Math.ceil( ( (long)n * (long)d / NATURAL_LOG_OF_2 ) / 32 );
107         this.m = len*32L;
108         if ( m >= 1L<<32 ) {
109             throw new IllegalArgumentException  ( "This filter would require " + m + " bits" );
110         }
111         bits = new int[ len ];
112 
113         if ( DEBUG ) System.err.println( "Number of bits: " + m );
114 
115         // seeded for reproduceable behavior in repeated runs; BUT: 
116         // SecureRandom's default implementation (as of 1.5) 
117         // seems to mix in its own seeding.
118         final SecureRandom   random = new SecureRandom  (new byte[] {19,96});
119         weight = new int[ d ][];
120         for( int i = 0; i < d; i++ ) {
121             weight[ i ] = new int[ NUMBER_OF_WEIGHTS ];
122             for( int j = 0; j < NUMBER_OF_WEIGHTS; j++ )
123                  weight[ i ][ j ] = random.nextInt();
124         }
125     }
126 
127     /** The number of character sequences in the filter.
128      *
129      * @return the number of character sequences in the filter (but see {@link #contains(CharSequence)}).
130      */
131 
132     public int size() {
133         return size;
134     }
135 
136     /** Hashes the given sequence with the given hash function.
137      *
138      * @param s a character sequence.
139      * @param l the length of <code>s</code>.
140      * @param k a hash function index (smaller than {@link #d}).
141      * @return the position in the filter corresponding to <code>s</code> for the hash function <code>k</code>.
142      */
143     private long hash( final CharSequence   s, final int l, final int k ) {
144         final int[] w = weight[ k ];
145         int h = 0, i = l;
146         while( i-- != 0 ) h ^= s.charAt( i ) * w[ i % NUMBER_OF_WEIGHTS ];
147         return ((long)h-Integer.MIN_VALUE) % m; 
148     }
149 
150     /** Checks whether the given character sequence is in this filter.
151      *
152      * <P>Note that this method may return true on a character sequence that is has
153      * not been added to the filter. This will happen with probability 2<sub>-<var>d</var></sub>,
154      * where <var>d</var> is the number of hash functions specified at creation time, if
155      * the number of the elements in the filter is less than <var>n</var>, the number
156      * of expected elements specified at creation time.
157      *
158      * @param s a character sequence.
159      * @return true if the sequence is in the filter (or if a sequence with the
160      * same hash sequence is in the filter).
161      */
162 
163     public boolean contains( final CharSequence   s ) {
164         int i = d, l = s.length();
165         while( i-- != 0 ) if ( ! getBit( hash( s, l, i ) ) ) return false;
166         return true;
167     }
168 
169     /** Adds a character sequence to the filter.
170      *
171      * @param s a character sequence.
172      * @return true if the character sequence was not in the filter (but see {@link #contains(CharSequence)}).
173      */
174 
175     public boolean add( final CharSequence   s ) {
176         boolean result = false;
177         int i = d, l = s.length();
178         long h;
179         while( i-- != 0 ) {
180             h = hash( s, l, i );
181             if ( ! getBit( h ) ) result = true;
182             setBit( h );
183         }
184         if ( result ) size++;
185         return result;
186     }
187     
188     protected final static long ADDRESS_BITS_PER_UNIT = 5; // 32=2^5
189     protected final static long BIT_INDEX_MASK = 31; // = BITS_PER_UNIT - 1;
190 
191     /**
192      * Returns from the local bitvector the value of the bit with 
193      * the specified index. The value is <tt>true</tt> if the bit 
194      * with the index <tt>bitIndex</tt> is currently set; otherwise, 
195      * returns <tt>false</tt>.
196      *
197      * (adapted from cern.colt.bitvector.QuickBitVector)
198      * 
199      * @param     bitIndex   the bit index.
200      * @return    the value of the bit with the specified index.
201      */
202     protected boolean getBit(long bitIndex) {
203         return ((bits[(int)(bitIndex >> ADDRESS_BITS_PER_UNIT)] & (1 << (bitIndex & BIT_INDEX_MASK))) != 0);
204     }
205 
206     /**
207      * Changes the bit with index <tt>bitIndex</tt> in local bitvector.
208      *
209      * (adapted from cern.colt.bitvector.QuickBitVector)
210      * 
211      * @param     bitIndex   the index of the bit to be set.
212      */
213     protected void setBit(long bitIndex) {
214             bits[(int)(bitIndex >> ADDRESS_BITS_PER_UNIT)] |= 1 << (bitIndex & BIT_INDEX_MASK);
215     }
216 
217     /* (non-Javadoc)
218      * @see org.archive.util.BloomFilter#getSizeBytes()
219      */
220     public long getSizeBytes() {
221         return bits.length*4;
222     }
223 }
224
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags