|
| 1 | +// ported from xxbloom - https://github.com/ceejbot/xxbloom/blob/master/LICENSE |
| 2 | +import { expect } from 'aegir/chai' |
| 3 | +import { fromString as uint8ArrayFromString } from 'uint8arrays/from-string' |
| 4 | +import { BloomFilter } from '../src/bloom-filter.js' |
| 5 | + |
| 6 | +function hasBitsSet (buffer: Uint8Array): number { |
| 7 | + let isset = 0 |
| 8 | + for (let i = 0; i < buffer.length; i++) { |
| 9 | + isset |= (buffer[i] !== 0 ? 1 : 0) |
| 10 | + } |
| 11 | + return isset |
| 12 | +} |
| 13 | + |
| 14 | +describe('bloom-filter', () => { |
| 15 | + it('constructs a filter of the requested size', () => { |
| 16 | + const filter = new BloomFilter({ hashes: 4, bits: 32 }) |
| 17 | + expect(filter.seeds).to.have.lengthOf(4) |
| 18 | + expect(filter.bits).to.equal(32) |
| 19 | + expect(filter.buffer).to.be.an.instanceOf(Uint8Array) |
| 20 | + }) |
| 21 | + |
| 22 | + it('zeroes out its storage buffer', () => { |
| 23 | + const filter = new BloomFilter({ hashes: 3, bits: 64 }) |
| 24 | + for (let i = 0; i < filter.buffer.length; i++) { |
| 25 | + expect(filter.buffer[i]).to.equal(0) |
| 26 | + } |
| 27 | + }) |
| 28 | + |
| 29 | + it('uses passed-in seeds if provided', () => { |
| 30 | + const filter = new BloomFilter({ bits: 256, seeds: [1, 2, 3, 4, 5] }) |
| 31 | + expect(filter.seeds.length).to.equal(5) |
| 32 | + expect(filter.seeds[0]).to.equal(1) |
| 33 | + expect(filter.seeds[4]).to.equal(5) |
| 34 | + }) |
| 35 | + |
| 36 | + describe('createOptimal()', () => { |
| 37 | + it('creates a filter with good defaults', () => { |
| 38 | + let filter = BloomFilter.create(95) |
| 39 | + expect(filter.bits).to.equal(1048) |
| 40 | + expect(filter.seeds.length).to.equal(8) |
| 41 | + |
| 42 | + filter = BloomFilter.create(148) |
| 43 | + expect(filter.bits).to.equal(1632) |
| 44 | + expect(filter.seeds.length).to.equal(8) |
| 45 | + |
| 46 | + filter = BloomFilter.create(10) |
| 47 | + expect(filter.bits).to.equal(110) |
| 48 | + expect(filter.seeds.length).to.equal(8) |
| 49 | + }) |
| 50 | + |
| 51 | + it('createOptimal() lets you specify an error rate', () => { |
| 52 | + let filter = BloomFilter.create(20000) |
| 53 | + expect(filter.bits).to.equal(220555) |
| 54 | + const previous = filter.bits |
| 55 | + |
| 56 | + filter = BloomFilter.create(20000, 0.2) |
| 57 | + expect(filter.bits).to.be.below(previous) |
| 58 | + }) |
| 59 | + }) |
| 60 | + |
| 61 | + describe('setbit() and getbit()', () => { |
| 62 | + it('sets the specified bit', () => { |
| 63 | + const filter = new BloomFilter({ hashes: 3, bits: 16 }) |
| 64 | + |
| 65 | + filter.setbit(0) |
| 66 | + let val = filter.getbit(0) |
| 67 | + expect(val).to.equal(true) |
| 68 | + |
| 69 | + filter.setbit(1) |
| 70 | + val = filter.getbit(1) |
| 71 | + expect(val).to.equal(true) |
| 72 | + |
| 73 | + val = filter.getbit(2) |
| 74 | + expect(val).to.equal(false) |
| 75 | + |
| 76 | + filter.setbit(10) |
| 77 | + val = filter.getbit(10) |
| 78 | + expect(val).to.equal(true) |
| 79 | + }) |
| 80 | + |
| 81 | + it('can set all bits', () => { |
| 82 | + let i: number |
| 83 | + let value: number |
| 84 | + |
| 85 | + const filter = new BloomFilter({ hashes: 3, bits: 16 }) |
| 86 | + expect(filter.buffer.length).to.equal(2) |
| 87 | + |
| 88 | + for (i = 0; i < 16; i++) { |
| 89 | + filter.setbit(i) |
| 90 | + } |
| 91 | + |
| 92 | + for (i = 0; i < 2; i++) { |
| 93 | + value = filter.buffer[i] |
| 94 | + expect(value).to.equal(255) |
| 95 | + } |
| 96 | + }) |
| 97 | + |
| 98 | + it('slides over into the next buffer slice when setting bits', () => { |
| 99 | + let val |
| 100 | + const filter = new BloomFilter({ hashes: 3, bits: 64 }) |
| 101 | + |
| 102 | + filter.setbit(8) |
| 103 | + val = filter.buffer[1] |
| 104 | + expect(val).to.equal(1) |
| 105 | + |
| 106 | + filter.setbit(17) |
| 107 | + val = filter.buffer[2] |
| 108 | + expect(val).to.equal(2) |
| 109 | + |
| 110 | + filter.setbit(34) |
| 111 | + val = filter.buffer[4] |
| 112 | + expect(val).to.equal(4) |
| 113 | + }) |
| 114 | + }) |
| 115 | + |
| 116 | + describe('add()', () => { |
| 117 | + it('can store buffers', () => { |
| 118 | + const filter = new BloomFilter({ hashes: 4, bits: 128 }) |
| 119 | + |
| 120 | + expect(hasBitsSet(filter.buffer)).to.equal(0) |
| 121 | + filter.add(uint8ArrayFromString('cat')) |
| 122 | + expect(hasBitsSet(filter.buffer)).to.equal(1) |
| 123 | + }) |
| 124 | + |
| 125 | + it('can store strings', () => { |
| 126 | + const filter = new BloomFilter({ hashes: 4, bits: 128 }) |
| 127 | + filter.add('cat') |
| 128 | + |
| 129 | + expect(hasBitsSet(filter.buffer)).to.equal(1) |
| 130 | + }) |
| 131 | + |
| 132 | + it('can add a hundred random items', () => { |
| 133 | + const alpha = '0123456789abcdefghijklmnopqrstuvwxyz' |
| 134 | + function randomWord (length?: number): string { |
| 135 | + length = length ?? Math.ceil(Math.random() * 20) |
| 136 | + let result = '' |
| 137 | + for (let i = 0; i < length; i++) { |
| 138 | + result += alpha[Math.floor(Math.random() * alpha.length)] |
| 139 | + } |
| 140 | + |
| 141 | + return result |
| 142 | + } |
| 143 | + |
| 144 | + const filter = BloomFilter.create(100) |
| 145 | + const words: string[] = [] |
| 146 | + |
| 147 | + for (let i = 0; i < 100; i++) { |
| 148 | + const w = randomWord() |
| 149 | + words.push(w) |
| 150 | + filter.add(w) |
| 151 | + } |
| 152 | + |
| 153 | + for (let i = 0; i < words.length; i++) { |
| 154 | + expect(filter.has(words[i])).to.equal(true) |
| 155 | + } |
| 156 | + }) |
| 157 | + }) |
| 158 | + |
| 159 | + describe('has()', () => { |
| 160 | + it('returns true when called on a stored item', () => { |
| 161 | + const filter = new BloomFilter({ hashes: 3, bits: 16 }) |
| 162 | + filter.add('cat') |
| 163 | + |
| 164 | + expect(hasBitsSet(filter.buffer)).to.equal(1) |
| 165 | + expect(filter.has('cat')).to.be.true() |
| 166 | + }) |
| 167 | + |
| 168 | + it('returns false for items not in the set (mostly)', () => { |
| 169 | + const filter = new BloomFilter({ hashes: 4, bits: 50 }) |
| 170 | + filter.add('cat') |
| 171 | + expect(filter.has('dog')).to.be.false() |
| 172 | + }) |
| 173 | + |
| 174 | + it('responds appropriately for arrays of added items', () => { |
| 175 | + const filter = new BloomFilter({ hashes: 3, bits: 128 }) |
| 176 | + filter.add('cat') |
| 177 | + filter.add('dog') |
| 178 | + filter.add('wallaby') |
| 179 | + |
| 180 | + expect(filter.has('cat')).to.equal(true) |
| 181 | + expect(filter.has('dog')).to.equal(true) |
| 182 | + expect(filter.has('wallaby')).to.equal(true) |
| 183 | + expect(filter.has('orange')).to.equal(false) |
| 184 | + }) |
| 185 | + }) |
| 186 | + |
| 187 | + describe('clear()', () => { |
| 188 | + it('clears the filter', () => { |
| 189 | + const filter = new BloomFilter({ hashes: 3, bits: 128 }) |
| 190 | + filter.add('cat') |
| 191 | + filter.add('dog') |
| 192 | + filter.add('wallaby') |
| 193 | + expect(hasBitsSet(filter.buffer)).to.equal(1) |
| 194 | + |
| 195 | + filter.clear() |
| 196 | + expect(hasBitsSet(filter.buffer)).to.equal(0) |
| 197 | + }) |
| 198 | + }) |
| 199 | +}) |
0 commit comments