Convert MurmurHash3_64 to an ES6 class

Notable changes:
 - Remove the `return this;` from the `MurmurHash3_64.update` method, since it's completely unused and doesn't make a lot of sense.
 - Remove the loop(s) from the `MurmurHash3_64.hexdigest` method, since creating a temporary array and then looping over it is wasteful given how simple this can be written with modern JavaScript.
This commit is contained in:
Jonas Jenwald 2019-03-09 16:50:50 +01:00
parent e41c4aece4
commit 6b1ac44aea

View File

@ -16,88 +16,85 @@
* Based on https://code.google.com/p/smhasher/wiki/MurmurHash3.
* Hashes roughly 100 KB per millisecond on i7 3.4 GHz.
*/
/* eslint no-var: error */
import { isArrayBuffer, isString } from '../shared/util';
var MurmurHash3_64 = (function MurmurHash3_64Closure(seed) {
// Workaround for missing math precision in JS.
var MASK_HIGH = 0xffff0000;
var MASK_LOW = 0xffff;
const SEED = 0xc3d2e1f0;
// Workaround for missing math precision in JS.
const MASK_HIGH = 0xffff0000;
const MASK_LOW = 0xffff;
function MurmurHash3_64(seed) {
var SEED = 0xc3d2e1f0;
class MurmurHash3_64 {
constructor(seed) {
this.h1 = seed ? seed & 0xffffffff : SEED;
this.h2 = seed ? seed & 0xffffffff : SEED;
}
MurmurHash3_64.prototype = {
update: function MurmurHash3_64_update(input) {
let data, length;
if (isString(input)) {
data = new Uint8Array(input.length * 2);
length = 0;
for (let i = 0, ii = input.length; i < ii; i++) {
var code = input.charCodeAt(i);
if (code <= 0xff) {
data[length++] = code;
} else {
data[length++] = code >>> 8;
data[length++] = code & 0xff;
}
}
} else if (isArrayBuffer(input)) {
data = input;
length = data.byteLength;
} else {
throw new Error('Wrong data format in MurmurHash3_64_update. ' +
'Input must be a string or array.');
}
var blockCounts = length >> 2;
var tailLength = length - blockCounts * 4;
// we don't care about endianness here
var dataUint32 = new Uint32Array(data.buffer, 0, blockCounts);
var k1 = 0;
var k2 = 0;
var h1 = this.h1;
var h2 = this.h2;
var C1 = 0xcc9e2d51;
var C2 = 0x1b873593;
var C1_LOW = C1 & MASK_LOW;
var C2_LOW = C2 & MASK_LOW;
for (let i = 0; i < blockCounts; i++) {
if (i & 1) {
k1 = dataUint32[i];
k1 = (k1 * C1 & MASK_HIGH) | (k1 * C1_LOW & MASK_LOW);
k1 = k1 << 15 | k1 >>> 17;
k1 = (k1 * C2 & MASK_HIGH) | (k1 * C2_LOW & MASK_LOW);
h1 ^= k1;
h1 = h1 << 13 | h1 >>> 19;
h1 = h1 * 5 + 0xe6546b64;
update(input) {
let data, length;
if (isString(input)) {
data = new Uint8Array(input.length * 2);
length = 0;
for (let i = 0, ii = input.length; i < ii; i++) {
const code = input.charCodeAt(i);
if (code <= 0xff) {
data[length++] = code;
} else {
k2 = dataUint32[i];
k2 = (k2 * C1 & MASK_HIGH) | (k2 * C1_LOW & MASK_LOW);
k2 = k2 << 15 | k2 >>> 17;
k2 = (k2 * C2 & MASK_HIGH) | (k2 * C2_LOW & MASK_LOW);
h2 ^= k2;
h2 = h2 << 13 | h2 >>> 19;
h2 = h2 * 5 + 0xe6546b64;
data[length++] = code >>> 8;
data[length++] = code & 0xff;
}
}
} else if (isArrayBuffer(input)) {
data = input;
length = data.byteLength;
} else {
throw new Error('Wrong data format in MurmurHash3_64_update. ' +
'Input must be a string or array.');
}
k1 = 0;
const blockCounts = length >> 2;
const tailLength = length - blockCounts * 4;
// We don't care about endianness here.
const dataUint32 = new Uint32Array(data.buffer, 0, blockCounts);
let k1 = 0, k2 = 0;
let h1 = this.h1, h2 = this.h2;
const C1 = 0xcc9e2d51, C2 = 0x1b873593;
const C1_LOW = C1 & MASK_LOW, C2_LOW = C2 & MASK_LOW;
for (let i = 0; i < blockCounts; i++) {
if (i & 1) {
k1 = dataUint32[i];
k1 = (k1 * C1 & MASK_HIGH) | (k1 * C1_LOW & MASK_LOW);
k1 = k1 << 15 | k1 >>> 17;
k1 = (k1 * C2 & MASK_HIGH) | (k1 * C2_LOW & MASK_LOW);
h1 ^= k1;
h1 = h1 << 13 | h1 >>> 19;
h1 = h1 * 5 + 0xe6546b64;
} else {
k2 = dataUint32[i];
k2 = (k2 * C1 & MASK_HIGH) | (k2 * C1_LOW & MASK_LOW);
k2 = k2 << 15 | k2 >>> 17;
k2 = (k2 * C2 & MASK_HIGH) | (k2 * C2_LOW & MASK_LOW);
h2 ^= k2;
h2 = h2 << 13 | h2 >>> 19;
h2 = h2 * 5 + 0xe6546b64;
}
}
k1 = 0;
switch (tailLength) {
case 3:
k1 ^= data[blockCounts * 4 + 2] << 16;
/* falls through */
case 2:
k1 ^= data[blockCounts * 4 + 1] << 8;
/* falls through */
case 1:
k1 ^= data[blockCounts * 4];
/* falls through */
switch (tailLength) {
case 3:
k1 ^= data[blockCounts * 4 + 2] << 16;
/* falls through */
case 2:
k1 ^= data[blockCounts * 4 + 1] << 8;
/* falls through */
case 1:
k1 ^= data[blockCounts * 4];
/* falls through */
k1 = (k1 * C1 & MASK_HIGH) | (k1 * C1_LOW & MASK_LOW);
k1 = k1 << 15 | k1 >>> 17;
k1 = (k1 * C2 & MASK_HIGH) | (k1 * C2_LOW & MASK_LOW);
@ -106,41 +103,29 @@ var MurmurHash3_64 = (function MurmurHash3_64Closure(seed) {
} else {
h2 ^= k1;
}
}
}
this.h1 = h1;
this.h2 = h2;
return this;
},
this.h1 = h1;
this.h2 = h2;
}
hexdigest: function MurmurHash3_64_hexdigest() {
var h1 = this.h1;
var h2 = this.h2;
hexdigest() {
let h1 = this.h1, h2 = this.h2;
h1 ^= h2 >>> 1;
h1 = (h1 * 0xed558ccd & MASK_HIGH) | (h1 * 0x8ccd & MASK_LOW);
h2 = (h2 * 0xff51afd7 & MASK_HIGH) |
(((h2 << 16 | h1 >>> 16) * 0xafd7ed55 & MASK_HIGH) >>> 16);
h1 ^= h2 >>> 1;
h1 = (h1 * 0x1a85ec53 & MASK_HIGH) | (h1 * 0xec53 & MASK_LOW);
h2 = (h2 * 0xc4ceb9fe & MASK_HIGH) |
(((h2 << 16 | h1 >>> 16) * 0xb9fe1a85 & MASK_HIGH) >>> 16);
h1 ^= h2 >>> 1;
h1 ^= h2 >>> 1;
h1 = (h1 * 0xed558ccd & MASK_HIGH) | (h1 * 0x8ccd & MASK_LOW);
h2 = (h2 * 0xff51afd7 & MASK_HIGH) |
(((h2 << 16 | h1 >>> 16) * 0xafd7ed55 & MASK_HIGH) >>> 16);
h1 ^= h2 >>> 1;
h1 = (h1 * 0x1a85ec53 & MASK_HIGH) | (h1 * 0xec53 & MASK_LOW);
h2 = (h2 * 0xc4ceb9fe & MASK_HIGH) |
(((h2 << 16 | h1 >>> 16) * 0xb9fe1a85 & MASK_HIGH) >>> 16);
h1 ^= h2 >>> 1;
for (var i = 0, arr = [h1, h2], str = ''; i < arr.length; i++) {
var hex = (arr[i] >>> 0).toString(16);
while (hex.length < 8) {
hex = '0' + hex;
}
str += hex;
}
return str;
},
};
return MurmurHash3_64;
})();
const hex1 = (h1 >>> 0).toString(16), hex2 = (h2 >>> 0).toString(16);
return hex1.padStart(8, '0') + hex2.padStart(8, '0');
}
}
export {
MurmurHash3_64,