Skip to content

dtinth/it.js

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

41 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

it.js Build Status

it.js is a library to make it easier to create accessor/iterator functions, for use with things like _.map, _.filter, _.sortBy, _.each, and so on... It complements nicely with Underscore.js (but it does not depend on, and can be used without underscore).

This library is inspired by:

See also: my blog post about cool things you can do with it.js (a more concise, less complete version of this example)

tldr

In short, this library allows you to write this:

It.send('toLowerCase')

instead of this:

function(name) { return name.toLowerCase() }

And this:

It.instantiate(Status)

instead of this:

function(json) { return new Status(json) }

They are chainable, so you can also write things like this:

It.get('first').get('length')

instead of this:

function(person) { return person.first.length }

And a lot more functionalities (see the quick reference). Thanks to awesome functional programming and metaprogramming stuff, this library is just around ~200 lines of code in a single file, and ~3k minified.

Installation

npm:

npm install it.js

Node:

var It = require('it.js')

Bower:

bower install dtinth/it.js

Browser:

<script src="path/to/it.js"></script>

Quick Reference

It                             -> return it
It.self                        -> return this

.get('foo')                    -> return it.foo
.send/.invoke('foo', ...args)  -> return it.foo(...args)
.fcall(...args)                -> return it(...args)
.set/.put('foo', 'bar')        -> it.foo = 'bar'; return it
.del('foo')                    -> delete it.foo; return it

.or(defaultValue)              -> return it || defaultValue
.instantiate(Klass)            -> return new Klass(it)
.tap(fun)                      -> fun(it); return it
.post('foo', [...args])        -> return it.foo(...args)
.fapply([...args])             -> return it(...args)

.maybe(fun)                    -> return it && fun(it)
.maybe('foo')                  -> return it && it.foo
.not(fun)                      -> return !fun(it)
.not('foo')                    -> return !it.foo

.splat(fun)                    -> return Array.prototype.map.call(it, fun)
.splat/.pluck('foo')           -> return Array.prototype.map.call(it, It.get('foo'))
.select/.filter(fun)           -> return Array.prototype.filter.call(it, fun)
.select/.filter('foo')         -> return Array.prototype.filter.call(it, It.get('foo'))
.reduce('foo')                 -> return Array.prototype.reduce.call(it, It.get('foo'))

['op'](value)                  -> return it op value
['op'](a, b)                    = a op b
     (where op === == !== != > >= < <= + - * /)

example.js

var _ = require('underscore')
var It = require('./')
var numbers = [3, 1, 4, 1, 5]
var strings = ['this', 'is', 'a', 'Book']

It

It provides an identity function, just like _.identity, but much shorter.

What you pass to it, you get it back!

console.log(It(123))
console.log(It('hello!'))
123
hello!

This code maps an array with itself... Pretty useless

console.log(_.map(numbers, It))
[ 3, 1, 4, 1, 5 ]

This code gets a sorted copy of an array. We just sort by itself! (This is nice since underscore has only _.sortBy, but not _.sort)

console.log(_.sortBy(numbers, It))
console.log(_.sortBy(strings, It))
[ 1, 1, 3, 4, 5 ]
[ 'Book', 'a', 'is', 'this' ]

.get

.get returns the value of a property. Here's where things get interesting...

Let's create a function that returns the length of a string (or an array, or whatever that has .length property).

// equivalent to function(x) { return x.length }
var getLength = It.get('length')

Now, getLength is a function that returns the length property.

console.log(getLength('this is a string'))
console.log(getLength(strings))
16
4

We can use it to sort the strings by their length.

console.log(_.sortBy(strings, getLength))
[ 'a', 'is', 'this', 'Book' ]

.send / .invoke

Use .send(...) to call a method on an object.

// equivalent to function(x) { return x.toUpperCase() }
var upcase = It.send('toUpperCase')

Now, upcase is a function that takes something and calls the toUpperCase() method on it.

console.log(upcase('this is a string'))
THIS IS A STRING

With this, we can map all these strings to uppercase:

console.log(_.map(strings, upcase))
[ 'THIS', 'IS', 'A', 'BOOK' ]

And with this, case-insensitive sorting is easy:

console.log(_.sortBy(strings, upcase))
[ 'a', 'Book', 'is', 'this' ]

.splat / .pluck

Use .splat(function) to map an array over that function.

// equivalent to function(x) { return Array.prototype.map.call(x, upcase) }
var upcaseAll = It.splat(upcase)
console.log(upcaseAll(strings))
[ 'THIS', 'IS', 'A', 'BOOK' ]

Note that .pluck('foo') is equivalent to .splat(It.get('foo'))

Chaining

Of course, all of these are chainable.

// equivalent to function(x) { return x.substr(0, 1).toUpperCase() }
var firstCharacterCapitalized = It.send('substr', 0, 1).send('toUpperCase')

Get the first character of each string, capitalized.

console.log(_.map(strings, firstCharacterCapitalized))
[ 'T', 'I', 'A', 'B' ]

Now, let's move on and try to use chaining to do something more practical...

Here we have a list of people. (name generation thanks to chance.js)

var addressBook = [
  { first: 'Sifwa',    last: 'Duhav', phone: '(416) 984-4454' },
  { first: 'Moc',                     phone: '(898) 983-5755' },
  { first: 'Diblacbo', last: 'Li',    phone: '(258) 838-8314' },
  { first: 'Betu',     last: 'Jol',   phone: '(219) 234-9591' },
  { first: 'Fuhetu',   last: 'Ra',    phone: '(631) 437-2332' }
]

Let's sort them by the length of first name!

// equivalent to function(x) { return x.first.length }
var firstNameLength = It.get('first').get('length')
console.log(_.sortBy(addressBook, firstNameLength))
[ { first: 'Moc', phone: '(898) 983-5755' },
  { first: 'Betu', last: 'Jol', phone: '(219) 234-9591' },
  { first: 'Sifwa', last: 'Duhav', phone: '(416) 984-4454' },
  { first: 'Fuhetu', last: 'Ra', phone: '(631) 437-2332' },
  { first: 'Diblacbo', last: 'Li', phone: '(258) 838-8314' } ]

.set / .put

.set(property, value) sets a property on an object. The result of this operation will be the invoked object, so you can chain more operations (something like .set('a','b').set('c','d')).

Let's set everyone's score to zero! Yes, scores in an address book!

_.each(addressBook, It.set('score', 0))
console.log(addressBook)
[ { first: 'Sifwa',
    last: 'Duhav',
    phone: '(416) 984-4454',
    score: 0 },
  { first: 'Moc', phone: '(898) 983-5755', score: 0 },
  { first: 'Diblacbo',
    last: 'Li',
    phone: '(258) 838-8314',
    score: 0 },
  { first: 'Betu', last: 'Jol', phone: '(219) 234-9591', score: 0 },
  { first: 'Fuhetu', last: 'Ra', phone: '(631) 437-2332', score: 0 } ]

.not

.not(func) invokes a passed function with current value, and returns the logical NOT of the result. If you don't put in any function, .not() is equivalent to .not(It).

For example, one of the contact in the address book above does not have a last name. Let's find out who.

// these three are equivalent to function(x) { return !x.last }
console.log(_.select(addressBook, It.not(It.get('last'))))
console.log(_.select(addressBook, It.get('last').not()))
console.log(_.select(addressBook, It.not('last')))
[ { first: 'Moc', phone: '(898) 983-5755', score: 0 } ]
[ { first: 'Moc', phone: '(898) 983-5755', score: 0 } ]
[ { first: 'Moc', phone: '(898) 983-5755', score: 0 } ]

.maybe

.maybe(func) invokes a passed function with current value only if the current value is truthy.

Let's say we want to get everyone's last name, lowercased. Since "Moc" doesn't have a last name, attempt to call .toLowerCase() on undefined will throw an Error. We want to call .toLowerCase() only when we have something to call on.

// equivalent to function(x) { return x.last && x.last.toLowerCase() }
var lastNameLowered = It.get('last').maybe(It.send('toLowerCase'))
console.log(_.map(addressBook, lastNameLowered))
[ 'duhav', undefined, 'li', 'jol', 'ra' ]

Then you can filter out falsy value by using _.filter(..., It).

console.log(_.filter(_.map(addressBook, lastNameLowered), It))
[ 'duhav', 'li', 'jol', 'ra' ]

Note that .maybe('foo') is a shorthand for .maybe(It.get('foo')).

.or

Instead of using .maybe, we can use .or to put a default value.

var lastNameLowered2 = It.get('last').or('None').send('toLowerCase')
console.log(_.map(addressBook, lastNameLowered2))
[ 'duhav', 'none', 'li', 'jol', 'ra' ]

.instantiate

.instantiate(Constructor) can be used to quickly map things into an instance.

Here we have a Person class.

function Person(info) {
  this.info = info
}
Person.prototype.getName = function() {
  return this.info.first + ' ' + this.info.last
}
Person.prototype.greet = function() {
  console.log('Hello! I am "' + this.getName() + '"')
}

We can map everyone in the address book into a new Person instance!

// equivalent to function(x) { return new Person(x) }
var people = _.map(addressBook, It.instantiate(Person))
_.each(people, It.send('greet'))
Hello! I am "Sifwa Duhav"
Hello! I am "Moc undefined"
Hello! I am "Diblacbo Li"
Hello! I am "Betu Jol"
Hello! I am "Fuhetu Ra"

It.self

You can use It.self instead of It to create a function that uses the value of this instead of the value of passed argument.

You can use it to quickly make an accessor function

Person#getFirstName returns the first name.

// equivalent to function() { return this.info.first }
Person.prototype.getFirstName = It.self.get('info').get('first')

This function takes a last name, and returns a name suffix. No need to check of null here, we'll let .maybe do it.

function initial(string) {
  return ' ' + string.substr(0, 1) + '.'
}

Person#getLastInitial returns the initial of last name. If the person does not have last name, then return empty string.

// equivalent to function() { return (this.info.last && initial(this.info.last)) || '' }
Person.prototype.getLastInitial = It.self.get('info').get('last').maybe(initial).or('')

We can then redefine the getName method to make use of them:

Person.prototype.getName = function() {
  return this.getFirstName() + this.getLastInitial()
}
_.each(people, It.send('greet'))
Hello! I am "Sifwa D."
Hello! I am "Moc"
Hello! I am "Diblacbo L."
Hello! I am "Betu J."
Hello! I am "Fuhetu R."

.compose

You can use .compose to compose your own functionality. In fact, most of the functionality are in this library is built on top of .compose.

Here we have these vectors...

var vectors = [ [1, 5], [5, 1], [2, -3] ]

We also have a square function...

function square(x) {
  return x * x
}

Let's get the square of x and y components of these vectors!

console.log(_.map(vectors, It.get(0).compose(square)))
console.log(_.map(vectors, It.get(1).compose(square)))
[ 1, 25, 4 ]
[ 25, 1, 9 ]

You can also use .compose to chain functions together.

var test = { a: { b: 1 }, b: { a: 2 } }
var getA = It.get('a')
var getB = It.get('b')
var getAB = getA.compose(getB)
var getBA = getB.compose(getA)
console.log(test)
console.log(getA(test))
console.log(getB(test))
console.log(getAB(test))
console.log(getBA(test))
{ a: { b: 1 }, b: { a: 2 } }
{ b: 1 }
{ a: 2 }
1
2

Let's bring back that array of Person instances, and the firstNameLength function we created earlier before.

The firstNameLength function works with JSON data, not Person instances. Luckily, the Person class stores the original JSON data in the .info property. So we can still sort these people by their first name length.

_.each(_.sortBy(people, It.get('info').compose(firstNameLength)), It.send('greet'))
Hello! I am "Moc"
Hello! I am "Betu J."
Hello! I am "Sifwa D."
Hello! I am "Fuhetu R."
Hello! I am "Diblacbo L."

Here we have a list of things, initially sorted alphabetically...

var things = [
  { name: 'Apple',          type: 'fruit' },
  { name: 'CoffeeScript',   type: 'language' },
  { name: 'Cat',            type: 'animal' },
  { name: 'Dog',            type: 'animal' },
  { name: 'Guava',          type: 'fruit' },
  { name: 'JavaScript',     type: 'language' },
  { name: 'Mountain Lion',  type: 'animal' },
  { name: 'Pineapple',      type: 'fruit' },
  { name: 'Ruby',           type: 'language' } ]

Let's say we want to order them. We want languages first, then we want animals, and finally we want fruits. In that order.

Thanks to .compose and _.partial, this is easy!

var where = _.partial(_.indexOf, ['language', 'animal', 'fruit'])
console.log(_.sortBy(things, It.get('type').compose(where)))
[ { name: 'CoffeeScript', type: 'language' },
  { name: 'JavaScript', type: 'language' },
  { name: 'Ruby', type: 'language' },
  { name: 'Cat', type: 'animal' },
  { name: 'Dog', type: 'animal' },
  { name: 'Mountain Lion', type: 'animal' },
  { name: 'Apple', type: 'fruit' },
  { name: 'Guava', type: 'fruit' },
  { name: 'Pineapple', type: 'fruit' } ]

['==='], ['=='], ['!=='], ['!='], ['>'], ['>='], ['<'], ['<='], ['+'], ['-'], ['*'], ['/'], ['%']

These functions can be used to check against the given value...

There are so many different conventions on how to name each of these functions.

  • Should I use equal, eq, eql, equals, strictEqual, or strictlyEquals for ===? How about ==?
  • add or plus? gt, above, greaterThan, more, moreThan?

After I think about them, I think that these names will only add more confusion. So I just use the original operator name, which are symbols.

var addOne = It['+'](1)
console.log(addOne(41))
42

These "operator" functions are a little bit special. If you pass 2 parameters to these functions, it will calculate the answer right away.

console.log(It['+'](41, 1))
42

...so that we can put them into _.reduce:

console.log(_.reduce([1,2,3,4,5], It['+']))
15

.select / .filter

Just like .splat(fun) that runs [].map(fun) over passed array, .select(fun) runs [].filter(fun) over passed array.


Using ['%'], we can generate even and odd functions quickly.

var odd = It['%'](2)
var even = It.not(odd)

Then we can filter the even and odd numbers.

var onlyEven = It.select(even)
var onlyOdd = It.select(odd)
console.log(onlyEven(numbers))
console.log(onlyOdd(numbers))
[ 4 ]
[ 3, 1, 1, 5 ]

From the above list of things, let's get all the languages.

var getLanguageNames = It.select(It.get('type')['==']('language')).pluck('name')
console.log(getLanguageNames(things))
[ 'CoffeeScript', 'JavaScript', 'Ruby' ]

.reduce

Yeah. Just like .splat and .select...

var sum = It.reduce(It['+'])
console.log(sum([1,2,3,4,5]))
15

.tap

.tap invokes the passed function with the current value, and returns the current value.

log the numbers and while mapping to get the squares

console.log(numbers)
console.log(_.map(numbers, It.tap(console.log).compose(square)))
[ 3, 1, 4, 1, 5 ]
3
1
4
1
5
[ 9, 1, 16, 1, 25 ]

make everyone greet while mapping to get their first name

console.log(_.map(people, It.tap(It.send('greet')).send('getFirstName')))
Hello! I am "Sifwa D."
Hello! I am "Moc"
Hello! I am "Diblacbo L."
Hello! I am "Betu J."
Hello! I am "Fuhetu R."
[ 'Sifwa', 'Moc', 'Diblacbo', 'Betu', 'Fuhetu' ]

Performance

While these generated functions are fast (only a little function call overhead), the process of creating these functions is expensive.

For best performance, you can generate the functions you want to use ahead of time, and just use them, instead of generating these functions on the fly. As a simple guideline, don't generate these functions inside a loop.

A very simple benchmark is included in benchmark.js to illustrate the point.

License

MIT Licensed

About

Chainable object-oriented functional combinators

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published