predict-v8-randomness

A huge shout-out to PwnFunction for the inspiration!

predict-v8-randomness uses z3 — a Satisfiability Modulo Theories (SMT) solver developed by Microsoft — to predict the output of Math.random() in V8, the JavaScript engine used by Chrome and Node.js.

Install

npm

npm i predict-v8-randomness

yarn

yarn add predict-v8-randomness

Use with CommonJS or ESM

CJS

const predictV8Randomness = require("predict-v8-randomness");
const predictor = new predictV8Randomness.Predictor(...);
// or
const { Predictor } = require("predict-v8-randomness");
const predictor = new Predictor(...);

ESM

import predictV8Randomness from "predict-v8-randomness";
const predictor = new predictV8Randomness.Predictor(...);
// or
import { Predictor } from "predict-v8-randomness";
const predictor = new Predictor(...);

Usage

For every 10 predictions it takes ~3 seconds to compute, so be mindful of performance when predicting large amounts. Keep in mind, the max we can predict next is 60.

We always return an array of predictions (number[]). If you are only predicting one item you can destructure the return for simplicity.

Dynamically Generated Sequence

// If no parameters are provided, we generate the sequence dynamically
const predictor = new Predictor();

// Predict next Math.random() output. Destructure return.
const [nextRand] = predictor.predictNext();
// is equivalent to:
const [nextRand] = predictor.predictNext(1);
// Validate prediction right now, right here, in real time.
console.log("Accurate?", nextRand === Math.random());

// Predict next 10 Math.random() outputs
const nextTenRand = predictor.predictNext(10);
const actuals = Array.from({ length: 10 }, Math.random);
// Validate those 10
console.log(
  "Accurate?",
  nextTenRand.every((e, i) => actuals[i] === e),
);

Provide Your Own Seed Sequence

The provided sequence must contain EXACTLY 4 numbers. Anything other than 4 and we lose prediction accuracy.

const initialSequence = [
  /* Your sequence here */
  /* HAS to be 4 numbers */
];
const predictor = new Predictor(initialSequence);
const [nextRand] = predictor.predictNext();
// Where N <= 60 && N > 0
const futureN = predictor.predictNext(N);

CLI

You can use the following methods to run as CLI

Method	Instructions	Info
`npx`	`> npx predict-v8-randomness [args]`	Probably the simplest method
Global `npm`	`> npm i -G predict-v8-randomness` `> predict-v8-randomness [args]`	Can run from any terminal on your machine
Local Project Path	`> node_modules/.bin/predict-v8-randomness [args]`	In a project that has this paackage installed

Dynamically Generated Sequence

Command

predict-v8-randomness --predictions 5

Output

{
  generatedSequence: [
    0.3408102678742442,
    0.08789933352314194,
    0.03272179228873395,
    0.4842101806240975
  ],
  predictions: [
    0.18459229023584944,
    0.00115362787609663,
    0.2168134565216322,
    0.4505911192744063,
    0.0861285118521693
  ],
  actual: [
    0.18459229023584944,
    0.00115362787609663,
    0.2168134565216322,
    0.4505911192744063,
    0.0861285118521693
  ],
  isCorrect: true
}

Provide Your Own Sequence

Generate sequence via Node REPL (among other ways):

generate-sequence-for-cli

Command

Using our generated sequence from Node REPL as --sequence

predict-v8-randomness --predictions 5 --sequence \
0.7491279279338094 \
0.44015510494242127 \
0.7346597255565754 \
0.36002618846830314

Output

{
  sequence: [
    0.7491279279338094,
    0.44015510494242127,
    0.7346597255565754,
    0.36002618846830314
  ],
  predictions: [
    0.6959327841483642,
    0.04084705024171864,
    0.16434259075016922,
    0.8110024854456912,
    0.023498283488221583
  ],
  actual: "You'll need to get this yourself via the same way you generated the sequence"
}

Validation

Generate actual random numbers and compare to predictions above:

actual-next-values-for-cli

包详细信息