Skip to content

Latest commit

 

History

History
221 lines (142 loc) · 11.8 KB

README.md

File metadata and controls

221 lines (142 loc) · 11.8 KB
Raw

async-hofs

CircleCI

JS Promise & async / await higher order functions utils.

About

This repo contains utilities (mostly higher order functions) intending to help with common asynchronous tasks where async / await or promises are used.

Compatibility

At present, and for the 0.x versions, the entire module expects a standard Promise implementation to be available (Node v4+), and also doesn't itself use async / await in any of it's source - so as to not require transpilation when used as a dependancy (it's intended to be used as a dep for google cloud functions), which at the time of writing this, is pinned to Node v8 LTS.

However from 1.x onward async / await is permitted (although not yet necessarily used) and Node v8.x compatibility is targeted and Node v6.x compatility may break without major version changes.

Setup

To clone run git clone https://github.com/raywhite/async-hofs.

Tests are written using ava, and can be run with npm run test:node. The full test suite includes linting with eslint . (npm run test:lint), and can be run with npm run test.

To install as a dep, run npm install @raywhite/async-hofs --save.

API

compose(..fns) => fn

  • ...fns - (...Function) - any number of functions.
  • fn - (Function) - the composed function.

While async functions are expected, synchronous functions will also be composed. Note that the composed function will alway return a promise. compose will compose functions from right to left.

sequence(...fns) => fn

  • ...fns - (...Function) - any number of functions.
  • fn - (Function) - the composed function.

While async functions are expected, synchronous functions will also be composed. Note that the composed function will alway return a promise. sequence will compose functions from left to right.

  • ...fns (...Function)

memoize(fn, *[stringify], [timeout = -1]) => memoized

  • fn - (Function) - the composed function.
  • stringify - (Function) - the function use to stringify arguments.
  • timeout - (Number) - the time to live, for cached results.
  • memoized - (Function) - the memoized function.

A basic memoizer implementation, except that it will memoize the resolved or rejected values for async functions. The stringify argument will be called with the arguments passed to the memoized function, and by default, will simply call String(value) on the first argument to produce internal cache keys... which assumes the first argument is an primitive value - the user should be carful to provide a more appropriate stringify function where this is not the case.

If a Number is passed instead of a function for stringify, it will be treated as timeout, and any third argument will be ignored.

Where timeout is an absolute Number (in milliseconds), it will determine the time for which the resolution (or rejection) of memoized is cached.

createAsyncFnQueue([concurrency = 1]) => enqueue

  • concurrency - (Number) - how many times to spawn the async function - defaults to 1.
  • enqueue - (Function) - adds a function to the internal queue.

Provides a queue that executes given async functions with a maximum concurrency. Async functions are given by calling the returned function, which returns a promise that resolves or rejects when the function is eventually called.

Rejections do not impact the queue (other given async functions will continue to be called), but the user of the queue is responsible for handling rejections.

createAsyncFnPool(fn, [concurrency = 1]) => pool

  • fn - (Function) - an async function to be invoked - where it requires parameters, used Array.prototype.bind.
  • concurrency - (Number) - how many times to spawn the async function - defaults to 1.
  • args = (...Mixed) - any extra arguments to pass to fn.
  • pool - (Promise)

Wraps an async function, and takes an optional concurrency. fn will be used to create a "green" thread (think of it like a goroutine or something)... and it will limit the concurrency with which that function is called. Consider the following example:

const { createAsyncFnPool } = require('async-hofs')

const sleep = function (value) {
  return new Promise(function (resolve) {
    setTimeout(function () {
        resolve(value)
    }, 100) // Network delay.
  })
}

const inputs = [1, 2, 3, 4, 5, 6]
const outputs = []

const thread = async function () {
  while (inputs.length) {
    const value = await sleep(inputs.shift())
    outputs.push(value)
  }
}

const fn = async function () {
  await createAsyncFnPool(thread, 2)
  console.log.call(console, outputs)
}

fn().catch(console.error.bind(console))

retry(fn, [curve = 2], [limit = 2], [shouldRetry = undefined]) => retrier

Wraps an async function so that it will be attempted limit times before it actually rejects.

Where the wrapped function rejects multiple times (exceeding limit), the error that it finally rejects with will always be value that the last attempt rejected with.

  • fn - (Function) - an async function to be wrapped for retrying.
  • curve - (Function|Number) - the number of times to retry - defaults to 2.
  • limit - (Number) - the number of times to retry - defaults to 2; if falsy, will retry indefinitely.
  • shouldRetry - function (Error) => boolean - an optional function to decide whether to retry; can be used to abort earlier than limit.
  • retrier - (Function) - the wrapped function.

If not present, or a Number is passed, the curve argument will be treated as the limit, and a curve will be generated internally (y = x => 0) so that subsequent attempts are always invoked immediately after a failure.

For ease of use, this module provides some built in helpers for the generation of common curve generators (see createLinear and createExponential below).

NOTE: this method is aliased as createRetrierFn.

createLinear({ m, b }) => line

Intended for use with createRetierFn, to create a function to be used as a linear curve generator, or simply a line.

  • { m, b } - the constants m and b or the gradient and y intercept in the, respectively, in the equation y = m * x + b.
  • line - a function that takes an x value and returns a y value.

createExponential({ a = 2, b = 1 }, m = 1) => curve

Intended for use with createRetrierFn, to create a function to be used as an exponential curve generator.

  • { a, b } - the constants a and b in the equation y = ((a * b) ** x) * m
  • m - the multiplier in the equation y = ((a * b) ** x) * m - to allow scaling between milliseconds and seconds.
  • curve - a function that takes an x value and returns a y value.

mutex([concurrency = 1]) => lock

Given a concurrency, this function will return a lock, which is itself a function that resolves a release function. The lock function will not resolve unless a mutex is available - the user must ensure that they call release whenever work is complete, otherwise any pending lock()s will not resolve.

NOTE: this method is aliased as createCLock and createConcurrencyLock - which are just more verbose names.

  • concurrency - (Number) - the number of concurrent lock()s that may be resolved at any given time.
  • lock - (Function) - allocates a mutex promise.

Consider the following example, where the second async function cannot proceed until release has been called inside the first:

  const lock = mutex(1)

  const values = []

  Promise.all([
    // First promise.
    (async function () {
      const release = await lock()
      await sleep(2000)
      values.push(1)
      release()
    }())

    // Second promise.
    (async function () {
      const release = await lock()
      values.push(2)
      release()
    }())
  ]).then(function () {
    console.log(values) // => [1, 2]
  })

clock(fn, [concurrency = 1]) => clocked

Given a function fn and an optional concurrency, this function will return a version of fn that will schedule invocation so as to allow a maximum of concurrency concurrent invocations of that function. This is intended for use case where you don't want to exceed some memory or IO limit, or create a mutex (for instance to prevent concurrent access to files).

NOTE: this method is aliased as createCLockedFn and createConcurrencyLockedFn - which are just more verbose names.

  • fn - (Function) - an async function to lock / release.
  • concurrency - (Number) - the number of concurrent invocations allowed - defaults to 1.
  • clocked - (Function) - the concurrency locked function.
    • clocked.pending - a getter for the number of invocations pending resolution.
    • clocked.queued - a getter for the number of calls awaiting invocation.

limit(fn, [rate = 1], [interval = 1000]) => limited

Given a function fn and a rate and iterval, the returned version of fn will be rate limited such that invokation will be limited to a maximum of rate calls per any rolling interval period.

  • fn - (Function) - an async function to lock / release.
  • rate - (Number) - the number of concurrent invocations allowed - defaults to 1.
  • interval - (Number) - the interval for the rate limit - defaults to 1000.
  • limited - (Function) - the rate limited function.

NOTE: this method is aliased as createRateLimitedFn - which is just a more verbose name.

benchmark(fn, [precision = 'ms'], [...args]) => res

The returned value (res) is a Promise that resolves with a tuple in the form (time, value) where value is the value resolved by calling fn, and time is the measured execuition time of fn with a precision of precision. Where fn rejects, benchmark itself with reject with the same value ツ.

  • fn - (Function) - the async function to be invoked.
  • precision - (String) - a constant (s|ms|ns) representing the precision of the timing.
  • args - (...Mixed) - extras arguments to pass to the fn invokcation.
  • res - (Array) - the time and value tuple.

buffer(readable, [limit = 1000 * 1024]) => buf

Given a stdlib stream.Readable, this function will continue to read from the stream until the end event is emitted by the stream, and then resolve the returned promise. The returned promise will reject if the limit is exceeded, and will also reject with any errors emitted by the underlying stream.

NOTE: This function will actually consume the stream, meaning that the stream shouldn't also be consumed by another function, unless the event handlers are attached prior to calling buffer. Importantly, buffer itself can't actually consume a stream that is or was being consumed by buffer - so subsequent calls to buffer using the same stream will error.

  • readable - (stream.Readable) - the readable stream to be buffered.
  • limit - (Number) - the max number of bytes to buffer.
  • buf - (Promise) - resolves with the buffer contents.

constant buffer.LIMIT_EXCEEDED

The value of the error message and type upon rejection of the promise returned by buffer where the reason for rejection was exceeding of the limit parameter. Should be used for asserting whether or not this was the type of error.

License

• MIT © Ray White, 2017-2018 •