ca-bloom-filter

A lightweight, performant bloom filter implementation with a simple API.

Installing

Via NPM: npm install ca-bloom-filter --save.

Getting Started

After installing

import BloomFilter from 'ca-bloom-filter';

const bloomFilter = new BloomFilter(8, 1); // 8 bit filter using 1 hash.

Example Usage:

import BloomFilter from 'ca-bloom-filter';

const bloomFilter = new BloomFilter(8, 1);

bloom.contains('cheese'); // false
bloom.add('cheese');
bloom.contains('cheese'); // true

Condensed Documentation

Below is a condensed form of the documentation, each is a function that can be found on the BloomFilter object, called like so.

const bloom = new BloomFilter(42, 4);
bloom.contains('cheese'); // false
bloom.add('cheese');
bloom.contains('cheese'); // true

BloomFilter

Basic bloom filter implementation.

const bloom = new BloomFilter(42, 4);
bloom.contains('cheese'); // false
bloom.add('cheese');
bloom.contains('cheese'); // true

| Method | Parameters | Return | | ----------- | -------- | ------ | | .add(key) | key:String | Returns BloomFilter for chaining. | | .contains(key) | key:String | Returns true if the item is within the filter, false otherwise. | | .equals(bloomFilter) | bloomFilter:BloomFilter | Returns true if the bloom filters are equal (same pattern of 1s and 0s), false otherwise.| | .falsePositiveRate() | None | Returns Number false positive rate 0.0 <= fpr <= 1.0. | | .calculateBitIndices(key) | key:String | Returns an array of indices {0 <= index < this.bits} which need to be set. |

SafeBloomFilter

Extends BloomFilter, implements a 'safe' version automatically sized to provide the desired false positive rate over the expected number of inserts. After the expected number of inserts has been passed, attempted inserts will throw errors.

const bloom = new SafeBloomFilter(10000, 0.05);
bloom.contains('cheese'); // false
bloom.add('cheese');
bloom.contains('cheese'); // true

| Method | Parameters | Return | | ----------- | -------- | ------ | | .estimateNumberBits(expectedInserts, falsePositiveRate) | expectedInserts:Number, falsePositiveRate:Number | Returns Number, number of bits this filter requires for safe operation. | | .optimalNumHashFunctions(expectedInserts, bits) | expectedInserts:Number, bits:Number | Returns Number, number of Hash functions this filter requires. | | .add(key) | key:String | Returns BloomFilter for chaining. |

Full Documentation

BloomFilter

const bloom = new BloomFilter(42, 4);
bloom.contains('cheese'); // false
bloom.add('cheese');
bloom.contains('cheese'); // true

add

bloomFilter.add(key)

Adds the given key to the filter and increments the number of inserts.

Parameters

• key -> An item to add to the filter.

Returns

Returns BloomFilter for chaining.

Example

bloomFilter.add('cheese');

contains

bloomFilter.contains(key)

Tests whether the key is stored in the filter.

Parameters

• key -> The item to be tested for filter membership.

Returns

Returns true if the item is within the filter, false otherwise.

Example

bloomFilter.contains('cheese');

equals

bloomFilter.equals(bloomFilter)

Tests whether the key is stored in the filter.

Parameters

• bloomFilter -> A bloom filter instance.

Returns

Returns true if the bloom filters are equal (same pattern of 1s and 0s), false otherwise.

Example

bloomFilter.equals(otherBloom);

falsePositiveRate

bloomFilter.falsePositiveRate() Provides an estimate for the false positive rate with the current inserted elements. This will most likely be lower than the expected false positive rate when the filter is not near the capacity but will trend towards 100% as it fills up.

probFalsePositive = (s / m) ^ k

s - Number of Bits Set.

m - Number of Bits in the Filter

k - Number of Hash Functions used.

http://ws680.nist.gov/publication/get_pdf.cfm?pub_id=903775 http://cglab.ca/~morin/publications/ds/bloom-submitted.pdf

Parameters

• None

Returns

Returns Number false positive rate 0.0 <= fpr <= 1.0.

Example

bloomFilter.falsePositiveRate();

calculateBitIndices

bloomFilter.calculateBitIndices(key)

Calculate the indices at which we set the bits to 1 in the bit array. https://willwhim.wordpress.com/2011/09/03/producing-n-hash-functions-by-hashing-only-once/

Parameters

• key -> Item for which we calculate the bits to set.

Returns

Returns an array of indices {0 <= index < this.bits} which need to be set.

Example

bloomFilter.equals(otherBloom);

SafeBloomFilter

const bloom = new SafeBloomFilter(10000, 0.05);
bloom.contains('cheese'); // false
bloom.add('cheese');
bloom.contains('cheese'); // true

add-safe

bloomFilter.add(key)

Only adds an item to the filter if we are below the capacity of the filter, this avoids increasing the actual error rate of the filter above the desired error rate. Throws an error if there are more than expectedInserts made to the filter.

Parameters

• key -> An item to add to the filter.

Returns

Returns BloomFilter for chaining.

Example

bloomFilter.add('cheese');

estimateNumberBits

SafeBloomFilter.estimateNumberOfBits(expectedInserts, falsePositiveRate)

Estimates the number of bits required to store the given number of elements while maintaining the given false positive rate.

m = - (n Ln P / (Ln 2)^2)

https://en.wikipedia.org/wiki/Bloom_filter

https://stackoverflow.com/questions/658439/how-many-hash-functions-does-my-bloom-filter-need

Parameters

• expectedInserts -> Expected number of inserts that will be made.
• falsePositiveRate -> Desired maximum false positive rate.

Returns

Returns Number, number of bits this filter requires for safe operation.

Example

SafeBloomFilter.estimateNumberBits(5000, 0.02);

optimalNumHashFunctions

SafeBloomFilter.optimalNumHashFunctions(expectedInserts, bits)

Calculates the optimal number of hash functions to minimise the false probability for the given m (size) and n (expectedInserts).

k = (m / n) * ln(2).

https://en.wikipedia.org/wiki/Bloom_filter

https://stackoverflow.com/questions/658439/how-many-hash-functions-does-my-bloom-filter-need

Parameters

• expectedInserts -> Expected number of inserts that will be made.
• bits -> Number of bits used in the filter.

Returns

Returns Number, number of Hash functions this filter requires.

Example

SafeBloomFilter.optimalNumHashFunctions(5000, 250);

License

See LICENSE file.

Resources

• Bloom Filter Overview