这里是目前解决方案的一个基准，参考纳米基准

import { v4 as uuid4 } from 'uuid'
import benchmark from 'benchmark'
import shortid from 'shortid'

let suite = new benchmark.Suite()

suite
  .add('crypto.randomUUID', () => {
    crypto.randomUUID()
  })
  .add('nanoid', () => {
    nanoid()
  })
  .add('uuid v4', () => {
    uuid4()
  })
  .add("math.random", () => {
    (new Date()).getTime().toString(36) + Math.random().toString(36).slice(2)
  })
  .add('crypto.randomBytes', () => {
    crypto.randomBytes(32).toString('hex')
  })
  .add('shortid', () => {
    shortid()
  })
  .on('cycle', event => {
    let name = event.target.name
    let hz = formatNumber(event.target.hz.toFixed(0)).padStart(10)

    process.stdout.write(`${name}${pico.bold(hz)}${pico.dim(' ops/sec')}\n`)
  })
  .run()

结果是

node ./test/benchmark.js
crypto.randomUUID         13,281,440 ops/sec
nanoid                     3,278,757 ops/sec
uuid v4                    1,117,140 ops/sec
math.random                1,206,105 ops/sec
crypto.randomBytes           280,199 ops/sec
shortid                       30,728 ops/sec

测试env:

2.6 GHz 6Cores Intel酷睿i7 MacOS 节点v16.17.0

节点uuid已弃用，请使用uuid

npm install uuid --save
// Generate a v1 UUID (time-based) 
const uuidV1 = require('uuid/v1');
uuidV1(); // -> '6c84fb90-12c4-11e1-840d-7b25c5ee775a' 

// Generate a v4 UUID (random) 
const uuidV4 = require('uuid/v4');
uuidV4(); // -> '110ec58a-a0f2-4ac4-8393-c866d813b8d1' 

Npm链接

从YaroslavGaponov的答案延伸出来，最简单的实现就是使用Math.random()。

Math.random()

从数学上讲，分数在实空间[0,1]中相同的概率理论上为0。在node.js中，它的默认长度为16个小数，概率接近于0。这种实现还应该减少算术溢出，因为不执行任何操作。此外，与字符串相比，它的内存效率更高，因为小数占用的内存比字符串少。

我称之为“分数-唯一id”。

编写代码生成1,000,000个Math.random()数字，但无法找到任何重复的数字(至少对于默认的小数点为16)。参见下面的代码(如果有请提供反馈):

random_numbers = [] 
for (i = 0; i < 1000000; i++) { 
   random_numbers.push(Math.random()); 
   //random_numbers.push(Math.random().toFixed(13)) //depends decimals default 16 
} 

if (i === 1000000) { 
   console.log("Before checking duplicate"); 
   console.log(random_numbers.length); 
   console.log("After checking duplicate"); 
   random_set = new Set(random_numbers); // Set removes duplicates
   console.log([...random_set].length); // length is still the same after removing
} 

从14.17.0开始，你现在可以使用内置的加密模块来生成uuid (UUIDv4 flavour):

const { randomUUID } = require('crypto'); // Added in: node v14.17.0

console.log(randomUUID());

// '89rct5ac2-8493-49b0-95d8-de843d90e6ca'

你可以探索更多 https://nodejs.org/api/crypto.html#crypto_crypto_randomuuid_options

注意:密码。randomUUID比uuid快三倍。而且不需要增加额外的依赖。

对我来说，获取唯一id的最简单方法是时间，我在这个例子中使用哈希

const hash = require('object-hash');
const user = { ..., iat: Date.now() }
const user_id = hash(user)

console.log(user_id) // ex. 49686bab1a2276e0b1bd61ccc86f8156

这就是我在任何方面获得唯一标识符的方法。因为在现实世界中，时间是不会重复的。

如果有人需要加密强UUID，也有解决方案。

https://www.npmjs.com/package/generate-safe-id

npm install generate-safe-id

Why not UUIDs? Random UUIDs (UUIDv4) do not have enough entropy to be universally unique (ironic, eh?). Random UUIDs have only 122 bits of entropy, which suggests that a duplicate will occur after only 2^61 IDs. Additionally, some UUIDv4 implementations do not use a cryptographically strong random number generator. This library generates 240-bit IDs using the Node.js crypto RNG, suggesting the first duplicate will occur after generating 2^120 IDs. Based on the current energy production of the human race, this threshold will be impossible to cross for the foreseeable future.

var generateSafeId = require('generate-safe-id');

var id = generateSafeId();
// id == "zVPkWyvgRW-7pSk0iRzEhdnPcnWfMRi-ZcaPxrHA"