I ❤ Ramda - Partial Application with Curry

May 11, 2018 0 Comments

I ❤ Ramda - Partial Application with Curry

 

 

The I ❤ Ramda series will differ slightly from others you may have seen. Instead of using the typical foo, bar, baz examples, or articles demonstrating isolated functions, this series aims to showcase real world applications of Ramda.js.

This is the first article and will focus Partial Application with Curry.

Partial Application with Bind

If you have used bind, you may already be familiar with the concept of Partial Application.

const multiply = (a, b) => a  b // partially apply 2 to multiply to create double 
const double = multiply.bind(null, 2) double(3) //=> 6

Using bind we were able to partially apply the value 2 to the argument a in the function multiply to create a completely new function double.

The above solution (with bind) works just fine, even if we do have to call bind and also pass in the context (in this case it's null).

Partial Application with Curry

The Ramda way of handling Partial Application would be to Curry the multiply function.

import curry from 'ramda/src/curry' const multiply = curry((a, b) => a  b) // partially apply 2 to multiply to create double 
const double = multiply(2) double(3) //=> 6

Okay, so now the code is... different, but you're not excited, and I get that. But hang on because the 🧙 magic is just about to start.

Where bind falls short

Let's consider this example...

import fs from 'fs' const output = (err, data) => { if (err) return console.log('ERROR:', err) console.log(data) 
} // How do I partially apply utf8?
fs.readFile('README.md', 'utf8', output)
fs.readFile('package.json', 'utf8', output)

All calls to readFile are going to be utf8 and now that I think about it, they are all going to use the same output

function too.

For this example bind cannot be used because bind will only apply arguments from left to right. A completely new function would have to be created.

// Harrumph! 
const outputUtf8File = file => fs.readFile(file, 'utf8', output)

Special Placeholder

Ramda has what they refer to as a Special Placeholder which is used to to partially apply arguments regardless of position.

To use this Special Placeholder, the function must first be curried.

import curry from 'ramda/src/curry' 
import from 'ramda/src/' // Special Placeholder
import fs from 'fs' // curry fs.readFile to add the magic
const readFile = curry(fs.readFile)

Now that fs.readFile has been curried into readFile, we can use Ramda's Special Placeholder () for the filename argument and partially apply, utf8 and output to readFile.

// partially apply utf8 and output. 
const outputUtf8File = readFile(
, 'utf8', output) outputUtf8File('README.md')
outputUtf8File('package.json')

😍 If you are not excited at this point, go see a doctor because you dead inside.

Let's finish this example up with some refactoring.

// let's refactor this: 
outputUtf8File('README.md')
outputUtf8File('package.json') // into this:
const files = ['README.md', 'package.json']
files.map(outputUtf8File)

readFile can also be used like this because now we have more options!

files.map(readFile(, 'utf8', output)) 

Homework

The best way to learn is to do 😉. So for this excercise try to create two new functions fetchGet and fetchPost based off of the function fetch.

import curry from 'ramda/src/curry' 
import
from 'ramda/src/'
import fetch from 'node-fetch' // might need this in node.js land. const url = '???' // pick a url to test // fetch(url, { method: 'GET' }) const fetchGet = /* ??? /
const fetchPost = /
??? */

If you need help, post questions in the comments section, I'll do my best to answer every question!

What else can we do?

Here's a simplified snippet from a project I am working on. First I'll stub out the tokenContract so you can also run this example.

Again, don't read too much into this part of the code, it's just the setup. But notice to how the curry function is used below.

import curry from 'ramda/src/curry' 
import
from 'ramda/src/__' // example stub so you can run the code below
const tokenContract = { getTokenIndex: curry((account, tokenId) => account === 1 && tokenId === 123 ? Promise.resolve(1000) : Promise.reject()), getTokenAtIndex: curry((account, index) => account === 1 && index === 1000 ? Promise.resolve('success') : Promise.reject('fail')),
}

Now the meat 🍖

const account = 1 // without partial application 
tokenContract.getTokenIndex(account, 123) //=> then 1000 .then(i => tokenContract.getTokenAtIndex(account, i)) //=> then "success" // with partial application
tokenContract.getTokenIndex(account, 123) //=> then 1000 .then(tokenContract.getTokenAtIndex(account)) //=> then "success"

Partial Application allowed us to trim i => and , i off the ends of our function inside then. This is a pattern you

will see often. The single argument supplied from the then function (in this case it's i) will become the final

argument of the function to be called in tokenContract.getTokenAtIndex.

And since I know the work I am doing will always be for account, I could apply the account value to both functions.

const getToken = tokenContract.getTokenIndex(account) 
const getTokenAtIndex = tokenContract.getTokenAtIndex(account) getToken(123) //=> then 1000 .then(getTokenAtIndex) //=> then "success"

Extra credit

Function composition will be covered in future articles, but right now I'd like leave this as extra credit for you

to research and tinker with:

const account = 1 const getTokenIndex = tokenContract.getTokenIndex(account) 
const getTokenAtIndex = tokenContract.getTokenAtIndex(account) // Ramda's pipeP asynchronous function composer
const getToken = pipeP( getTokenIndex, getTokenAtIndex,
) getToken(123) //=> then "success"

Summary

This article briefly covered Partial Application, Currying, and even Function Composition. Head over to Ramda to learn more. Ask questions in the comments, I will respond to every question!

If you found this interesting, I have more articles on Medium and dev.to.

Subscribe on Twitter, Medium, dev.to and steemit or joel.net to be notified of future articles.

subscribe to @joelnet

originally posted here

Cheers!

license CC BY 4.0


Tag cloud