Rambda

Rambda is smaller and faster alternative to the popular functional programming library Ramda. - Documentation

❯ Example use

import { compose, map, filter } from 'rambda'

const result = compose(
  map(x => x * 2),
  filter(x => x > 2)
)([1, 2, 3, 4])
// => [6, 8]

You can test this example in Rambda's REPL

• Differences between Rambda and Ramda
• API
• Changelog

❯ Rambda's advantages

TypeScript included

TypeScript definitions are included in the library, in comparison to Ramda, where you need to additionally install @types/ramda.

Still, you need to be aware that functional programming features in TypeScript are in development, which means that using R.compose/R.pipe can be problematic.

Important - Rambda version 7.1.0(or higher) requires TypeScript version 4.3.3(or higher).

Understandable source code due to little usage of internals

Ramda uses a lot of internals, which hides a lot of logic. Reading the full source code of a method can be challenging.

Better VSCode experience

If the project is written in Javascript, then go to source definition action will lead you to actual implementation of the method.

Immutable TS definitions

You can use immutable version of Rambda definitions, which is linted with ESLint functional/prefer-readonly-type plugin.

import {add} from 'rambda/immutable'

Deno support

Latest version of Ramba available for Deno users is 3 years old. This is not the case with Rambda as most of recent releases are available for Deno users.

Also, Rambda provides you with included TS definitions:

// Deno extension(https://marketplace.visualstudio.com/items?itemName=denoland.vscode-deno)
// is installed and initialized
import * as R from "https://deno.land/x/rambda/mod.ts";
import * as Ramda from "https://deno.land/x/ramda/mod.ts";

R.add(1)('foo') // => will trigger warning in VSCode as it should
Ramda.add(1)('foo') // => will not trigger warning in VSCode

Dot notation for R.path, R.paths, R.assocPath and R.lensPath

Standard usage of R.path is R.path(['a', 'b'], {a: {b: 1} }).

In Rambda you have the choice to use dot notation(which is arguably more readable):

R.path('a.b', {a: {b: 1} })

Please note that since path input is turned into array, i.e. if you want R.path(['a','1', 'b'], {a: {'1': {b: 2}}}) to return 2, you will have to pass array path, not string path. If you pass a.1.b, it will turn path input to ['a', 1, 'b']. The other side effect is in R.assocPath and R.dissocPath, where inputs such as ['a', '1', 'b'] will be turned into ['a', 1, 'b'].

Comma notation for R.pick and R.omit

Similar to dot notation, but the separator is comma(,) instead of dot(.).

R.pick('a,b', {a: 1 , b: 2, c: 3} })
// No space allowed between properties

Speed

Rambda is generally more performant than Ramda as the benchmarks can prove that.

Support

One of the main issues with Ramda is the slow process of releasing new versions. This is not the case with Rambda as releases are made on regular basis.

❯ Missing Ramda methods

• construct - Using classes is not very functional programming oriented.
• constructN - same as above
• into - no support for transducer as it is overly complex to implement, understand and read.
• invert - overly complicated and limited use case
• invertObj
• invoker
• keysIn - we shouldn't encourage extending object with .prototype
• lift
• liftN
• mapAccum - Ramda example doesn't looks convincing
• mapAccumRight
• memoizeWith - hard to imagine its usage in context of R.pipe/R.compose
• mergeDeepWith - limited use case
• mergeDeepWithKey
• mergeWithKey
• nAry - hard to argument about and hard to create meaningful TypeScript definitions
• nthArg - limited use case
• o - enough TypeScript issues with R.pipe/R.compose to add more composition methods
• otherwise - naming is confusing
• pair - left-pad types of debacles happens partially because of such methods that should not be hidden, bur rather part of your code base even if they need to exist.
• partialRight - I dislike R.partial, so I don't want to add more methods that are based on it
• pipeWith
• project - naming is confusing, but also limited use case
• promap
• reduceRight - I find right/left methods confusing so I added them only where it makes sense.
• reduceWhile - functions with 4 inputs - I think that even 3 is too much
• reduced
• remove - nice name but it is too generic. Also, Rambdax has such method and there it works very differently
• scan - hard to explain
• sequence
• splitWhenever
• symmetricDifferenceWith
• andThen
• toPairsIn
• transduce - currently is out of focus
• traverse - same as above
• unary
• uncurryN
• unfold - similar to R.scan and I find that it doesn't help with readability
• unionWith - why it has its usage, I want to limit number of methods that accept more than 2 arguments
• until
• useWith - hard to explain
• valuesIn
• xprod - limited use case
• thunkify
• __ - placeholder method allows user to further customize the method call. While, it seems useful initially, the price is too high in terms of complexity for TypeScript definitions. If it is not easy exressable in TypeScript, it is not worth it as Rambda is a TypeScript first library.

The following methods are not going to be added(reason for exclusion is provided as a comment):

❯ Install

• yarn add rambda

• For UMD usage either use ./dist/rambda.umd.js or the following CDN link:

https://unpkg.com/rambda@CURRENT_VERSION/dist/rambda.umd.js

• with deno

import {add} from "https://deno.land/x/rambda/mod.ts";

Differences between Rambda and Ramda

• Rambda's type detects async functions and unresolved Promises. The returned values are 'Async' and 'Promise'.

• Rambda's type handles NaN input, in which case it returns NaN.

• Rambda's forEach can iterate over objects not only arrays.

• Rambda's map, filter, partition when they iterate over objects, they pass property and input object as predicate's argument.

• Rambda's filter returns empty array with bad input(null or undefined), while Ramda throws.

• Ramda's clamp work with strings, while Rambda's method work only with numbers.

• Ramda's indexOf/lastIndexOf work with strings and lists, while Rambda's method work only with lists as iterable input.

• Error handling, when wrong inputs are provided, may not be the same. This difference will be better documented once all brute force tests are completed.

• TypeScript definitions between rambda and @types/ramda may vary.

If you need more Ramda methods in Rambda, you may either submit a PR or check the extended version of Rambda - Rambdax. In case of the former, you may want to consult with Rambda contribution guidelines.

❯ Benchmarks

There are methods which are benchmarked only with Ramda and Rambda(i.e. no Lodash).

Note that some of these methods, are called with and without curring. This is done in order to give more detailed performance feedback.

The benchmarks results are produced from latest versions of Rambda, Lodash(4.17.21) and Ramda(0.30.1).

method | Rambda | Ramda | Lodash --- |--- | --- | --- add | 🚀 Fastest | 21.52% slower | 82.15% slower adjust | 8.48% slower | 🚀 Fastest | 🔳 all | 🚀 Fastest | 7.18% slower | 🔳 allPass | 🚀 Fastest | 88.25% slower | 🔳 allPass | 🚀 Fastest | 98.56% slower | 🔳 and | 🚀 Fastest | 89.09% slower | 🔳 any | 🚀 Fastest | 92.87% slower | 45.82% slower anyPass | 🚀 Fastest | 98.25% slower | 🔳 append | 🚀 Fastest | 2.07% slower | 🔳 applySpec | 🚀 Fastest | 80.43% slower | 🔳 assoc | 72.32% slower | 60.08% slower | 🚀 Fastest clone | 🚀 Fastest | 91.86% slower | 86.48% slower compose | 6.07% slower | 16.89% slower | 🚀 Fastest converge | 78.63% slower | 🚀 Fastest | 🔳 curry | 🚀 Fastest | 28.86% slower | 🔳 curryN | 🚀 Fastest | 41.05% slower | 🔳 defaultTo | 🚀 Fastest | 48.91% slower | 🔳 drop | 🚀 Fastest | 82.35% slower | 🔳 dropLast | 🚀 Fastest | 86.74% slower | 🔳 equals | 58.37% slower | 96.73% slower | 🚀 Fastest filter | 6.7% slower | 72.03% slower | 🚀 Fastest find | 🚀 Fastest | 85.14% slower | 42.65% slower findIndex | 🚀 Fastest | 86.48% slower | 72.27% slower flatten | 🚀 Fastest | 85.68% slower | 3.57% slower ifElse | 🚀 Fastest | 58.56% slower | 🔳 includes | 🚀 Fastest | 81.64% slower | 🔳 indexOf | 🚀 Fastest | 80.17% slower | 🔳 indexOf | 🚀 Fastest | 82.2% slower | 🔳 init | 🚀 Fastest | 92.24% slower | 13.3% slower is | 🚀 Fastest | 57.69% slower | 🔳 isEmpty | 🚀 Fastest | 97.14% slower | 54.99% slower last | 🚀 Fastest | 93.43% slower | 5.28% slower lastIndexOf | 🚀 Fastest | 85.19% slower | 🔳 map | 🚀 Fastest | 86.6% slower | 11.73% slower match | 🚀 Fastest | 44.83% slower | 🔳 merge | 🚀 Fastest | 12.21% slower | 55.76% slower none | 🚀 Fastest | 96.48% slower | 🔳 objOf | 🚀 Fastest | 38.05% slower | 🔳 omit | 🚀 Fastest | 69.95% slower | 97.34% slower over | 🚀 Fastest | 56.23% slower | 🔳 path | 37.81% slower | 77.81% slower | 🚀 Fastest pick | 🚀 Fastest | 19.07% slower | 80.2% slower pipe | 🚀 Fastest | 0.11% slower | 🔳 prop | 🚀 Fastest | 87.95% slower | 🔳 propEq | 🚀 Fastest | 91.92% slower | 🔳 range | 🚀 Fastest | 61.8% slower | 57.44% slower reduce | 60.48% slower | 77.1% slower | 🚀 Fastest repeat | 48.57% slower | 68.98% slower | 🚀 Fastest replace | 33.45% slower | 33.99% slower | 🚀 Fastest set | 🚀 Fastest | 50.35% slower | 🔳 sort | 🚀 Fastest | 40.23% slower | 🔳 sortBy | 🚀 Fastest | 25.29% slower | 56.88% slower split | 🚀 Fastest | 55.37% slower | 17.64% slower splitEvery | 🚀 Fastest | 71.98% slower | 🔳 take | 🚀 Fastest | 91.96% slower | 4.72% slower takeLast | 🚀 Fastest | 93.39% slower | 19.22% slower test | 🚀 Fastest | 82.34% slower | 🔳 type | 🚀 Fastest | 48.6% slower | 🔳 uniq | 🚀 Fastest | 84.9% slower | 🔳 uniqBy | 51.93% slower | 🚀 Fastest | 🔳 uniqWith | 8.29% slower | 🚀 Fastest | 🔳 uniqWith | 14.23% slower | 🚀 Fastest | 🔳 update | 🚀 Fastest | 52.35% slower | 🔳 view | 🚀 Fastest | 76.15% slower | 🔳

❯ Used by

• ESLint plugin Mocha

• WatermelonDB

• Walmart Canada reported by w-b-dev

• VSCode Slack integration

• Webpack PostCSS

• MobX-State-Tree decorators

• Rewrite of the Betaflight configurator

• MineFlayer plugin

API

add

It adds a and b.

Try this R.add example in Rambda REPL

addIndex

Try this R.addIndex example in Rambda REPL

addIndexRight

Same as R.addIndex, but it will passed indexes are decreasing, instead of increasing.

adjust

adjust<T>(index: number, replaceFn: (x: T) => T, list: T[]): T[]

It replaces index in array list with the result of replaceFn(list[i]).

Try this R.adjust example in Rambda REPL

adjust<T>(index: number, replaceFn: (x: T) => T, list: T[]): T[];
adjust<T>(index: number, replaceFn: (x: T) => T): (list: T[]) => T[];

import { cloneList } from './_internals/cloneList.js'
import { curry } from './curry.js'

function adjustFn(
  index, replaceFn, list
){
  const actualIndex = index < 0 ? list.length + index : index
  if (index >= list.length || actualIndex < 0) return list

  const clone = cloneList(list)
  clone[ actualIndex ] = replaceFn(clone[ actualIndex ])

  return clone
}

export const adjust = curry(adjustFn)

import { add } from './add.js'
import { adjust } from './adjust.js'
import { pipe } from './pipe.js'

const list = [ 0, 1, 2 ]
const expected = [ 0, 11, 2 ]

test('happy', () => {})

test('happy', () => {
  expect(adjust(
    1, add(10), list
  )).toEqual(expected)
})

test('with curring type 1 1 1', () => {
  expect(adjust(1)(add(10))(list)).toEqual(expected)
})

test('with curring type 1 2', () => {
  expect(adjust(1)(add(10), list)).toEqual(expected)
})

test('with curring type 2 1', () => {
  expect(adjust(1, add(10))(list)).toEqual(expected)
})

test('with negative index', () => {
  expect(adjust(
    -2, add(10), list
  )).toEqual(expected)
})

test('when index is out of bounds', () => {
  const list = [ 0, 1, 2, 3 ]
  expect(adjust(
    4, add(1), list
  )).toEqual(list)
  expect(adjust(
    -5, add(1), list
  )).toEqual(list)
})

all

all<T>(predicate: (x: T) => boolean, list: T[]): boolean

It returns true, if all members of array list returns true, when applied as argument to predicate function.

Try this R.all example in Rambda REPL

all<T>(predicate: (x: T) => boolean, list: T[]): boolean;
all<T>(predicate: (x: T) => boolean): (list: T[]) => boolean;

export function all(predicate, list){
  if (arguments.length === 1) return _list => all(predicate, _list)

  for (let i = 0; i < list.length; i++){
    if (!predicate(list[ i ])) return false
  }

  return true
}

import { all } from './all.js'

const list = [ 0, 1, 2, 3, 4 ]

test('when true', () => {
  const fn = x => x > -1

  expect(all(fn)(list)).toBeTrue()
})

test('when false', () => {
  const fn = x => x > 2

  expect(all(fn, list)).toBeFalse()
})

import {all} from 'rambda'

describe('all', () => {
  it('happy', () => {
    const result = all(
      x => {
        x // $ExpectType number
        return x > 0
      },
      [1, 2, 3]
    )
    result // $ExpectType boolean
  })
  it('curried needs a type', () => {
    const result = all<number>(x => {
      x // $ExpectType number
      return x > 0
    })([1, 2, 3])
    result // $ExpectType boolean
  })
})

allPass

allPass<T>(predicates: ((x: T) => boolean)[]): (input: T) => boolean

It returns true, if all functions of predicates return true, when input is their argument.

Try this R.allPass example in Rambda REPL

allPass<T>(predicates: ((x: T) => boolean)[]): (input: T) => boolean;
allPass<T>(predicates: ((...inputs: T[]) => boolean)[]): (...inputs: T[]) => boolean;

export function allPass(predicates){
  return (...input) => {
    let counter = 0
    while (counter < predicates.length){
      if (!predicates[ counter ](...input)){
        return false
      }
      counter++
    }

    return true
  }
}

import { allPass } from './allPass.js'

test('happy', () => {
  const rules = [ x => typeof x === 'number', x => x > 10, x => x * 7 < 100 ]

  expect(allPass(rules)(11)).toBeTrue()

  expect(allPass(rules)(undefined)).toBeFalse()
})

test('when returns true', () => {
  const conditionArr = [ val => val.a === 1, val => val.b === 2 ]

  expect(allPass(conditionArr)({
    a : 1,
    b : 2,
  })).toBeTrue()
})

test('when returns false', () => {
  const conditionArr = [ val => val.a === 1, val => val.b === 3 ]

  expect(allPass(conditionArr)({
    a : 1,
    b : 2,
  })).toBeFalse()
})

test('works with multiple inputs', () => {
  const fn = function (
    w, x, y, z
  ){
    return w + x === y + z
  }
  expect(allPass([ fn ])(
    3, 3, 3, 3
  )).toBeTrue()
})

import {allPass, filter} from 'rambda'

describe('allPass', () => {
  it('happy', () => {
    const x = allPass<number>([
      y => {
        y // $ExpectType number
        return typeof y === 'number'
      },
      y => {
        return y > 0
      },
    ])(11)

    x // $ExpectType boolean
  })
  it('issue #642', () => {
    const isGreater = (num: number) => num > 5
    const pred = allPass([isGreater])
    const xs = [0, 1, 2, 3]

    const filtered1 = filter(pred)(xs)
    filtered1 // $ExpectType number[]
    const filtered2 = xs.filter(pred)
    filtered2 // $ExpectType number[]
  })
  it('issue #604', () => {
    const plusEq = function(w: number, x: number, y: number, z: number) {
      return w + x === y + z
    }
    const result = allPass([plusEq])(3, 3, 3, 3)

    result // $ExpectType boolean
  })
})

always

It returns function that always returns x.

Try this R.always example in Rambda REPL

and

Logical AND

Try this R.and example in Rambda REPL

any

any<T>(predicate: (x: T) => boolean, list: T[]): boolean

It returns true, if at least one member of list returns true, when passed to a predicate function.

Try this R.any example in Rambda REPL

any<T>(predicate: (x: T) => boolean, list: T[]): boolean;
any<T>(predicate: (x: T) => boolean): (list: T[]) => boolean;

export function any(predicate, list){
  if (arguments.length === 1) return _list => any(predicate, _list)

  let counter = 0
  while (counter < list.length){
    if (predicate(list[ counter ], counter)){
      return true
    }
    counter++
  }

  return false
}

import { any } from './any.js'

const list = [ 1, 2, 3 ]

test('happy', () => {
  expect(any(x => x < 0, list)).toBeFalse()
})

test('with curry', () => {
  expect(any(x => x > 2)(list)).toBeTrue()
})

import {any} from 'rambda'

describe('R.any', () => {
  it('happy', () => {
    const result = any(
      x => {
        x // $ExpectType number
        return x > 2
      },
      [1, 2, 3]
    )
    result // $ExpectType boolean
  })

  it('when curried needs a type', () => {
    const result = any<number>(x => {
      x // $ExpectType number
      return x > 2
    })([1, 2, 3])
    result // $ExpectType boolean
  })
})

anyPass

anyPass<T>(predicates: ((x: T) => boolean)[]): (input: T) => boolean

It accepts list of predicates and returns a function. This function with its input will return true, if any of predicates returns true for this input.

Try this R.anyPass example in Rambda REPL

anyPass<T>(predicates: ((x: T) => boolean)[]): (input: T) => boolean;
anyPass<T>(predicates: ((...inputs: T[]) => boolean)[]): (...inputs: T[]) => boolean;

export function anyPass(predicates){
  return (...input) => {
    let counter = 0
    while (counter < predicates.length){
      if (predicates[ counter ](...input)){
        return true
      }
      counter++
    }

    return false
  }
}

import { anyPass } from './anyPass.js'

test('happy', () => {
  const rules = [ x => typeof x === 'string', x => x > 10 ]
  const predicate = anyPass(rules)
  expect(predicate('foo')).toBeTrue()
  expect(predicate(6)).toBeFalse()
})

test('happy', () => {
  const rules = [ x => typeof x === 'string', x => x > 10 ]

  expect(anyPass(rules)(11)).toBeTrue()
  expect(anyPass(rules)(undefined)).toBeFalse()
})

const obj = {
  a : 1,
  b : 2,
}

test('when returns true', () => {
  const conditionArr = [ val => val.a === 1, val => val.a === 2 ]

  expect(anyPass(conditionArr)(obj)).toBeTrue()
})

test('when returns false + curry', () => {
  const conditionArr = [ val => val.a === 2, val => val.b === 3 ]

  expect(anyPass(conditionArr)(obj)).toBeFalse()
})

test('with empty predicates list', () => {
  expect(anyPass([])(3)).toBeFalse()
})

test('works with multiple inputs', () => {
  const fn = function (
    w, x, y, z
  ){
    console.log(
      w, x, y, z
    )

    return w + x === y + z
  }
  expect(anyPass([ fn ])(
    3, 3, 3, 3
  )).toBeTrue()
})

import {anyPass, filter} from 'rambda'

describe('anyPass', () => {
  it('happy', () => {
    const x = anyPass<number>([
      y => {
        y // $ExpectType number
        return typeof y === 'number'
      },
      y => {
        return y > 0
      },
    ])(11)

    x // $ExpectType boolean
  })
  it('issue #604', () => {
    const plusEq = function(w: number, x: number, y: number, z: number) {
      return w + x === y + z
    }
    const result = anyPass([plusEq])(3, 3, 3, 3)

    result // $ExpectType boolean
  })
  it('issue #642', () => {
    const isGreater = (num: number) => num > 5
    const pred = anyPass([isGreater])
    const xs = [0, 1, 2, 3]

    const filtered1 = filter(pred)(xs)
    filtered1 // $ExpectType number[]
    const filtered2 = xs.filter(pred)
    filtered2 // $ExpectType number[]
  })
  it('functions as a type guard', () => {
    const isString = (x: unknown): x is string => typeof x === 'string'
    const isNumber = (x: unknown): x is number => typeof x === 'number'
    const isBoolean = (x: unknown): x is boolean => typeof x === 'boolean'

    const isStringNumberOrBoolean = anyPass([isString, isNumber, isBoolean])

    isStringNumberOrBoolean // $ExpectType (input: unknown) => boolean

    const aValue: unknown = 1

    if (isStringNumberOrBoolean(aValue)) {
      aValue // $ExpectType unknown
    }
  })
})

ap

ap<T, U>(fns: Array<(a: T) => U>[], vs: T[]): U[]

It takes a list of functions and a list of values. Then it returns a list of values obtained by applying each function to each value.

Try this R.ap example in Rambda REPL

ap<T, U>(fns: Array<(a: T) => U>[], vs: T[]): U[];
ap<T, U>(fns: Array<(a: T) => U>): (vs: T[]) => U[];
ap<R, A, B>(fn: (r: R, a: A) => B, fn1: (r: R) => A): (r: R) => B;

export function ap(functions, input){
  if (arguments.length === 1){
    return _inputs => ap(functions, _inputs)
  }

  return functions.reduce((acc, fn) => [ ...acc, ...input.map(fn) ], [])
}

import { ap } from './ap.js'

function mult2(x){
  return x * 2
}
function plus3(x){
  return x + 3
}

test('happy', () => {
  expect(ap([ mult2, plus3 ], [ 1, 2, 3 ])).toEqual([ 2, 4, 6, 4, 5, 6 ])
})

aperture

aperture<N extends number, T>(n: N, list: T[]): Array<Tuple<T, N>> | []

It returns a new list, composed of consecutive n-tuples from a list.

Try this R.aperture example in Rambda REPL

aperture<N extends number, T>(n: N, list: T[]): Array<Tuple<T, N>> | [];
aperture<N extends number>(n: N): <T>(list: T[]) => Array<Tuple<T, N>> | [];

export function aperture(step, list){
  if (arguments.length === 1){
    return _list => aperture(step, _list)
  }
  if (step > list.length) return []
  let idx = 0
  const limit = list.length - (step - 1)
  const acc = new Array(limit)
  while (idx < limit){
    acc[ idx ] = list.slice(idx, idx + step)
    idx += 1
  }

  return acc
}

import { aperture } from './aperture.js'

const list = [ 1, 2, 3, 4, 5, 6, 7 ]

test('happy', () => {
  expect(aperture(1, list)).toEqual([ [ 1 ], [ 2 ], [ 3 ], [ 4 ], [ 5 ], [ 6 ], [ 7 ] ])
  expect(aperture(2, list)).toEqual([
    [ 1, 2 ],
    [ 2, 3 ],
    [ 3, 4 ],
    [ 4, 5 ],
    [ 5, 6 ],
    [ 6, 7 ],
  ])
  expect(aperture(3, list)).toEqual([
    [ 1, 2, 3 ],
    [ 2, 3, 4 ],
    [ 3, 4, 5 ],
    [ 4, 5, 6 ],
    [ 5, 6, 7 ],
  ])
  expect(aperture(8, list)).toEqual([])
})

append

append<T>(xToAppend: T, iterable: T[]): T[]

It adds element x at the end of iterable.

Try this R.append example in Rambda REPL

append<T>(xToAppend: T, iterable: T[]): T[];
append<T, U>(xToAppend: T, iterable: IsFirstSubtypeOfSecond<T, U>[]) : U[];
append<T>(xToAppend: T): <U>(iterable: IsFirstSubtypeOfSecond<T, U>[]) => U[];
append<T>(xToAppend: T): (iterable: T[]) => T[];

import { cloneList } from './_internals/cloneList.js'

export function append(x, input){
  if (arguments.length === 1) return _input => append(x, _input)

  if (typeof input === 'string') return input.split('').concat(x)

  const clone = cloneList(input)
  clone.push(x)

  return clone
}

import { append } from './append.js'

test('happy', () => {
  expect(append('tests', [ 'write', 'more' ])).toEqual([
    'write',
    'more',
    'tests',
  ])
})

test('append to empty array', () => {
  expect(append('tests')([])).toEqual([ 'tests' ])
})

test('with strings', () => {
  expect(append('o', 'fo')).toEqual([ 'f', 'o', 'o' ])
})

import {append, prepend} from 'rambda'

const listOfNumbers = [1, 2, 3]
const listOfNumbersAndStrings = [1, 'b', 3]

describe('R.append/R.prepend', () => {
  describe("with the same primitive type as the array's elements", () => {
    it('uncurried', () => {
      // @ts-expect-error
      append('d', listOfNumbers)
      // @ts-expect-error
      prepend('d', listOfNumbers)
      append(4, listOfNumbers) // $ExpectType number[]
      prepend(4, listOfNumbers) // $ExpectType number[]
    })

    it('curried', () => {
      // @ts-expect-error
      append('d')(listOfNumbers)
      append(4)(listOfNumbers) // $ExpectType number[]
      prepend(4)(listOfNumbers) // $ExpectType number[]
    })
  })

  describe("with a subtype of the array's elements", () => {
    it('uncurried', () => {
      // @ts-expect-error
      append(true, listOfNumbersAndStrings)
      append(4, listOfNumbersAndStrings) // $ExpectType (string | number)[]
      prepend(4, listOfNumbersAndStrings) // $ExpectType (string | number)[]
    })

    it('curried', () => {
      // @ts-expect-error
      append(true)(listOfNumbersAndStrings)
      append(4)(listOfNumbersAndStrings) // $ExpectType (string | number)[]
      prepend(4)(listOfNumbersAndStrings) // $ExpectType (string | number)[]
    })
  })

  describe("expanding the type of the array's elements", () => {
    it('uncurried', () => {
      // @ts-expect-error
      append('d', listOfNumbers)
      append<string | number>('d', listOfNumbers) // $ExpectType (string | number)[]
      prepend<string | number>('d', listOfNumbers) // $ExpectType (string | number)[]
    })

    it('curried', () => {
      // @ts-expect-error
      append('d')(listOfNumbers)
      const appendD = append('d')
      appendD<string | number>(listOfNumbers) // $ExpectType (string | number)[]
      const prependD = prepend('d')
      prependD<string | number>(listOfNumbers) // $ExpectType (string | number)[]
    })
  })
})

apply

apply<T = any>(fn: (...args: any[]) => T, args: any[]): T

It applies function fn to the list of arguments.

This is useful for creating a fixed-arity function from a variadic function. fn should be a bound function if context is significant.

Try this R.apply example in Rambda REPL

apply<T = any>(fn: (...args: any[]) => T, args: any[]): T;
apply<T = any>(fn: (...args: any[]) => T): (args: any[]) => T;

export function apply(fn, args){
  if (arguments.length === 1){
    return _args => apply(fn, _args)
  }

  return fn.apply(this, args)
}

import { apply } from './apply.js'
import { bind } from './bind.js'
import { identity } from './identity.js'

test('happy', () => {
  expect(apply(identity, [ 1, 2, 3 ])).toBe(1)
})

test('applies function to argument list', () => {
  expect(apply(Math.max, [ 1, 2, 3, -99, 42, 6, 7 ])).toBe(42)
})

test('provides no way to specify context', () => {
  const obj = {
    method (){
      return this === obj
    },
  }
  expect(apply(obj.method, [])).toBeFalse()
  expect(apply(bind(obj.method, obj), [])).toBeTrue()
})

import {apply, identity} from 'rambda'

describe('R.apply', () => {
  it('happy', () => {
    const result = apply<number>(identity, [1, 2, 3])

    result // $ExpectType number
  })
  it('curried', () => {
    const fn = apply<number>(identity)
    const result = fn([1, 2, 3])

    result // $ExpectType number
  })
})

applySpec

applySpec<Spec extends Record<string, AnyFunction>>(
  spec: Spec
): (
  ...args: Parameters<ValueOfRecord<Spec>>
) => { [Key in keyof Spec]: ReturnType<Spec[Key]> }

Try this R.applySpec example in Rambda REPL

applySpec<Spec extends Record<string, AnyFunction>>(
  spec: Spec
): (
  ...args: Parameters<ValueOfRecord<Spec>>
) => { [Key in keyof Spec]: ReturnType<Spec[Key]> };
applySpec<T>(spec: any): (...args: unknown[]) => T;

import { isArray } from './_internals/isArray.js'

// recursively traverse the given spec object to find the highest arity function
export function __findHighestArity(spec, max = 0){
  for (const key in spec){
    if (spec.hasOwnProperty(key) === false || key === 'constructor') continue

    if (typeof spec[ key ] === 'object'){
      max = Math.max(max, __findHighestArity(spec[ key ]))
    }

    if (typeof spec[ key ] === 'function'){
      max = Math.max(max, spec[ key ].length)
    }
  }

  return max
}

function __filterUndefined(){
  const defined = []
  let i = 0
  const l = arguments.length
  while (i < l){
    if (typeof arguments[ i ] === 'undefined') break
    defined[ i ] = arguments[ i ]
    i++
  }

  return defined
}

function __applySpecWithArity(
  spec, arity, cache
){
  const remaining = arity - cache.length

  if (remaining === 1)
    return x =>
      __applySpecWithArity(
        spec, arity, __filterUndefined(...cache, x)
      )
  if (remaining === 2)
    return (x, y) =>
      __applySpecWithArity(
        spec, arity, __filterUndefined(
          ...cache, x, y
        )
      )
  if (remaining === 3)
    return (
      x, y, z
    ) =>
      __applySpecWithArity(
        spec, arity, __filterUndefined(
          ...cache, x, y, z
        )
      )
  if (remaining === 4)
    return (
      x, y, z, a
    ) =>
      __applySpecWithArity(
        spec,
        arity,
        __filterUndefined(
          ...cache, x, y, z, a
        )
      )
  if (remaining > 4)
    return (...args) =>
      __applySpecWithArity(
        spec, arity, __filterUndefined(...cache, ...args)
      )

  // handle spec as Array
  if (isArray(spec)){
    const ret = []
    let i = 0
    const l = spec.length
    for (; i < l; i++){
      // handle recursive spec inside array
      if (typeof spec[ i ] === 'object' || isArray(spec[ i ])){
        ret[ i ] = __applySpecWithArity(
          spec[ i ], arity, cache
        )
      }
      // apply spec to the key
      if (typeof spec[ i ] === 'function'){
        ret[ i ] = spec[ i ](...cache)
      }
    }

    return ret
  }

  // handle spec as Object
  const ret = {}
  // apply callbacks to each property in the spec object
  for (const key in spec){
    if (spec.hasOwnProperty(key) === false || key === 'constructor') continue

    // apply the spec recursively
    if (typeof spec[ key ] === 'object'){
      ret[ key ] = __applySpecWithArity(
        spec[ key ], arity, cache
      )
      continue
    }

    // apply spec to the key
    if (typeof spec[ key ] === 'function'){
      ret[ key ] = spec[ key ](...cache)
    }
  }

  return ret
}

export function applySpec(spec, ...args){
  // get the highest arity spec function, cache the result and pass to __applySpecWithArity
  const arity = __findHighestArity(spec)

  if (arity === 0){
    return () => ({})
  }
  const toReturn = __applySpecWithArity(
    spec, arity, args
  )

  return toReturn
}

import { applySpec as applySpecRamda, nAry } from 'ramda'

import {
  add,
  always,
  compose,
  dec,
  inc,
  map,
  path,
  prop,
  T,
} from '../rambda.js'
import { applySpec } from './applySpec.js'

test('different than Ramda when bad spec', () => {
  const result = applySpec({ sum : { a : 1 } })(1, 2)
  const ramdaResult = applySpecRamda({ sum : { a : 1 } })(1, 2)
  expect(result).toEqual({})
  expect(ramdaResult).toEqual({ sum : { a : {} } })
})

test('works with empty spec', () => {
  expect(applySpec({})()).toEqual({})
  expect(applySpec([])(1, 2)).toEqual({})
  expect(applySpec(null)(1, 2)).toEqual({})
})

test('works with unary functions', () => {
  const result = applySpec({
    v : inc,
    u : dec,
  })(1)
  const expected = {
    v : 2,
    u : 0,
  }
  expect(result).toEqual(expected)
})

test('works with binary functions', () => {
  const result = applySpec({ sum : add })(1, 2)
  expect(result).toEqual({ sum : 3 })
})

test('works with nested specs', () => {
  const result = applySpec({
    unnested : always(0),
    nested   : { sum : add },
  })(1, 2)
  const expected = {
    unnested : 0,
    nested   : { sum : 3 },
  }
  expect(result).toEqual(expected)
})

test('works with arrays of nested specs', () => {
  const result = applySpec({
    unnested : always(0),
    nested   : [ { sum : add } ],
  })(1, 2)

  expect(result).toEqual({
    unnested : 0,
    nested   : [ { sum : 3 } ],
  })
})

test('works with arrays of spec objects', () => {
  const result = applySpec([ { sum : add } ])(1, 2)

  expect(result).toEqual([ { sum : 3 } ])
})

test('works with arrays of functions', () => {
  const result = applySpec([ map(prop('a')), map(prop('b')) ])([
    {
      a : 'a1',
      b : 'b1',
    },
    {
      a : 'a2',
      b : 'b2',
    },
  ])
  const expected = [
    [ 'a1', 'a2' ],
    [ 'b1', 'b2' ],
  ]
  expect(result).toEqual(expected)
})

test('works with a spec defining a map key', () => {
  expect(applySpec({ map : prop('a') })({ a : 1 })).toEqual({ map : 1 })
})

test('cannot retains the highest arity', () => {
  const f = applySpec({
    f1 : nAry(2, T),
    f2 : nAry(5, T),
  })
  const fRamda = applySpecRamda({
    f1 : nAry(2, T),
    f2 : nAry(5, T),
  })
  expect(f).toHaveLength(0)
  expect(fRamda).toHaveLength(5)
})

test('returns a curried function', () => {
  expect(applySpec({ sum : add })(1)(2)).toEqual({ sum : 3 })
})

// Additional tests
// ============================================
test('arity', () => {
  const spec = {
    one   : x1 => x1,
    two   : (x1, x2) => x1 + x2,
    three : (
      x1, x2, x3
    ) => x1 + x2 + x3,
  }
  expect(applySpec(
    spec, 1, 2, 3
  )).toEqual({
    one   : 1,
    two   : 3,
    three : 6,
  })
})

test('arity over 5 arguments', () => {
  const spec = {
    one   : x1 => x1,
    two   : (x1, x2) => x1 + x2,
    three : (
      x1, x2, x3
    ) => x1 + x2 + x3,
    four : (
      x1, x2, x3, x4
    ) => x1 + x2 + x3 + x4,
    five : (
      x1, x2, x3, x4, x5
    ) => x1 + x2 + x3 + x4 + x5,
  }
  expect(applySpec(
    spec, 1, 2, 3, 4, 5
  )).toEqual({
    one   : 1,
    two   : 3,
    three : 6,
    four  : 10,
    five  : 15,
  })
})

test('curried', () => {
  const spec = {
    one   : x1 => x1,
    two   : (x1, x2) => x1 + x2,
    three : (
      x1, x2, x3
    ) => x1 + x2 + x3,
  }
  expect(applySpec(spec)(1)(2)(3)).toEqual({
    one   : 1,
    two   : 3,
    three : 6,
  })
})

test('curried over 5 arguments', () => {
  const spec = {
    one   : x1 => x1,
    two   : (x1, x2) => x1 + x2,
    three : (
      x1, x2, x3
    ) => x1 + x2 + x3,
    four : (
      x1, x2, x3, x4
    ) => x1 + x2 + x3 + x4,
    five : (
      x1, x2, x3, x4, x5
    ) => x1 + x2 + x3 + x4 + x5,
  }
  expect(applySpec(spec)(1)(2)(3)(4)(5)).toEqual({
    one   : 1,
    two   : 3,
    three : 6,
    four  : 10,
    five  : 15,
  })
})

test('undefined property', () => {
  const spec = { prop : path([ 'property', 'doesnt', 'exist' ]) }
  expect(applySpec(spec, {})).toEqual({ prop : undefined })
})

test('restructure json object', () => {
  const spec = {
    id          : path('user.id'),
    name        : path('user.firstname'),
    profile     : path('user.profile'),
    doesntExist : path('user.profile.doesntExist'),
    info        : { views : compose(inc, prop('views')) },
    type        : always('playa'),
  }

  const data = {
    user : {
      id        : 1337,
      firstname : 'john',
      lastname  : 'shaft',
      profile   : 'shaft69',
    },
    views : 42,
  }

  expect(applySpec(spec, data)).toEqual({
    id          : 1337,
    name        : 'john',
    profile     : 'shaft69',
    doesntExist : undefined,
    info        : { views : 43 },
    type        : 'playa',
  })
})

import {multiply, applySpec, inc, dec, add} from 'rambda'

describe('applySpec', () => {
  it('ramda 1', () => {
    const result = applySpec({
      v: inc,
      u: dec,
    })(1)
    result // $ExpectType { v: number; u: number; }
  })
  it('ramda 1', () => {
    interface Output {
      sum: number,
      multiplied: number,
    }
    const result = applySpec<Output>({
      sum: add,
      multiplied: multiply,
    })(1, 2)

    result // $ExpectType Output
  })
})

applyTo

Try this R.applyTo example in Rambda REPL

ascend

Try this R.ascend example in Rambda REPL

assoc

It makes a shallow clone of obj with setting or overriding the property prop with newValue.

Try this R.assoc example in Rambda REPL

assocPath

assocPath<Output>(path: Path, newValue: any, obj: object): Output

It makes a shallow clone of obj with setting or overriding with newValue the property found with path.

Try this R.assocPath example in Rambda REPL

assocPath<Output>(path: Path, newValue: any, obj: object): Output;
assocPath<Output>(path: Path, newValue: any): (obj: object) => Output;
assocPath<Output>(path: Path): (newValue: any) => (obj: object) => Output;

import { cloneList } from './_internals/cloneList.js'
import { createPath } from './_internals/createPath.js'
import { isArray } from './_internals/isArray.js'
import { isIndexInteger } from './_internals/isInteger.js'
import { assocFn } from './assoc.js'
import { curry } from './curry.js'

export function assocPathFn(
  path, newValue, input
){
  const pathArrValue = createPath(path)
  if (pathArrValue.length === 0) return newValue

  const index = pathArrValue[ 0 ]
  if (pathArrValue.length > 1){
    const condition =
      typeof input !== 'object' ||
      input === null ||
      !input.hasOwnProperty(index)

    const nextInput = condition ?
      isIndexInteger(pathArrValue[ 1 ]) ?
        [] :
        {} :
      input[ index ]

    newValue = assocPathFn(
      Array.prototype.slice.call(pathArrValue, 1),
      newValue,
      nextInput
    )
  }

  if (isIndexInteger(index) && isArray(input)){
    const arr = cloneList(input)
    arr[ index ] = newValue

    return arr
  }

  return assocFn(
    index, newValue, input
  )
}

export const assocPath = curry(assocPathFn)

import { assocPathFn } from './assocPath.js'

test('happy', () => {
  const path = 'a.c.1'
  const input = {
    a : {
      b : 1,
      c : [ 1, 2 ],
    },
  }
  assocPathFn(
    path, 3, input
  )
  expect(input).toEqual({
    a : {
      b : 1,
      c : [ 1, 2 ],
    },
  })
})

test('string can be used as path input', () => {
  const testObj = {
    a : [ { b : 1 }, { b : 2 } ],
    d : 3,
  }
  const result1 = assocPathFn(
    [ 'a', 0, 'b' ], 10, testObj
  )
  const result2 = assocPathFn(
    'a.0.b', 10, testObj
  )

  const expected = {
    a : [ { b : 10 }, { b : 2 } ],
    d : 3,
  }
  expect(result1).toEqual(expected)
  expect(result2).toEqual(expected)
})

test('difference with ramda - doesn\'t overwrite primitive values with keys in the path', () => {
  const obj = { a : 'str' }
  const result = assocPathFn(
    [ 'a', 'b' ], 42, obj
  )

  expect(result).toEqual({
    a : {
      0 : 's',
      1 : 't',
      2 : 'r',
      b : 42,
    },
  })
})

test('adds a key to an empty object', () => {
  expect(assocPathFn(
    [ 'a' ], 1, {}
  )).toEqual({ a : 1 })
})

test('adds a key to a non-empty object', () => {
  expect(assocPathFn(
    'b', 2, { a : 1 }
  )).toEqual({
    a : 1,
    b : 2,
  })
})

test('adds a nested key to a non-empty object', () => {
  expect(assocPathFn(
    'b.c', 2, { a : 1 }
  )).toEqual({
    a : 1,
    b : { c : 2 },
  })
})

test('adds a nested key to a nested non-empty object', () => {
  expect(assocPathFn('b.d',
    3,{
    a : 1,
    b : { c : 2 },
  })).toEqual({
    a : 1,
    b : {
      c : 2,
      d : 3,
    },
  })
})

test('adds a key to a non-empty object', () => {
  expect(assocPathFn('b', 2, { a : 1 })).toEqual({
    a : 1,
    b : 2,
  })
})

test('adds a nested key to a non-empty object', () => {
  expect(assocPathFn('b.c', 2, { a : 1 })).toEqual({
    a : 1,
    b : { c : 2 },
  })
})

test('changes an existing key', () => {
  expect(assocPathFn(
    'a', 2, { a : 1 }
  )).toEqual({ a : 2 })
})

test('undefined is considered an empty object', () => {
  expect(assocPathFn(
    'a', 1, undefined
  )).toEqual({ a : 1 })
})

test('null is considered an empty object', () => {
  expect(assocPathFn(
    'a', 1, null
  )).toEqual({ a : 1 })
})

test('value can be null', () => {
  expect(assocPathFn(
    'a', null, null
  )).toEqual({ a : null })
})

test('value can be undefined', () => {
  expect(assocPathFn(
    'a', undefined, null
  )).toEqual({ a : undefined })
})

test('assignment is shallow', () => {
  expect(assocPathFn(
    'a', { b : 2 }, { a : { c : 3 } }
  )).toEqual({ a : { b : 2 } })
})

test('empty array as path', () => {
  const result = assocPathFn(
    [], 3, {
      a : 1,
      b : 2,
    }
  )
  expect(result).toBe(3)
})

test('happy', () => {
  const expected = { foo : { bar : { baz : 42 } } }
  const result = assocPathFn(
    [ 'foo', 'bar', 'baz' ], 42, { foo : null }
  )
  expect(result).toEqual(expected)
})

import {assocPath} from 'rambda'

interface Output {
  a: number,
  foo: {bar: number},
}

describe('R.assocPath - user must explicitly set type of output', () => {
  it('with array as path input', () => {
    const result = assocPath<Output>(['foo', 'bar'], 2, {a: 1})

    result // $ExpectType Output
  })
  it('with string as path input', () => {
    const result = assocPath<Output>('foo.bar', 2, {a: 1})

    result // $ExpectType Output
  })
})

describe('R.assocPath - curried', () => {
  it('with array as path input', () => {
    const result = assocPath<Output>(['foo', 'bar'], 2)({a: 1})

    result // $ExpectType Output
  })
  it('with string as path input', () => {
    const result = assocPath<Output>('foo.bar', 2)({a: 1})

    result // $ExpectType Output
  })
})

binary

Try this R.binary example in Rambda REPL

bind

bind<F extends AnyFunction, T>(fn: F, thisObj: T): (...args: Parameters<F>) => ReturnType<F>

Creates a function that is bound to a context.

Try this R.bind example in Rambda REPL

bind<F extends AnyFunction, T>(fn: F, thisObj: T): (...args: Parameters<F>) => ReturnType<F>;
bind<F extends AnyFunction, T>(fn: F): (thisObj: T) => (...args: Parameters<F>) => ReturnType<F>;

import { curryN } from './curryN.js'

export function bind(fn, thisObj){
  if (arguments.length === 1){
    return _thisObj => bind(fn, _thisObj)
  }

  return curryN(fn.length, (...args) => fn.apply(thisObj, args))
}

import { bind } from './bind.js'

function Foo(x){
  this.x = x
}
function add(x){
  return this.x + x
}
function Bar(x, y){
  this.x = x
  this.y = y
}
Bar.prototype = new Foo()
Bar.prototype.getX = function (){
  return 'prototype getX'
}

test('returns a function', () => {
  expect(typeof bind(add)(Foo)).toBe('function')
})

test('returns a function bound to the specified context object', () => {
  const f = new Foo(12)
  function isFoo(){
    return this instanceof Foo
  }
  const isFooBound = bind(isFoo, f)
  expect(isFoo()).toBeFalse()
  expect(isFooBound()).toBeTrue()
})

test('works with built-in types', () => {
  const abc = bind(String.prototype.toLowerCase, 'ABCDEFG')
  expect(typeof abc).toBe('function')
  expect(abc()).toBe('abcdefg')
})

test('works with user-defined types', () => {
  const f = new Foo(12)
  function getX(){
    return this.x
  }
  const getXFooBound = bind(getX, f)
  expect(getXFooBound()).toBe(12)
})

test('works with plain objects', () => {
  const pojso = { x : 100 }
  function incThis(){
    return this.x + 1
  }
  const incPojso = bind(incThis, pojso)
  expect(typeof incPojso).toBe('function')
  expect(incPojso()).toBe(101)
})

test('does not interfere with existing object methods', () => {
  const b = new Bar('a', 'b')
  function getX(){
    return this.x
  }
  const getXBarBound = bind(getX, b)
  expect(b.getX()).toBe('prototype getX')
  expect(getXBarBound()).toBe('a')
})

test('preserves arity', () => {
  const f0 = function (){
    return 0
  }
  const f1 = function (a){
    return a
  }
  const f2 = function (a, b){
    return a + b
  }
  const f3 = function (
    a, b, c
  ){
    return a + b + c
  }

  expect(bind(f0, {})).toHaveLength(0)
  expect(bind(f1, {})).toHaveLength(1)
  expect(bind(f2, {})).toHaveLength(2)
  expect(bind(f3, {})).toHaveLength(3)
})

import {bind} from 'rambda'

class Foo {}
function isFoo<T = any>(this: T): boolean {
  return this instanceof Foo
}

describe('R.bind', () => {
  it('happy', () => {
    const foo = new Foo()
    const result = bind(isFoo, foo)()

    result // $ExpectType boolean
  })
})

both

both(pred1: Pred, pred2: Pred): Pred

It returns a function with input argument.

This function will return true, if both firstCondition and secondCondition return true when input is passed as their argument.

Try this R.both example in Rambda REPL

both(pred1: Pred, pred2: Pred): Pred;
both<T>(pred1: Predicate<T>, pred2: Predicate<T>): Predicate<T>;
both<T>(pred1: Predicate<T>): (pred2: Predicate<T>) => Predicate<T>;
both(pred1: Pred): (pred2: Pred) => Pred;

export function both(f, g){
  if (arguments.length === 1) return _g => both(f, _g)

  return (...input) => f(...input) && g(...input)
}

import { both } from './both.js'

const firstFn = val => val > 0
const secondFn = val => val < 10

test('with curry', () => {
  expect(both(firstFn)(secondFn)(17)).toBeFalse()
})

test('without curry', () => {
  expect(both(firstFn, secondFn)(7)).toBeTrue()
})

test('with multiple inputs', () => {
  const between = function (
    a, b, c
  ){
    return a < b && b < c
  }
  const total20 = function (
    a, b, c
  ){
    return a + b + c === 20
  }
  const fn = both(between, total20)
  expect(fn(
    5, 7, 8
  )).toBeTrue()
})

test('skip evaluation of the second expression', () => {
  let effect = 'not evaluated'
  const F = function (){
    return false
  }
  const Z = function (){
    effect = 'Z got evaluated'
  }
  both(F, Z)()

  expect(effect).toBe('not evaluated')
})

import {both} from 'rambda'

describe('R.both', () => {
  it('with passed type', () => {
    const fn = both<number>(
      x => x > 1,
      x => x % 2 === 0
    )
    fn // $ExpectType Predicate<number>
    const result = fn(2) // $ExpectType boolean
    result // $ExpectType boolean
  })
  it('with passed type - curried', () => {
    const fn = both<number>(x => x > 1)(x => x % 2 === 0)
    fn // $ExpectType Predicate<number>
    const result = fn(2)
    result // $ExpectType boolean
  })
  it('no type passed', () => {
    const fn = both(
      x => {
        x // $ExpectType any
        return x > 1
      },
      x => {
        x // $ExpectType any
        return x % 2 === 0
      }
    )
    const result = fn(2)
    result // $ExpectType boolean
  })
  it('no type passed - curried', () => {
    const fn = both((x: number) => {
      x // $ExpectType number
      return x > 1
    })((x: number) => {
      x // $ExpectType number
      return x % 2 === 0
    })
    const result = fn(2)
    result // $ExpectType boolean
  })
})

call

Try this R.call example in Rambda REPL

chain

chain<T, U>(fn: (n: T) => U[], list: T[]): U[]

The method is also known as flatMap.

Try this R.chain example in Rambda REPL

chain<T, U>(fn: (n: T) => U[], list: T[]): U[];
chain<T, U>(fn: (n: T) => U[]): (list: T[]) => U[];

export function chain(fn, list){
  if (arguments.length === 1){
    return _list => chain(fn, _list)
  }

  return [].concat(...list.map(fn))
}

import { chain as chainRamda } from 'ramda'

import { chain } from './chain.js'

const duplicate = n => [ n, n ]

test('happy', () => {
  const fn = x => [ x * 2 ]
  const list = [ 1, 2, 3 ]

  const result = chain(fn, list)

  expect(result).toEqual([ 2, 4, 6 ])
})

test('maps then flattens one level', () => {
  expect(chain(duplicate, [ 1, 2, 3 ])).toEqual([ 1, 1, 2, 2, 3, 3 ])
})

test('maps then flattens one level - curry', () => {
  expect(chain(duplicate)([ 1, 2, 3 ])).toEqual([ 1, 1, 2, 2, 3, 3 ])
})

test('flattens only one level', () => {
  const nest = n => [ [ n ] ]
  expect(chain(nest, [ 1, 2, 3 ])).toEqual([ [ 1 ], [ 2 ], [ 3 ] ])
})

test('can compose', () => {
  function dec(x){
    return [ x - 1 ]
  }
  function times2(x){
    return [ x * 2 ]
  }

  const mdouble = chain(times2)
  const mdec = chain(dec)
  expect(mdec(mdouble([ 10, 20, 30 ]))).toEqual([ 19, 39, 59 ])
})

test('@types/ramda broken test', () => {
  const score = {
    maths   : 90,
    physics : 80,
  }

  const calculateTotal = score => {
    const { maths, physics } = score

    return maths + physics
  }

  const assocTotalToScore = (total, score) => ({
    ...score,
    total,
  })

  const calculateAndAssocTotalToScore = chainRamda(assocTotalToScore,
    calculateTotal)
  expect(() =>
    calculateAndAssocTotalToScore(score)).toThrowErrorMatchingInlineSnapshot('"fn(...) is not a function"')
})

import {chain} from 'rambda'

const list = [1, 2, 3]
const fn = (x: number) => [`${x}`, `${x}`]

describe('R.chain', () => {
  it('without passing type', () => {
    const result = chain(fn, list)
    result // $ExpectType string[]

    const curriedResult = chain(fn)(list)
    curriedResult // $ExpectType string[]
  })
})

clamp

Restrict a number input to be within min and max limits.

If input is bigger than max, then the result is max.

If input is smaller than min, then the result is min.

Try this R.clamp example in Rambda REPL

clone

It creates a deep copy of the input, which may contain (nested) Arrays and Objects, Numbers, Strings, Booleans and Dates.

<a title="red