@libp2p/utils 5.3.2 → 5.4.0-4ad63bb79

package/src/filters/bloom-filter.ts

@@ -0,0 +1,142 @@
+// ported from xxbloom - https://github.com/ceejbot/xxbloom/blob/master/LICENSE
+import { randomBytes } from '@libp2p/crypto'
+import mur from 'murmurhash3js-revisited'
+import { Uint8ArrayList } from 'uint8arraylist'
+import { alloc } from 'uint8arrays/alloc'
+import { fromString as uint8ArrayFromString } from 'uint8arrays/from-string'
+import type { Filter } from './index.js'
+
+const LN2_SQUARED = Math.LN2 * Math.LN2
+
+export interface BloomFilterOptions {
+  seeds?: number[]
+  hashes?: number
+  bits?: number
+}
+
+export class BloomFilter implements Filter {
+  public readonly seeds: number[]
+  public readonly bits: number
+  public buffer: Uint8Array
+
+  constructor (options: BloomFilterOptions = {}) {
+    if (options.seeds != null) {
+      this.seeds = options.seeds
+    } else {
+      this.seeds = generateSeeds(options.hashes ?? 8)
+    }
+
+    this.bits = options.bits ?? 1024
+    this.buffer = alloc(Math.ceil(this.bits / 8))
+  }
+
+  /**
+   * Add an item to the filter
+   */
+  add (item: Uint8Array | string): void {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    for (let i = 0; i < this.seeds.length; i++) {
+      const hash = mur.x86.hash32(item, this.seeds[i])
+      const bit = hash % this.bits
+
+      this.setbit(bit)
+    }
+  }
+
+  /**
+   * Test if the filter has an item. If it returns false it definitely does not
+   * have the item. If it returns true, it probably has the item but there's
+   * an `errorRate` chance it doesn't.
+   */
+  has (item: Uint8Array | string): boolean {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    for (let i = 0; i < this.seeds.length; i++) {
+      const hash = mur.x86.hash32(item, this.seeds[i])
+      const bit = hash % this.bits
+
+      const isSet = this.getbit(bit)
+
+      if (!isSet) {
+        return false
+      }
+    }
+
+    return true
+  }
+
+  /**
+   * Reset the filter
+   */
+  clear (): void {
+    this.buffer.fill(0)
+  }
+
+  setbit (bit: number): void {
+    let pos = 0
+    let shift = bit
+    while (shift > 7) {
+      pos++
+      shift -= 8
+    }
+
+    let bitfield = this.buffer[pos]
+    bitfield |= (0x1 << shift)
+    this.buffer[pos] = bitfield
+  }
+
+  getbit (bit: number): boolean {
+    let pos = 0
+    let shift = bit
+    while (shift > 7) {
+      pos++
+      shift -= 8
+    }
+
+    const bitfield = this.buffer[pos]
+    return (bitfield & (0x1 << shift)) !== 0
+  }
+}
+
+/**
+ * Create a `BloomFilter` with the smallest `bits` and `hashes` value for the
+ * specified item count and error rate.
+ */
+export function createBloomFilter (itemcount: number, errorRate: number = 0.005): Filter {
+  const opts = optimize(itemcount, errorRate)
+  return new BloomFilter(opts)
+}
+
+function optimize (itemCount: number, errorRate: number = 0.005): { bits: number, hashes: number } {
+  const bits = Math.round(-1 * itemCount * Math.log(errorRate) / LN2_SQUARED)
+  const hashes = Math.round((bits / itemCount) * Math.LN2)
+
+  return { bits, hashes }
+}
+
+function generateSeeds (count: number): number[] {
+  let buf: Uint8ArrayList
+  let j: number
+  const seeds = []
+
+  for (let i = 0; i < count; i++) {
+    buf = new Uint8ArrayList(randomBytes(4))
+    seeds[i] = buf.getUint32(0, true)
+
+    // Make sure we don't end up with two identical seeds,
+    // which is unlikely but possible.
+    for (j = 0; j < i; j++) {
+      if (seeds[i] === seeds[j]) {
+        i--
+        break
+      }
+    }
+  }
+
+  return seeds
+}

package/src/filters/bucket.ts

@@ -0,0 +1,64 @@
+import { Fingerprint } from './fingerprint.js'
+import { getRandomInt } from './utils.js'
+
+export class Bucket {
+  private readonly contents: Array<Fingerprint | null>
+
+  constructor (size: number) {
+    this.contents = new Array(size).fill(null)
+  }
+
+  has (fingerprint: Fingerprint): boolean {
+    if (!(fingerprint instanceof Fingerprint)) {
+      throw new TypeError('Invalid Fingerprint')
+    }
+
+    return this.contents.some((fp) => {
+      return fingerprint.equals(fp)
+    })
+  }
+
+  add (fingerprint: Fingerprint): boolean {
+    if (!(fingerprint instanceof Fingerprint)) {
+      throw new TypeError('Invalid Fingerprint')
+    }
+
+    for (let i = 0; i < this.contents.length; i++) {
+      if (this.contents[i] == null) {
+        this.contents[i] = fingerprint
+        return true
+      }
+    }
+
+    return true
+  }
+
+  swap (fingerprint: Fingerprint): Fingerprint | null {
+    if (!(fingerprint instanceof Fingerprint)) {
+      throw new TypeError('Invalid Fingerprint')
+    }
+
+    const i = getRandomInt(0, this.contents.length - 1)
+    const current = this.contents[i]
+    this.contents[i] = fingerprint
+
+    return current
+  }
+
+  remove (fingerprint: Fingerprint): boolean {
+    if (!(fingerprint instanceof Fingerprint)) {
+      throw new TypeError('Invalid Fingerprint')
+    }
+
+    const found = this.contents.findIndex((fp) => {
+      return fingerprint.equals(fp)
+    })
+
+    if (found > -1) {
+      this.contents[found] = null
+      return true
+    } else {
+      return false
+    }
+  }
+}

package/src/filters/cuckoo-filter.ts

@@ -0,0 +1,197 @@
+import { fromString as uint8ArrayFromString } from 'uint8arrays/from-string'
+import { Bucket } from './bucket.js'
+import { Fingerprint, MAX_FINGERPRINT_SIZE } from './fingerprint.js'
+import { fnv1a, type Hash } from './hashes.js'
+import { getRandomInt } from './utils.js'
+import type { Filter } from './index.js'
+
+const maxCuckooCount = 500
+
+export interface CuckooFilterInit {
+  /**
+   * How many items the filter is expected to contain
+   */
+  filterSize: number
+
+  /**
+   * How many items to put in each bucket
+   */
+  bucketSize?: number
+
+  /**
+   * How many bytes the fingerprint is expected to be
+   */
+  fingerprintSize?: number
+
+  /**
+   * A non-cryptographic hash implementation
+   */
+  hash?: Hash
+
+  /**
+   * A number used to seed the hash
+   */
+  seed?: number
+}
+
+export class CuckooFilter implements Filter {
+  private readonly bucketSize: number
+  private readonly filterSize: number
+  private readonly fingerprintSize: number
+  private readonly buckets: Bucket[]
+  public count: number
+  private readonly hash: Hash
+  private readonly seed: number
+
+  constructor (init: CuckooFilterInit) {
+    this.filterSize = init.filterSize
+    this.bucketSize = init.bucketSize ?? 4
+    this.fingerprintSize = init.fingerprintSize ?? 2
+    this.count = 0
+    this.buckets = []
+    this.hash = init.hash ?? fnv1a
+    this.seed = init.seed ?? getRandomInt(0, Math.pow(2, 10))
+  }
+
+  add (item: Uint8Array | string): boolean {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    const fingerprint = new Fingerprint(item, this.hash, this.seed, this.fingerprintSize)
+    const j = this.hash.hash(item, this.seed) % this.filterSize
+    const k = (j ^ fingerprint.hash()) % this.filterSize
+
+    if (this.buckets[j] == null) {
+      this.buckets[j] = new Bucket(this.bucketSize)
+    }
+
+    if (this.buckets[k] == null) {
+      this.buckets[k] = new Bucket(this.bucketSize)
+    }
+
+    if (this.buckets[j].add(fingerprint) || this.buckets[k].add(fingerprint)) {
+      this.count++
+      return true
+    }
+
+    const rand = [j, k]
+    let i = rand[getRandomInt(0, rand.length - 1)]
+
+    if (this.buckets[i] == null) {
+      this.buckets[i] = new Bucket(this.bucketSize)
+    }
+
+    for (let n = 0; n < maxCuckooCount; n++) {
+      const swapped = this.buckets[i].swap(fingerprint)
+
+      if (swapped == null) {
+        continue
+      }
+
+      i = (i ^ swapped.hash()) % this.filterSize
+
+      if (this.buckets[i] == null) {
+        this.buckets[i] = new Bucket(this.bucketSize)
+      }
+
+      if (this.buckets[i].add(swapped)) {
+        this.count++
+
+        return true
+      } else {
+        continue
+      }
+    }
+
+    return false
+  }
+
+  has (item: Uint8Array | string): boolean {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    const fingerprint = new Fingerprint(item, this.hash, this.seed, this.fingerprintSize)
+    const j = this.hash.hash(item, this.seed) % this.filterSize
+    const inJ = this.buckets[j]?.has(fingerprint) ?? false
+
+    if (inJ) {
+      return inJ
+    }
+
+    const k = (j ^ fingerprint.hash()) % this.filterSize
+
+    return this.buckets[k]?.has(fingerprint) ?? false
+  }
+
+  remove (item: Uint8Array | string): boolean {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    const fingerprint = new Fingerprint(item, this.hash, this.seed, this.fingerprintSize)
+    const j = this.hash.hash(item, this.seed) % this.filterSize
+    const inJ = this.buckets[j]?.remove(fingerprint) ?? false
+
+    if (inJ) {
+      this.count--
+      return inJ
+    }
+
+    const k = (j ^ fingerprint.hash()) % this.filterSize
+    const inK = this.buckets[k]?.remove(fingerprint) ?? false
+
+    if (inK) {
+      this.count--
+    }
+
+    return inK
+  }
+
+  get reliable (): boolean {
+    return Math.floor(100 * (this.count / this.filterSize)) <= 95
+  }
+}
+
+// max load constants, defined in the cuckoo paper
+const MAX_LOAD = {
+  1: 0.5,
+  2: 0.84,
+  4: 0.95,
+  8: 0.98
+}
+
+function calculateBucketSize (errorRate: number = 0.001): 2 | 4 | 8 {
+  if (errorRate > 0.002) {
+    return 2
+  }
+
+  if (errorRate > 0.00001) {
+    return 4
+  }
+
+  return 8
+}
+
+export function optimize (maxItems: number, errorRate: number = 0.001): CuckooFilterInit {
+  // https://www.eecs.harvard.edu/~michaelm/postscripts/cuckoo-conext2014.pdf
+  // Section 5.1 Optimal Bucket Size
+  const bucketSize = calculateBucketSize(errorRate)
+  const load = MAX_LOAD[bucketSize]
+
+  // https://stackoverflow.com/questions/57555236/how-to-size-a-cuckoo-filter/57617208#57617208
+  const filterSize = Math.round(maxItems / load)
+  const fingerprintSize = Math.min(Math.ceil(Math.log(filterSize / bucketSize)) + 2, MAX_FINGERPRINT_SIZE)
+
+  return {
+    filterSize,
+    bucketSize,
+    fingerprintSize
+  }
+}
+
+export function createCuckooFilter (maxItems: number, errorRate: number = 0.005): Filter {
+  const opts = optimize(maxItems, errorRate)
+  return new CuckooFilter(opts)
+}

package/src/filters/fingerprint.ts

@@ -0,0 +1,44 @@
+import { alloc as uint8ArrayAlloc } from 'uint8arrays/alloc'
+import { equals as uint8ArrayEquals } from 'uint8arrays/equals'
+import type { Hash } from './hashes'
+
+export const MAX_FINGERPRINT_SIZE = 64
+
+export class Fingerprint {
+  private readonly fp: Uint8Array
+  private readonly h: Hash
+  private readonly seed: number
+
+  constructor (buf: Uint8Array, hash: Hash, seed: number, fingerprintSize: number = 2) {
+    if (fingerprintSize > MAX_FINGERPRINT_SIZE) {
+      throw new TypeError('Invalid Fingerprint Size')
+    }
+
+    const fnv = hash.hashV(buf, seed)
+    const fp = uint8ArrayAlloc(fingerprintSize)
+
+    for (let i = 0; i < fp.length; i++) {
+      fp[i] = fnv[i]
+    }
+
+    if (fp.length === 0) {
+      fp[0] = 7
+    }
+
+    this.fp = fp
+    this.h = hash
+    this.seed = seed
+  }
+
+  hash (): number {
+    return this.h.hash(this.fp, this.seed)
+  }
+
+  equals (other?: any): boolean {
+    if (!(other?.fp instanceof Uint8Array)) {
+      return false
+    }
+
+    return uint8ArrayEquals(this.fp, other.fp)
+  }
+}

package/src/filters/hashes.ts

@@ -0,0 +1,38 @@
+import fnv1aHash from '@sindresorhus/fnv1a'
+import mur from 'murmurhash3js-revisited'
+import { fromString as uint8ArrayFromString } from 'uint8arrays/from-string'
+
+export interface Hash {
+  hash(input: Uint8Array, seed: number): number
+  hashV(input: Uint8Array, seed: number): Uint8Array
+}
+
+export const murmur3: Hash = {
+  hash: (input, seed) => {
+    return mur.x86.hash32(input, seed)
+  },
+  hashV: (input, seed) => {
+    return numberToBuffer(murmur3.hash(input, seed))
+  }
+}
+
+export const fnv1a: Hash = {
+  hash: (input) => {
+    return Number(fnv1aHash(input, {
+      size: 32
+    }))
+  },
+  hashV: (input, seed) => {
+    return numberToBuffer(fnv1a.hash(input, seed))
+  }
+}
+
+export function numberToBuffer (num: bigint | number): Uint8Array {
+  let hex = num.toString(16)
+
+  if (hex.length % 2 === 1) {
+    hex = `0${hex}`
+  }
+
+  return uint8ArrayFromString(hex, 'base16')
+}

package/src/filters/index.ts

@@ -0,0 +1,9 @@
+export { BloomFilter, createBloomFilter, type BloomFilterOptions } from './bloom-filter.js'
+export { CuckooFilter, createCuckooFilter, type CuckooFilterInit } from './cuckoo-filter.js'
+export { ScalableCuckooFilter, createScalableCuckooFilter, type ScalableCuckooFilterInit } from './scalable-cuckoo-filter.js'
+
+export interface Filter {
+  add(item: Uint8Array | string): void
+  has(item: Uint8Array | string): boolean
+  remove?(buf: Uint8Array | string): boolean
+}

package/src/filters/scalable-cuckoo-filter.ts

@@ -0,0 +1,111 @@
+import { fromString as uint8ArrayFromString } from 'uint8arrays/from-string'
+import { CuckooFilter, optimize, type CuckooFilterInit } from './cuckoo-filter.js'
+import { fnv1a, type Hash } from './hashes.js'
+import { getRandomInt } from './utils.js'
+import type { Filter } from './index.js'
+
+export interface ScalableCuckooFilterInit extends CuckooFilterInit {
+  /**
+   * A number to multiply maxItems by when adding new sub-filters
+   */
+  scale?: number
+}
+
+export class ScalableCuckooFilter implements Filter {
+  private readonly filterSize: number
+  private readonly bucketSize: number
+  private readonly fingerprintSize: number
+  private readonly scale: number
+  private readonly filterSeries: CuckooFilter[]
+  private readonly hash: Hash
+  private readonly seed: number
+
+  constructor (init: ScalableCuckooFilterInit) {
+    this.bucketSize = init.bucketSize ?? 4
+    this.filterSize = init.filterSize ?? (1 << 18) / this.bucketSize
+    this.fingerprintSize = init.fingerprintSize ?? 2
+    this.scale = init.scale ?? 2
+    this.hash = init.hash ?? fnv1a
+    this.seed = init.seed ?? getRandomInt(0, Math.pow(2, 10))
+    this.filterSeries = [
+      new CuckooFilter({
+        filterSize: this.filterSize,
+        bucketSize: this.bucketSize,
+        fingerprintSize: this.fingerprintSize,
+        hash: this.hash,
+        seed: this.seed
+      })
+    ]
+  }
+
+  add (item: Uint8Array | string): boolean {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    if (this.has(item)) {
+      return true
+    }
+
+    let current = this.filterSeries.find((cuckoo) => {
+      return cuckoo.reliable
+    })
+
+    if (current == null) {
+      const curSize = this.filterSize * Math.pow(this.scale, this.filterSeries.length)
+
+      current = new CuckooFilter({
+        filterSize: curSize,
+        bucketSize: this.bucketSize,
+        fingerprintSize: this.fingerprintSize,
+        hash: this.hash,
+        seed: this.seed
+      })
+
+      this.filterSeries.push(current)
+    }
+
+    return current.add(item)
+  }
+
+  has (item: Uint8Array | string): boolean {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    for (let i = 0; i < this.filterSeries.length; i++) {
+      if (this.filterSeries[i].has(item)) {
+        return true
+      }
+    }
+
+    return false
+  }
+
+  remove (item: Uint8Array | string): boolean {
+    if (typeof item === 'string') {
+      item = uint8ArrayFromString(item)
+    }
+
+    for (let i = 0; i < this.filterSeries.length; i++) {
+      if (this.filterSeries[i].remove(item)) {
+        return true
+      }
+    }
+
+    return false
+  }
+
+  get count (): number {
+    return this.filterSeries.reduce((acc, curr) => {
+      return acc + curr.count
+    }, 0)
+  }
+}
+
+export function createScalableCuckooFilter (maxItems: number, errorRate: number = 0.001, options?: Pick<ScalableCuckooFilterInit, 'hash' | 'seed' | 'scale'>): Filter {
+  return new ScalableCuckooFilter({
+    ...optimize(maxItems, errorRate),
+    ...(options ?? {})
+  })
+}

package/src/filters/utils.ts

@@ -0,0 +1,3 @@
+export function getRandomInt (min: number, max: number): number {
+  return Math.floor(Math.random() * (max - min)) + min
+}

package/src/moving-average.ts

@@ -0,0 +1,45 @@
+/**
+ * Implements exponential moving average. Ported from `moving-average`.
+ *
+ * @see https://en.wikipedia.org/wiki/Moving_average#Exponential_moving_average
+ * @see https://www.npmjs.com/package/moving-average
+ */
+export class MovingAverage {
+  public movingAverage: number
+  public variance: number
+  public deviation: number
+  public forecast: number
+  private readonly timespan: number
+  private previousTime?: number
+
+  constructor (timespan: number) {
+    this.timespan = timespan
+    this.movingAverage = 0
+    this.variance = 0
+    this.deviation = 0
+    this.forecast = 0
+  }
+
+  alpha (t: number, pt: number): number {
+    return 1 - (Math.exp(-(t - pt) / this.timespan))
+  }
+
+  push (value: number, time: number = Date.now()): void {
+    if (this.previousTime != null) {
+      // calculate moving average
+      const a = this.alpha(time, this.previousTime)
+      const diff = value - this.movingAverage
+      const incr = a * diff
+      this.movingAverage = a * value + (1 - a) * this.movingAverage
+      // calculate variance & deviation
+      this.variance = (1 - a) * (this.variance + diff * incr)
+      this.deviation = Math.sqrt(this.variance)
+      // calculate forecast
+      this.forecast = this.movingAverage + a * diff
+    } else {
+      this.movingAverage = value
+    }
+
+    this.previousTime = time
+  }
+}