A standard Bloom filter.
| 5 | * |
| 6 | */ |
| 7 | public class Bloom { |
| 8 | |
| 9 | public static Bloom construct(long[] keys, double bitsPerKey) { |
| 10 | long n = keys.length; |
| 11 | int k = getBestK(bitsPerKey); |
| 12 | Bloom f = new Bloom((int) n, bitsPerKey, k); |
| 13 | for(long x : keys) { |
| 14 | f.add(x); |
| 15 | } |
| 16 | return f; |
| 17 | } |
| 18 | |
| 19 | private static int getBestK(double bitsPerKey) { |
| 20 | return Math.max(1, (int) Math.round(bitsPerKey * Math.log(2))); |
| 21 | } |
| 22 | |
| 23 | private final int k; |
| 24 | private final long bits; |
| 25 | private final long seed; |
| 26 | private final int arraySize; |
| 27 | private final long[] data; |
| 28 | |
| 29 | Bloom(int entryCount, double bitsPerKey, int k) { |
| 30 | entryCount = Math.max(1, entryCount); |
| 31 | this.k = k; |
| 32 | this.seed = Hash.randomSeed(); |
| 33 | this.bits = (long) (entryCount * bitsPerKey); |
| 34 | arraySize = (int) ((bits + 63) / 64); |
| 35 | data = new long[arraySize]; |
| 36 | } |
| 37 | |
| 38 | public void add(long key) { |
| 39 | long hash = Hash.hash64(key, seed); |
| 40 | long a = (hash >>> 32) | (hash << 32); |
| 41 | long b = hash; |
| 42 | for (int i = 0; i < k; i++) { |
| 43 | data[Hash.reduce((int) (a >>> 32), arraySize)] |= 1L << a; |
| 44 | a += b; |
| 45 | } |
| 46 | } |
| 47 | |
| 48 | public boolean mayContain(long key) { |
| 49 | long hash = Hash.hash64(key, seed); |
| 50 | long a = (hash >>> 32) | (hash << 32); |
| 51 | long b = hash; |
| 52 | for (int i = 0; i < k; i++) { |
| 53 | if ((data[Hash.reduce((int) (a >>> 32), arraySize)] & 1L << a) == 0) { |
| 54 | return false; |
| 55 | } |
| 56 | a += b; |
| 57 | } |
| 58 | return true; |
| 59 | } |
| 60 | |
| 61 | public byte[][] getData() { |
| 62 | byte[][] d = new byte[data.length][]; |
| 63 | for (int i = 0; i < data.length; i++) { |
| 64 | d[i] = tool.longToBytes(data[i]); |
nothing calls this directly
no outgoing calls
no test coverage detected