nodups

nodups is a small yet powerful library specialized in working with arrays basically to get unique values and to drop duplicates. But it's more than just that. Let's see this in action:

Usage

First of all, nodups basic usage, without any options, allows you to get unique values of the passed array:

nodups([ 1, { a: 5 }, 2, { a: 5 },  1 ])
// result is [ 1, { a: 5 }, 2 ]

It leaves the original array unchainged and returns resulting array with unique values in the order of appearence.

Next, to change the original array directly just pass inplace option:

const array = [ 1, { a: 5 }, 2, { a: 5 },  1 ]
nodups(array, { inplace: true })
// array now contains [ 1, { a: 5 }, 2 ]

nodups compares objects values in depth and only checks own enumerable properties of the objects. Primitive properties and values are compared with SameValueZero algorithm - that means nodups treats NaN values as equal to each other by default:

nodups([ NaN, 0, NaN, NaN ])
// result is [ NaN, 0 ]

To change the way nodups compares values you can pass compare option with the compare function:

nodups([ 1, 5, 7, 14, 19, 33, 36 ], { compare: (a, b) => ~~(a/10) === ~~(b/10) })
// result is [ 1, 14, 33 ]

You may also pass '===' and '==' shorthands with compare:

const obj = { a: 1 }
nodups([ obj, { a: 1 }, obj, { a: 1 }, obj ], { compare: '===' })
// result is [ obj, { a: 1 }, { a: 1 } ]
 
nodups([ 0, '', false, [] ], { compare: '==' })
// result is [ 0 ]

If you would like to change the strictness of comparison primitive properties and values are compared with, you may pass strict option with false value:

nodups([ { a: 1 }, { a: '1' } ], { strict: false })
// result is [ { a: 1 } ]

Please note that { strict: false } and { compare: '==' } are different concepts:

nodups([ 0, '', false, NaN, NaN, { a: 1 }, { a: '1' }, { a: '1' } ], { strict: false })
// result is [ 0, NaN, { a: 1 } ]
 
nodups([ 0, '', false, NaN, NaN, { a: 1 }, { a: '1' }, { a: '1' } ], { compare: '==' })
// result is [ 0, NaN, NaN, { a: 1 }, { a: '1' }, { a: '1' } ]

When you know that your array is sorted you can gain additional performance by passing sorted option:

nodups([ 1, 1, 1, 3, 3, 4, 5 ], { sorted: true })
// result is [ 1, 3, 4, 5 ]

But be careful, in case of not sorted array using sorted would lead to failure to drop all duplicates:

nodups([ 1, 3, 1, 1, 3, 4, 5 ], { sorted: true })
// result is [ 1, 3, 1, 3, 4, 5 ]

As already noted, nodups by default compares objects in depth by all their own enumerable properties. To restrict set of properties by which objects are compared you can use by:

nodups([
    { a: 1, b: 3 },
    { a: 2, b: 4 },
    { a: 1, b: 5 },
], { by: [ 'a' ] })
// result is [
//     { a: 1, b: 3 },
//     { a: 2, b: 4 },
// ]

And to ignore some of the properties you can use skip:

nodups([
    { a: 1, b: 3 },
    { a: 2, b: 4 },
    { a: 1, b: 5 },
], { skip: [ 'b' ] })
// result is [
//     { a: 1, b: 3 },
//     { a: 2, b: 4 },
// ]

If both are specified, by wins:

nodups([
    { a: 1, b: 3 },
    { a: 2, b: 4 },
    { a: 1, b: 5 },
], { by: [ 'a' ], skip: [ 'a' ] })
// result is [
//     { a: 1, b: 3 },
//     { a: 2, b: 4 },
// ]

by and skip options accept path or array of paths in objects. Each path may be represented as either .-separated keys of each object level or as array of keys of each level (similar to lodash notion of paths):

nodups([
    { a: { b: 1, c: 2 }, d: 7 },
    { a: { b: 5, c: 4 }, d: 8 },
    { a: { b: 1, c: 2 }, d: 9 },
], { by: [ 'a.b', [ 'a', 'c' ] ] })
// result is [
//     { a: { b: 1, c: 2 }, d: 7 },
//     { a: { b: 5, c: 4 }, d: 8 },
// ]

Though normally we want to drop duplicates, sometimes it is useful to know what are they or how many, so nodups provides you with onUnique option. onUnique option value is expected to be a function with the following signature: (unique, duplicates, index, uniques) and it is called for each unique value:

nodups([
    { a: 1 },
    { a: 2 },
    { a: 1 },
], { onUnique: (unique, duplicates) => unique.dups = duplicates.length })
// result is [
//     { a: 1, dups: 1 },
//     { a: 2, dups: 0 },
// ]

onUnique can be also uses to change resulting array of unique values:

nodups([ 1, 3, 2, 3 ], { onUnique: (uniq, dups, i, uniqs) => uniqs[i] = uniq * 2 })
// result is [ 2, 6, 4 ]

And, of course, options can be intermixed!

Installation

npm i nodups

Polyfill

If you would like nodups method to be available on array instances like that:

[ 1, 1, 1 ].nodups() // result is [ 1 ]

than you can polyfill it:

require('nodups').polyfill()
// or
require('nodups/polyfill')
// or
import nodups from 'nodups/polyfill'

In particular, it is handy in case of method chaining:

array
    .nodups()
    .map(x => x.category)
    .nodups()
    .filter(x => x.startsWith('_'))
    .map(x => getType(x))
    .nodups()

Credits to @zlumer for his proposal on the polyfill and example.

Options

To summarize usage section here is more formal description of nodups options:

• inplace (Boolean) - drop duplicates from original array.
• compare (Function(a, b)|'==='|'==') - custom comparison function of any two array elements or string shorthand.
• strict (Boolean) - true by default. Compare objects' primitive properties with sligtly changed == operation (the only difference is that NaN values are treated as equal).
• sorted (Boolean) - tells nodups that array is sorted and performance optimization can be applied.
• by (String|Array) - compare objects only by own enumerable properties specified by the paths.
• skip (String|Array) - compare object only by own enumerable properties execept ones specified by the paths.
• onUnique - (Function(unique, duplicated, index, uniques)) - callback fired for each unique element. Allows to do some additional work with duplicates.