rndgen

In a Nutshell

A class library returning random data, random integers, and random floating point number either using crypto as source, or deterministically deriving a sequence from sha512 hash function using the caller submitted key.

Reference

const RndGen = require('rndgen');

var rg = new RndGen(); // Non-deterministic instance
rg.getBytes(10);       // 10 bytes of random data in Buffer
rg.getInt();           // random integer 0..9007199254740991 inclusively
rg.getInt(42);         // random integer 0..42 inclusively
rg.getInt(100, 200);   // random integer 100..200 inclusively
rg.rand()              // random number 0.0 <= x < 1.0

var rg = new RndGen('secret key!'); // Deterministic instance
// Same interfaces are available.  Same key and same call order
// produces same sequence of return values.

RndGen()

Constructor for undeterministic random class. All functions called using an instance created like this, return non-reproducible data.

RndGen(key)

Constructor for deterministic random class. All call chains to class functions using an instance created like this produce a deterministic return data that can be reproduced using the same key and same call chain.

RndGen.prototype.getBytes(bytes)

Return a Buffer of random bytes. Number of bytes passed as an argument.

RndGen.prototype.getInt(min, max)

Return a random integer in range min..max inclusively. The limits must be nonnegative safe integers and min must not be bigger than max. If one argument is passed, it is considered the maximum and the minimum defaults to zero. If no arguments are passed, the minimum defaults to zero and the maximum defaults to Number.MAX_SAFE_INTEGER.

RndGen.prototype.rand()

Return a random number x greater or equal to 0 but less than 1.

Because of javascript number type, rand() function can return 2^53 different values and of course they can be enumerated. You'd be very lucky to see any of these particular ones, but rest assured, it is possible that any of those pop up. One that absolutely does NOT pop out is 1.00000000000000000000.

// 1                0.00000000000000000000
// 2                0.00000000000000011102
// 3                0.00000000000000022204
// 4                0.00000000000000033307
// 5                0.00000000000000044409
// 6                0.00000000000000055511
// 7                0.00000000000000066613
// 8                0.00000000000000077716
// 9                0.00000000000000088818
// 10               0.00000000000000099920
//                  .
//                  .
//                  .
// 9007199254740982 0.99999999999999877875
// 9007199254740983 0.99999999999999888978
// 9007199254740984 0.99999999999999900080
// 9007199254740985 0.99999999999999911182
// 9007199254740986 0.99999999999999922284
// 9007199254740987 0.99999999999999933387
// 9007199254740988 0.99999999999999944489
// 9007199254740989 0.99999999999999955591
// 9007199254740990 0.99999999999999966693
// 9007199254740991 0.99999999999999977796
// 9007199254740992 0.99999999999999988898

RndGen.prototype.shuffle(a)

Shuffles array or buffer into random order. The ordering is done inplace and the return value is the given array again.

const RndGen = require('rndgen');
var rg = new RndGen();

// Produce array of integers 0..99 in random order
var seq =  rg.shuffle(new Array(100).fill().map(function(x,i){return i;}));

BigNum

I don't want to introduce node-bignum package as a dependency in this fairly simple and small class since it would bloat it e.g. in case of serverless environments. However in case you are wondering, and why wouldn't you be, how this library could be used for generating evenly distributed bignums, this is how it is done:

"use strict";

const RndGen = require('./rndgen.js');
const BigNum = require('bignum');

var rg = new RndGen();

function rndBigNum(n) {
    if (! BigNum.isBigNum(n)) {
        n = BigNum(n);
    }
    if (! (BigNum.isBigNum(n) && n.ge(0))) {
        throw new Error('Bad limit');
    }
    if (n.eq(0)) {
        return BigNum(0);
    }
    var d, r, b = n.bitLength();
    do {
        d = rg.getBytes(Math.ceil(b / 8));
        if (b % 8) {
            d[0] >>= (8 - (b % 8))
        }
        r = BigNum.fromBuffer(d)
    } while (r.gt(n));
    return r;
}

module.exports = rndBigNum;

Naturally this would be more useful for scenarios where a deterministic sequence is required, in which case you'd of course be using a key for your RndGen instance. The principle is the same for other big integer packages like node-bigint.

Author

Timo J. Rinne [e-mail]

License

GPL-2.0