Iterate over promises, promise-returning or async/await functions in series or parallel. Support settle (fail-fast), concurrency (limiting) and hooks system (start, beforeEach, afterEach, finish)

Table of Contents

• Install
• Usage
• Background
  • Why not "promise fun"?
  • Why not separate libs?
• API
  • .serial
  • .parallel
  • .each
• Options
• Hooks
• Item
• Finish hook
• Related
• Contributing
• Building docs
• Running tests
• Author
• Logo
• License

(TOC generated by verb using markdown-toc)

Install

Install with npm

$ npm install each-promise --save

or install using yarn

$ yarn add each-promise

Usage

For more use-cases see the tests

const eachPromise = require('each-promise')
const arr = [
  123,
  'foo',
  () => 456,
  Promise.resolve(567)
  false,
  () => Promise.resolve(11)
]

eachPromise
  .serial(arr)
  .then((res) => {
    console.log(res) // => [123, 'foo', 456, 567, false, 11]
  })

Background

You may think why this exists, what is this for, why not Sindre's microlibs like p-map, p-map-series, p-settle, p-each-series or p-reduce.

Why not "promise fun"?

They do their jobs okey, but in some cases they don't. And that's the my case. I need control over "fast fail" behavior, also known as "settle" or "bail". I need serial and parallel iteration, but parallel with concurrency too. They requires node v4, and uses native Promise constructor. I believe in that we should not use modern things if we don't need them, it is just syntax sugar. This package is written in way that works in node versions below v4 and also you can pass custom Promise constructor through options.Promise if you want.

• node@4 required
• no hooks system
• no settle / fail-fast / bail
• no custom Promise
• no real and meaningful tests
• concurrency control

back to top

Why not separate libs?

Why not separate .serial and .parallel into own libs like Sindre did? Because the main core logic and difference is absolutely in just 2-3 lines of code and one if check. The main thing is that parallel uses for loop with concurrency combination, and series does not use loops, but recursive function calls.

For free you get hooks system. And really it cost nothing. It just able to be done, because the structure of the code and because I need such thing.

• node v0.10 and above
• custom Promise constructor
• real settle / fail fast
• hook system, through options
• very stable and well tested with real tests
• concurrency control

back to top

API

.serial

Iterate over iterable in series (serially) with optional opts (see options section) and optional mapper function (see item section).

Params

• <iterable> {Array}: iterable object like array with any type of values
• [mapper] {Function}: function to apply to each item in iterable, see item section
• [opts] {Object}: see options section
• returns {Promise}: Always resolved or rejected promise

Example

var delay = require('delay')
var eachPromise = require('each-promise')

var arr = [
  () => delay(500).then(() => 1),
  () => delay(200).then(() => { throw Error('foo') }),
  () => delay(10).then(() => 3),
  () => delay(350).then(() => 4),
  () => delay(150).then(() => 5)
]

eachPromise
  .serial(arr)
  .then((res) => {
    console.log(res) // [1, Error: foo, 3, 4, 5]
  })

// see what happens when parallel
eachPromise
  .parallel(arr)
  .then((res) => {
    console.log(res) // => [3, 5, Error: foo, 4, 1]
  })

// pass `settle: false` if you want
// to stop after first error
eachPromise
  .serial(arr, { settle: false })
  .catch((err) => console.log(err)) // => Error: foo

.parallel

Iterate concurrently over iterable in parallel (support limiting with opts.concurrency) with optional opts (see options section) and optional mapper function (see item section).

Params

• <iterable> {Array}: iterable object like array with any type of values
• [mapper] {Function}: function to apply to each item in iterable, see item section
• [opts] {Object}: see options section
• returns {Promise}: Always resolved or rejected promise

Example

var eachPromise = require('each-promise')

var arr = [
  function one () {
    return delay(200).then(() => {
      return 123
    })
  },
  Promise.resolve('foobar'),
  function two () {
    return delay(1500).then(() => {
      return 345
    })
  },
  delay(10).then(() => 'zero'),
  function three () {
    return delay(400).then(() => {
      coffffnsole.log(3) // eslint-disable-line no-undef
      return 567
    })
  },
  'abc',
  function four () {
    return delay(250).then(() => {
      return 789
    })
  },
  function five () {
    return delay(100).then(() => {
      sasasa // eslint-disable-line no-undef
      return 444
    })
  },
  function six () {
    return delay(80).then(() => {
      return 'last'
    })
  }
]

// does not stop after first error
// pass `settle: false` if you want
eachPromise
  .parallel(arr)
  .then((res) => {
    console.log(res)
    // => [
    //   'foobar',
    //   'abc',
    //   'zero',
    //   'last',
    //   ReferenceError: sasasa is not defined,
    //   123,
    //   789,
    //   ReferenceError: coffffnsole is not defined
    //   345
    // ]
  })

.each

Iterate over iterable in series or parallel (default), depending on default opts. Pass opts.serial: true if you want to iterate in series, pass opts.serial: false or does not pass anything for parallel.

Params

• <iterable> {Array}: iterable object like array with any type of values
• [mapper] {Function}: function to apply to each item in iterable, see item section
• [opts] {Object}: see options section
• returns {Promise}: Always resolved or rejected promise

Example

var delay = require('delay')
var eachPromise = require('each-promise')

var arr = [
  123,
  function () {
    return delay(500).then(() => 456)
  },
  Promise.resolve(678),
  function () {
    return 999
  },
  function () {
    return delay(200).then(() => 'foo')
  }
]

eachPromise
  .each(arr)
  .then(function (res) {
    console.log('done', res) // => [123, 678, 999, 'foo', 456]
  })

Options

You have control over everything, through options.

• Promise {Function}: custom Promise constructor to be used, defaults to native
• mapper {Function}: function to apply to each item in iterable, see item section
• settle {Boolean}: if false stops after first error (also known as "fail-fast" or "bail"), default true
• flat {Boolean}: result array to contain only values, default true
• concurrency {Number}: works only with .parallel method, defaults to iterable length
• start {Function}: on start hook, see hooks section
• beforeEach {Function}: called before each item in iterable, see hooks section
• afterEach {Function}: called after each item in iterable, see hooks section
• finish {Function}: called at the end of iteration, see hooks section
• context {Object}: custom context to be passed to each fn in iterable
• args {Array}: custom argument(s) to be pass to fn, given value is arrayified

back to top

Hooks

You can do what you want between stages through hooks - start, before each, after each, finish.

• start {Function}: called at the start of iteration, before anything
• beforeEach {Function}: passed with item, index, arr arguments
  • item is an object with value, reason and index properties, see item section
  • index is the same as item.index
  • arr is the iterable object - array or object
• afterEach {Function}: passed with item, index, arr arguments
  • item is an object with value, reason and index properties, see item section
  • index is the same as item.index
  • arr is the iterable object - array or object
• finish {Function}: called at the end of iteration, see finish hook section

back to top

Item

That object is special object, that is passed to beforeEach and afterEach hooks, also can be found in result object if you pass opts.flat: false option. And passed to opts.mapper function too.

• item.value resolved/rejected promise value, if at beforeEach hook it can be function
• item.reason may not exist if item.value, if exist it is standard Error object
• item.index is number, order of "executing", not the order that is defined in iterable

back to top

Finish hook

This hooks is called when everything is finished / completed. At the very end of iteration. It is passed with err, result arguments where:

• err is an Error object, if opts.settle: false, otherwise null
• result is always an array with values or item objects if opts.flat: false

back to top

Related

• always-done: Handle completion and errors with elegance! Support for streams, callbacks, promises, child processes, async/await and sync functions. A… more | homepage
• minibase-create-plugin: Utility for minibase and base that helps you create plugins | homepage
• minibase-is-registered: Plugin for minibase and base, that adds isRegistered method to your application to detect if plugin is already… more | homepage
• minibase: Minimalist alternative for Base. Build complex APIs with small units called plugins. Works well with most of the… more | homepage
• mukla: Small, parallel and fast test framework with suppport for async/await, promises, callbacks, streams and observables. Targets and works… more | homepage
• try-catch-callback: try/catch block with a callback, used in try-catch-core. Use it when you don't care about asyncness so much… more | homepage
• try-catch-core: Low-level package to handle completion and errors of sync or asynchronous functions, using once and dezalgo libs. Useful… more | homepage

Contributing

Pull requests and stars are always welcome. For bugs and feature requests, please create an issue.
Please read the contributing guidelines for advice on opening issues, pull requests, and coding standards.
If you need some help and can spent some cash, feel free to contact me at CodeMentor.io too.

In short: If you want to contribute to that project, please follow these things

1. Please DO NOT edit README.md, CHANGELOG.md and .verb.md files. See "Building docs" section.
2. Ensure anything is okey by installing the dependencies and run the tests. See "Running tests" section.
3. Always use npm run commit to commit changes instead of git commit, because it is interactive and user-friendly. It uses commitizen behind the scenes, which follows Conventional Changelog idealogy.
4. Do NOT bump the version in package.json. For that we use npm run release, which is standard-version and follows Conventional Changelog idealogy.

Thanks a lot! :)

Building docs

Documentation and that readme is generated using verb-generate-readme, which is a verb generator, so you need to install both of them and then run verb command like that

$ npm install verbose/verb#dev verb-generate-readme --global && verb

Please don't edit the README directly. Any changes to the readme must be made in .verb.md.

Running tests

Clone repository and run the following in that cloned directory

$ npm install && npm test

Author

Charlike Mike Reagent

• github/tunnckoCore
• twitter/tunnckoCore
• codementor/tunnckoCore

Logo

The logo is Cyclone Emoji from EmojiOne.com. Released under the CC BY 4.0 license.

License

Copyright © 2016-2017, Charlike Mike Reagent. Released under the MIT License.

This file was generated by verb-generate-readme, v0.4.3, on March 19, 2017.
Project scaffolded using charlike cli.