commonMain.com.goncalossilva.murmurhash.MurmurHash3.kt Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of SymSpellKt-jvm Show documentation
Show all versions of SymSpellKt-jvm Show documentation
A Kotlin Multiplatform implementation of the SymSpell Spell Checking algorithm.
package com.goncalossilva.murmurhash
// Copied from https://github.com/goncalossilva/kotlinx-murmurhash
public class MurmurHash3(private val seed: UInt = 0u) {
public fun hash32x86(key: ByteArray): UInt {
var h = seed
val len = key.size
val nblocks = len / 4
for (i in 0 until nblocks * 4 step 4) {
val k = key.getLittleEndianUInt(i)
h = h xor k.mix(R1_32, C1_32, C2_32)
h = h.rotateLeft(R2_32)
h = h * M_32 + N_32
}
val index = nblocks * 4
val rem = len - index
var k = 0u
if (rem == 3) {
k = k xor (key.getUInt(index + 2) shl 16)
}
if (rem >= 2) {
k = k xor (key.getUInt(index + 1) shl 8)
}
if (rem >= 1) {
k = k xor key.getUInt(index)
h = h xor k.mix(R1_32, C1_32, C2_32)
}
h = h xor len.toUInt()
h = h.fmix()
return h
}
public fun hash128x86(key: ByteArray): Array {
var h1 = seed
var h2 = seed
var h3 = seed
var h4 = seed
val len = key.size
val nblocks = len / 16
for (i in 0 until nblocks * 16 step 16) {
val k1 = key.getLittleEndianUInt(i)
val k2 = key.getLittleEndianUInt(i + 4)
val k3 = key.getLittleEndianUInt(i + 8)
val k4 = key.getLittleEndianUInt(i + 12)
h1 = h1 xor k1.mix(R1_128x86, C1_128x86, C2_128x86)
h1 = h1.rotateLeft(R5_128x86)
h1 += h2
h1 = h1 * M_128x86 + N1_128x86
h2 = h2 xor k2.mix(R2_128x86, C2_128x86, C3_128x86)
h2 = h2.rotateLeft(R3_128x86)
h2 += h3
h2 = h2 * M_128x86 + N2_128x86
h3 = h3 xor k3.mix(R3_128x86, C3_128x86, C4_128x86)
h3 = h3.rotateLeft(R1_128x86)
h3 += h4
h3 = h3 * M_128x86 + N3_128x86
h4 = h4 xor k4.mix(R4_128x86, C4_128x86, C1_128x86)
h4 = h4.rotateLeft(R6_128x86)
h4 += h1
h4 = h4 * M_128x86 + N4_128x86
}
val index = nblocks * 16
val rem = len - index
var k1 = 0u
var k2 = 0u
var k3 = 0u
var k4 = 0u
if (rem == 15) {
k4 = k4 xor (key.getUInt(index + 14) shl 16)
}
if (rem >= 14) {
k4 = k4 xor (key.getUInt(index + 13) shl 8)
}
if (rem >= 13) {
k4 = k4 xor key.getUInt(index + 12)
h4 = h4 xor k4.mix(R4_128x86, C4_128x86, C1_128x86)
}
if (rem >= 12) {
k3 = k3 xor (key.getUInt(index + 11) shl 24)
}
if (rem >= 11) {
k3 = k3 xor (key.getUInt(index + 10) shl 16)
}
if (rem >= 10) {
k3 = k3 xor (key.getUInt(index + 9) shl 8)
}
if (rem >= 9) {
k3 = k3 xor key.getUInt(index + 8)
h3 = h3 xor k3.mix(R3_128x86, C3_128x86, C4_128x86)
}
if (rem >= 8) {
k2 = k2 xor (key.getUInt(index + 7) shl 24)
}
if (rem >= 7) {
k2 = k2 xor (key.getUInt(index + 6) shl 16)
}
if (rem >= 6) {
k2 = k2 xor (key.getUInt(index + 5) shl 8)
}
if (rem >= 5) {
k2 = k2 xor key.getUInt(index + 4)
h2 = h2 xor k2.mix(R2_128x86, C2_128x86, C3_128x86)
}
if (rem >= 4) {
k1 = k1 xor (key.getUInt(index + 3) shl 24)
}
if (rem >= 3) {
k1 = k1 xor (key.getUInt(index + 2) shl 16)
}
if (rem >= 2) {
k1 = k1 xor (key.getUInt(index + 1) shl 8)
}
if (rem >= 1) {
k1 = k1 xor key.getUInt(index)
h1 = h1 xor k1.mix(R1_128x86, C1_128x86, C2_128x86)
}
h1 = h1 xor len.toUInt()
h2 = h2 xor len.toUInt()
h3 = h3 xor len.toUInt()
h4 = h4 xor len.toUInt()
h1 += h2
h1 += h3
h1 += h4
h2 += h1
h3 += h1
h4 += h1
h1 = h1.fmix()
h2 = h2.fmix()
h3 = h3.fmix()
h4 = h4.fmix()
h1 += h2
h1 += h3
h1 += h4
h2 += h1
h3 += h1
h4 += h1
return arrayOf(h1, h2, h3, h4)
}
public fun hash128x64(key: ByteArray): Array {
var h1 = seed.toULong()
var h2 = seed.toULong()
val len = key.size
val nblocks = len / 16
for (i in 0 until nblocks * 16 step 16) {
val k1 = key.getLittleEndianLong(i)
val k2 = key.getLittleEndianLong(i + 8)
h1 = h1 xor k1.mix(R1_128x64, C1_128x64, C2_128x64)
h1 = h1.rotateLeft(R2_128x64)
h1 += h2
h1 = h1 * M_128x64 + N1_128x64
h2 = h2 xor k2.mix(R3_128x64, C2_128x64, C1_128x64)
h2 = h2.rotateLeft(R1_128x64)
h2 += h1
h2 = h2 * M_128x64 + N2_128x64
}
val index = nblocks * 16
val rem = len - index
var k1 = 0uL
var k2 = 0uL
if (rem == 15) {
k2 = k2 xor (key.getULong(index + 14) shl 48)
}
if (rem >= 14) {
k2 = k2 xor (key.getULong(index + 13) shl 40)
}
if (rem >= 13) {
k2 = k2 xor (key.getULong(index + 12) shl 32)
}
if (rem >= 12) {
k2 = k2 xor (key.getULong(index + 11) shl 24)
}
if (rem >= 11) {
k2 = k2 xor (key.getULong(index + 10) shl 16)
}
if (rem >= 10) {
k2 = k2 xor (key.getULong(index + 9) shl 8)
}
if (rem >= 9) {
k2 = k2 xor key.getULong(index + 8)
h2 = h2 xor k2.mix(R3_128x64, C2_128x64, C1_128x64)
}
if (rem >= 8) {
k1 = k1 xor (key.getULong(index + 7) shl 56)
}
if (rem >= 7) {
k1 = k1 xor (key.getULong(index + 6) shl 48)
}
if (rem >= 6) {
k1 = k1 xor (key.getULong(index + 5) shl 40)
}
if (rem >= 5) {
k1 = k1 xor (key.getULong(index + 4) shl 32)
}
if (rem >= 4) {
k1 = k1 xor (key.getULong(index + 3) shl 24)
}
if (rem >= 3) {
k1 = k1 xor (key.getULong(index + 2) shl 16)
}
if (rem >= 2) {
k1 = k1 xor (key.getULong(index + 1) shl 8)
}
if (rem >= 1) {
k1 = k1 xor key.getULong(index)
h1 = h1 xor k1.mix(R1_128x64, C1_128x64, C2_128x64)
}
h1 = h1 xor len.toULong()
h2 = h2 xor len.toULong()
h1 += h2
h2 += h1
h1 = h1.fmix()
h2 = h2.fmix()
h1 += h2
h2 += h1
return arrayOf(h1, h2)
}
private fun ByteArray.getLittleEndianUInt(index: Int): UInt {
return this.getUInt(index) or
(this.getUInt(index + 1) shl 8) or
(this.getUInt(index + 2) shl 16) or
(this.getUInt(index + 3) shl 24)
}
private fun ByteArray.getLittleEndianLong(index: Int): ULong {
return this.getULong(index) or
(this.getULong(index + 1) shl 8) or
(this.getULong(index + 2) shl 16) or
(this.getULong(index + 3) shl 24) or
(this.getULong(index + 4) shl 32) or
(this.getULong(index + 5) shl 40) or
(this.getULong(index + 6) shl 48) or
(this.getULong(index + 7) shl 56)
}
private fun UInt.mix(r: Int, c1: UInt, c2: UInt): UInt {
var k = this
k *= c1
k = k.rotateLeft(r)
k *= c2
return k
}
private fun ULong.mix(r: Int, c1: ULong, c2: ULong): ULong {
var k = this
k *= c1
k = k.rotateLeft(r)
k *= c2
return k
}
private fun UInt.fmix(): UInt {
var h = this
h = h xor (h shr 16)
h *= 0x85ebca6bu
h = h xor (h shr 13)
h *= 0xc2b2ae35u
h = h xor (h shr 16)
return h
}
private fun ULong.fmix(): ULong {
var h = this
h = h xor (h shr 33)
h *= 0xff51afd7ed558ccduL
h = h xor (h shr 33)
h *= 0xc4ceb9fe1a85ec53uL
h = h xor (h shr 33)
return h
}
private fun ByteArray.getUInt(index: Int) = get(index).toUByte().toUInt()
private fun ByteArray.getULong(index: Int) = get(index).toUByte().toULong()
private companion object {
private const val C1_32: UInt = 0xcc9e2d51u
private const val C2_32: UInt = 0x1b873593u
private const val R1_32: Int = 15
private const val R2_32: Int = 13
private const val M_32: UInt = 5u
private const val N_32: UInt = 0xe6546b64u
private const val C1_128x86: UInt = 0x239b961bu
private const val C2_128x86: UInt = 0xab0e9789u
private const val C3_128x86: UInt = 0x38b34ae5u
private const val C4_128x86: UInt = 0xa1e38b93u
private const val R1_128x86: Int = 15
private const val R2_128x86: Int = 16
private const val R3_128x86: Int = 17
private const val R4_128x86: Int = 18
private const val R5_128x86: Int = 19
private const val R6_128x86: Int = 13
private const val M_128x86: UInt = 5u
private const val N1_128x86: UInt = 0x561ccd1bu
private const val N2_128x86: UInt = 0x0bcaa747u
private const val N3_128x86: UInt = 0x96cd1c35u
private const val N4_128x86: UInt = 0x32ac3b17u
private const val C1_128x64: ULong = 0x87c37b91114253d5uL
private const val C2_128x64: ULong = 0x4cf5ad432745937fuL
private const val R1_128x64: Int = 31
private const val R2_128x64: Int = 27
private const val R3_128x64: Int = 33
private const val M_128x64: ULong = 5u
private const val N1_128x64: ULong = 0x52dce729u
private const val N2_128x64: ULong = 0x38495ab5u
}
} © 2015 - 2025 Weber Informatics LLC | Privacy Policy