Group-items NPM

group-items

group-items is a TypeScript/JavaScript module for grouping arrays by some key, primitive or complex, and into whatever structure you desire.

Example:

import { group } from 'group-items'

// group names by length
const items = ['James', 'John', 'Robert', 'Michael', 'William', 'David']
const byLength = group(items).by('length').asObject()

console.log(byLength)

{
  4: ['John'],
  5: ['James', 'David'],
  6: ['Robert'],
  7: ['Michael', 'William']
}

This module can do a lot more though, as the keying can also be dynamically generated and there are other collectors in addition to .asObject().

Usage

Warning: From version 4.0.0 onwards, group-items is a native ES module and cannot be used in a CommonJS environment.

import { group } from 'group-items'

The basic workflow is as follows:

Start a grouping: group(Iterable)
Provide a keying: .by('property') or .by((item, index) => fn(item, index))
Collect the results: .asObject(), .asTuples(), .keys(), etc.

1. Group creation

Syntax: group(Iterable), where group is the main export of this module. Because this method takes an Iterable, you can give it arrays, sets, strings, ... and it will just work.

2. Keying

There are two ways of providing a keying:

(A) by property name, or
(B) by function.

You've seen an example for (A) in the code snippet above, where the length property was used to group strings. Examples for (B) can be found at the end of this document (in the Examples section). Put simply, you would provide a function that -- when given an element of the input Iterable -- computes some value that can be used as the key for that element. Later I will demonstrate the power of this concept.

Note that almost anything can be used as key. Key equality (and thereby grouping) is determined as per the deep-eql module.

3. Collection

After a grouping has been initialized and keyed, it can be collected in one of many ways:

.asArrays()

collects into an array of arrays (each child array is a group). E.g. if grouping integers by whether they are even or odd, the output might be: [[0, 2, 4, 6, 8], [1, 3, 5, 7, 9]].

.asEntries(options)

collects into an array of { key: <key>, items: [...] } objects. This is one of the more verbose collectors, but sometimes useful. The property names can be customized.

Options:

keyName: name of the key property (default: 'key')
itemsName: name of the items property (default: 'items')

Note that due to limitations of the TypeScript language, the return type of this method cannot be inferred correctly when customizing property names. In those cases, an array of Record will be returned which you have to cast yourself.

.asMap()

collects into a JavaScript Map, as you would expect. Maps have the advantage of supporting non-primitive keys.

.asObject()

collects into a JavaScript Object. You can find an example for this above. Note that due to the nature of JavaScript objects, only values of type string, number or symbol will be included in the result.

.asTuples()

collects into an array of 2-tuples, i.e., an array of the following form:

[
  [keyA, [itemA_0, itemA_1, itemA_2, /* etc ... */]],
  [keyB, [itemB_0, itemB_1, itemB_2, /* etc ... */]],
  // etc ...
]

.keys()

collects just the group keys into an array.

Examples

You've seen one example already. Here are a few more, demonstrating the capabilities.

Grouping numbers by remainder

group([0, 1, 2, 3, 4, 5])
  .by(i => i % 3)
  .asObject()

{
  0: [0, 3],
  1: [1, 4],
  2: [2, 5]
}

Grouping events by date

const events = [
  { date: [2020, 3,  1], title: 'EDC Mexico' },
  { date: [2020, 3, 21], title: 'Ultra Music Festival (ASOT)' },
  { date: [2020, 3, 21], title: 'This Is Me' },
  { date: [2020, 3, 29], title: 'Creamfields' },
  //...
]

group(events).by('date').asMap()

Map {
  [2020, 3, 1] => [ { date: [2020, 3, 1], title: 'EDC Mexico' } ],
  [2020, 3, 21] => [
    { date: [2020, 3, 21], title: 'Ultra Music Festival (ASOT)' },
    { date: [2020, 3, 21], title: 'This Is Me' }
  ],
  [2020, 3, 29] => [ { date: [2020, 3, 29], title: 'Creamfields' } ]
}

Or alternatively:

group(events).by('date').asEntries({ keyName: 'date', itemsName: 'events' })

[
  { date: [2020, 3, 1], events: [ /* ... */ ] },
  { date: [2020, 3, 21], events: [ /* ... */ ] },
  { date: [2020, 3, 29], events: [ /* ... */ ] }
]

Obtaining all unique Map values

This is certainly not the most efficient (or readable) way to do it, but you get the idea.

// initialize some mappings
const map = new Map([
  [0, 'foo'], [2, 'foo'], [3, 'bar'], [8, 'foo'], [9, 'qux'], [11, 'bar']
])

// create a reverse map (map each value to its respective keys)
group(map).by(entry => entry[1]).keys()

['foo', 'bar', 'qux']

Alternated string chunking

The following example makes use of the element index for grouping.

group('ax1by2cz3')
  // create chunks by alternating every 3rd character
  .by((char, index) => index % 3)
  .asArrays()
  // now join the inner arrays
  .map((arr) => arr.join(''))