Wondering what’s next for npm?Check out our public roadmap! »


1.5.1 • Public • Published


A number-hash ready to go for Talk Like A Pirate Day.

Circle CI codecov.io bitHound Score bitHound Dependencies MIT License XO code style

What is it?

Harsh is a tiny library for tokenizing lists of numbers, and then subsequently reversing those tokens back to the original numbers. You can also create a token from nothing by calling it with no arguments.

NOTE: This library should never be used for serious encryption. It is strictly a tokenizer.


v1.5 hashish now accepts arguments. You can only provide one id (so no array), both otherwise the arguments are the same as for hash.


Build process now uses Rollup, which reduces the size a bit but also makes it possible to export a native module. Dist folder now contains harsh.js which is the native module, harsh.umd.js which uses the universal module definition, capable for <script> and node usage, and harsh.min.js which is just the minified version of the umd file.


Added a new method hashish which just returns a single token (no object), to simplify the API for this scenario when you just need to create a single random token.


All functions have been made free-standing so you can now import them individually, import { hash } from 'harsh', etc.


$ git clone git@github.com:patrickfatrick/harsh.git
$ npm install harsh
$ bower install harsh

You can either use it as a module or include the minified file from the dist folder in a script tag.


Harsh basically stringifies the numbers you feed it, and appends randomly generated salts to them in a random order. The salts will always be the same for the entire array, which makes them trivial to reverse, provided that stored the salts. The implementation goes like so:

harsh.hash([number array[, number of salts[, base]]])
// number array defaults to an array containing one random number
// number of salts defaults to 2,
// base defaults to 36
hash() // Create a single token using a random number
hash([1234, 5678]) // Create two tokens that will reverse back to 1234 and 5678
hash([1234, 5678], 3) // Create two tokens that will have 3 salts appended to them
hash([1234, 5678], 3, 16) // Create two tokens with 3 salts and using base-16 (hexadecimal)

resulting in something like

// 1wapgwcgl
// j9exyab2rt, j9exb2rt4dq
// ggtdbrhfyaget1, 4dqbrhfget1ggtd
// 55e8a353c53b8124d2, 55e8a3b812353c5162e

Well really, what it returns is an object that contains the following properties

ids: [1234, 5678], // the array of numbers used
hashes: ['55e8a353c53b8124d2', '55e8a3b812353c5162e'], // the tokens created
salts: ['55e8a', '353c5', '3b812'], // the salts created
base: 16 // the base used for tokenization

If you want to eventually reverse tokens back to the original numbers you will need to store this object somewhere, or at least the salts (assuming you're consistent with the base).


This is a very simple API to just return one token as a string (no object).

hashish() // => 5hfda2fw79
hashish(1234, 2, 36) // Receives the same arguments as `hash`, but just for one id if specified

Creating a 'bunch'

To create a specified number of random tokens, use bunch(). Instead of an array of numbers as the first argument bunch accepts a number, and will create that many tokens based on random numbers.

bunch([number of tokens[, number of salts[, base]]])
// number of tokens defaults to 1
// number of salts defaults to 2,
// base defaults to 36

For example,

bunch() // Creates one token
bunch(5) // Creates five tokens
bunch(5, 3) // Creates five tokens with three salts appended to each one
bunch(5, 3, 16) // Creates five tokens with three salts, and using base-16 (hexadecimal)

This method returns an object with all the same properties as hash().


Reversing works very similarly, the format is

reverse([token array[, salts array[, base]]])
// base defaults to 36

So to reverse the first token from before,

reverse(['55e8a353c53b8124d2'], ['55e8a', '353c5', '3b812'], 16)
// [1234] Note this also returns an array.

The base and salts must match the original hashing, or this will not work.

Running the tests

$ npm install
$ npm test


Harsh is freely distributable under the terms of the MIT license.


npm i harsh

DownloadsWeekly Downloads






Last publish


  • avatar