utilise

Lean JavaScript Utilities as Micro-libraries

• This is not another pure functional utility library. These are just a set of common useful patterns that have evolved working with JS.

• Unfortunately the existing popular libraries (Underscore, Lodash, Ramda, Trine, is, etc) don't result in readable code that is easy to reason about. Instead of putting the data in the first, second, third argument or context variable, this library favours lambda's (partially applied functions) that complement the native Array operations, not compete with them.

For a flavour of what I mean, see:

// clean all js files in current dir, except reserved, logging before deletion
fs.readdirSync(__dirname)
  .filter(includes('js'))
  .filter(not(is.in(core)))
  .map(prepend(__dirname+'/'))
  .map(log('deleting'))
  .map(fs.unlinkSync)

// from mutation observer, redraw all added nodes that are custom elements
mutations
  .map(key('addedNodes'))
  .map(to.arr)
  .reduce(flatten)
  .filter(by('nodeName', includes('-')))
  .map(ripple.draw)

• Each function is in it's own repo. This library just has an automated link to all of them. This has a few benefits:

  • You can npm i utilise and finally write imports like <org>/<repo> in each file.
  • You don't have to make up esoteric names due to lack of <org>/<repo> in npm.
  • You can use utilise.js to import everything (in your application).
  • You can use { utility } = require('utilise/pure') to import what you need + tree shake.
  • You don't have to load a 0.5MB utility library just to use one function.
  • You can be a lot smarter with dead code elimination, even if you include the base file (but not use everything).

• There is no spraying your code with _. everywhere, since these functions are largely first-class additions to the grammar of the language that make your code a lot more fluent.

• These are mostly stable/fixed, a few new ones may still need experimenting with to get the API right.

• A smaller set of high power-to-weight ratio functions are preferred over many, many different functions that do similar things.

• Each micro-library is only just a few lines.

• Each micro-library has 100% coverage. See badges below.

• All browsers (IE >= 9) + node are supported. Tests are run on real browsers using popper.

• There is no polyfilling done here. Recommend using polyfill.io where needed. Some libraries will fail tests (like promise) which assume a native API like Promise, unless you shim first.

API Reference

Please also refer to the respective test.js for more cases and examples.

all

Select all elements based on a CSS selector, piercing shadow boundaries if the browser supports it.

all('li.class')

Narrow search space by also passing in a node

all('li.class', ul)

append

Append something to a string

['lorem', 'ipsum']
  .map(append('-foo')) // returns ['lorem-foo', 'ipsum-foo']

args

Cherry-pick arguments to pass to function by index. This is useful when iterating a list, and invoking a function which may be confused if passed the index and array arguments.

['lorem.txt', 'ipsum.txt']
  .map(args(0)(fs.createWriteStream))

This would fail without args since the second argument (index) would try to be read as the encoding.

You can pick out more than one argument using an array instead of a number.

attr

Get or set value of element attribute.

attr('key')(el)          // returns value for attribute key
attr('key', 'value')(el) // adds [key=value]
attr('key', false)(el)   // removes attribute key

az

Sorts array ascendingly based on the value of a key

array.sort(az('value'))

Since it uses key internally, you can also sort on a deep key.

body

Get the value of a resource from a ripple instance

body(ripple)('users')

This is used internally to avoid any type-specific convienience defaults. Always returns undefined if the resource does not exist. You should probably use ripple('resource') to get the value of a resource in your application.

by

Checks if a property matches a value

users = [ { name: 'foo' }, { name: 'bar' } ]
users
  .filter(by('name', 'foo'))

If the second value parameter is a function, you can run custom logic against each property (default is ==)

nodes
  .filter(by('nodeName', isCustomElement)) // checks if has '-' in tag

It is common to sometimes see filtering empty values via .filter(Boolean). If only one parameter is given, it filters out objects that do not have a value for the property.

nodes
  .filter(by('prop')) // removes object that do not have a value for prop

chainable

Takes a function, and returns a new function that evaluates the original function and also returns it again ignoring what the evaluated function returns

ripple('key', value)                // this normally registers a resource, and returns the value
ripple.resource = chainable(ripple) // ripple.resource now registers a resource, but returns ripple again

ripple 
  .resource('foo', 'bar')
  .resource('lorem', 'ipsum')

NB: I think this will be deprecated in favour of the more generic proxy function that is used to alter return values

client

Simple variable: Am I on the server or browser?

// browser
client == true

// server
client == false

Useful for isomorphic apps/libs, and also deadcode elimination.

clone

Returns a new deep copy of an object

copied = clone(original)

colorfill

Adds color to strings, standardising behaviour across server/client

require('colorfill')

'foo'.red  // server, returns string in red
'foo'.red  // client, returns string

copy

Copies properties from one object to another

keys(from)
  .filter(not(is.in(private)))
  .map(copy(from, to))

datum

Returns the D3 datum for a node. Useful in lists as you cannot d3.select(node).datum()

nodes
  .map(datum)

deb

Lightweight scoped version of console.log with a prefix, useful for per module identification

deb = deb('[module/prefix]')
deb('something went wrong!') // [module/prefix] something went wrong

Returns the input, so it is useful with intermediary logging whilst iterating over a list

list
  .map(op1)
  .map(deb)
  .map(op2)

You can filter debug logs on server-side with DEBUG (module, or module/submodule) or using the debug querystring parameter on the client-side.

debounce

Returns a debounced function. Specify time in ms, or defaults to 100ms.

debounced = debounce(fn)
// or
debounced = debounce(200)(fn)

def

Defines a property, if does not already exist, returning the value

def(object, prop, value[, writable])  // returns value

defaults

Sets default values for properties on an object if not already defined. Normally used at the beginning of components to default state values and expose API:

var state = defaults(this.state, {
  values: []
, focused: true
})  

// state.focused == this.state.focused == true
// state.values  == this.state.values  == []

To idempotently define API on an element:

defaults(this, { toggle, spin })  

// this.spin()
// this.toggle()

In case you need to default and reference a property for another property, you can default them individually rather than via an object literal.

defaults(state, 'numbers', [1,2,3,4,5])
defaults(state, 'odd', state.numbers.filter(d => d % 2))

// state.numbers == [1,2,3,4,5]
// state.odd == [1,3,5]

delay

Creates a promise that eventually delays after the specified milliseconds. You can also set to resolve to a specific value.

await delay(1000)
await delay(1000, 'some value')

done

Given a versioned object, attaches a one-time callback for a response on the latest change. This is an alternative and more robust pattern than using random correlation ID's to link request-responses over a decoupled channel.

done(push(newUser)(users))
  (d => !d.invalid && showConfirmation())

el

Creates a node from a CSS selector

el(div.foo.bar[lorem=ipsum]) // returns <el class="foo bar" lorem="ipsum" />

emitterify

Enhance any object with .on, .once and .emit

var o = emitterify({}) 
o.on('event')             // get listeners for event
o.on('event', fn)         // set listener on arbitrary event
o.once('event', fn)       // set listener that is only called once
o.on('event.ns', fn)      // set listener for event.namespace, unique listener per namespace
o.emit('event', payload)  // emit event with optional payload
o.emit('event', [array])  // emit event with optional multiple arguments

err

Lightweight scoped version of console.error with a prefix, useful for per module identification

err = err('[module/prefix]')
err('something went wrong!') // [module/prefix] something went wrong

escape

Escapes HTML

escape = escape('<div></div>') // '&lt;div&gt;&lt;/div&gt;'

extend

Extends an object with properties from another, not overwriting properties by default. See also extend.

to = { foo: 1 }
from = { foo: 2, bar: 3 }
extend(to)(from)  // to == { foo: 1, bar: 3 }

falsy

Function that returns false

shouldIContinue = falsy // when executed, returns false

file

Reads and returns a file. Server only.

var template = file('template.html')

filify

Browserify transform that resolves file('filename') to the actual file

file('foo') // converted to a string containing the contents of the file foo

filter

Filters an array

once(tr, filter(isEven))

first

Returns first element in array

first(array)  // returns array[0]

flatten

Flattens a 2D array

twoD = [[1], [2], [3]]
oneD = twoD.reduce(flatten)  // [1, 2, 3]

form

Converts a <form> to a JSON object as you would expect. This is useful to abstract the different ways of getting values from different controls (text inputs, radio elements, checkboxes, selects, files, custom elements). The keys come from the name attributes. Checkboxes are represented as arrays. Files as FileList. For Custom Elements, it takes the element.state.value property, so they can also participate in forms.

const { values, invalid } = form(element)

values == {
  foo: '..'         // text inputs, radio elements, selects as single value
, bar: ['..', '..'] // checkboxes as arrays
, baz: FileList     // file inputs   
}

invalid is an array containing all the elements that were marked by the is-invalid class.

fn

Turns a function as a string into a real function

foo = 'function(){ console.log("Hi!") }'
foo = fn(foo)   // returns function(){ console.log("foo") }
foo()           // logs out "Hi!"

from

Looks up and returns a property from an object. Useful for converting foreign keys to matching records.

users = [
  { name: 'foo', city: 1 }
, { name: 'bar', city: 2 }
, { name: 'baz', city: 1 }
]

cities: { 1: 'London', 2: 'New York', 3: 'Paris' }

// get a unique list of cities users live in
users
  .map(key('city'))
  .map(from(cities))
  .filter(unique)     // returns [ 'London', 'New York' ]

from.parent returns the value of a property from the parent datum. Useful if you generate a fixed number of columns, whose values depend on the parent.

processes = [ 
  { name: 'chrome', pid: '123', cpu: '50%' } 
, { name: 'safari', pid: '456', cpu: '50%' } 
]

// generate a list of rows, each with three columns
once('tr', processes)
  ('td', ['name', 'pid', 'cpu'])
    .text(from.parent)

<tr>
  <td>chrome</td><td>123</td></td>50%</td>
</tr>
<tr>
  <td>safari</td><td>456</td></td>50%</td>
</tr>

In general you should try to pass each element the data it needs and not reach outside of its own scope.

group

Grouped logging using groupCollapsed/groupEnd if it exists, or simple start/end demarcation logs using asterisk if not.

group('category', fn)

grep

Conditionally executes a function depending on the regex against its arguments. Returns the original unfiltered function. Useful for higher order modules to conditionally filter out logs of many smaller modules unobtrusively.

// filter out all internal logs from ripple (i.e. that don't start with "[ri/")
unfiltered = grep(console, 'log', /^(?!.*\[ri\/)/)

gt

Filters array depending if value of a key is greater than a threshold.

array.filter(gt(100, 'value'))

Since it uses key internally, you can also filter on a deep key.

has

Checks if object has property using in keyword

has(object, 'prop')

hashcode

Converts string to unique numerical equivalent - same as Java hashcode function.

hashcode('foobar') // -1268878963

header

Extract the value of a header from a ripple resource

header('content-type')(resource) // returns for example 'application/javascript'

Or if a second parameter is set, check for equality

// filter to get all data resources
resources 
  .filter(header('content-type', 'application/data'))

identity

Identity function, returns what it is passed in

identity(5) // returns 5

iff

Only invoke second function if condition fulfilled. Useful for finer-grained control over skipping certain operations

sel(el).property('value', iff(cond)(setValue))

includes

Checks if string or array contains the pattern or element (uses indexOf common to strings and arrays)

// filter down to all javascript files
files
  .filter(includes('.js'))

is

Various basic flavours of checking

is(v)(d)      // equality d == v
is.fn         // function
is.str        // string
is.num        // number
is.obj        // object (includes arrays)
is.lit        // object (excludes arrays)
is.bol        // boolean
is.truthy     // truthy
is.falsy      // falsy
is.arr        // array
is.null       // null
is.def        // undefined
is.promise    // promise
is.in(set)(d) // checks if d in set (string, array or object)

join

Replace a foreign key property with the full record or a value from the record

// doctors == [ { name: nick, grade: 1 .. }, .. ]
// ripple('grades') == [ { id: 1, name: 'FY1' }, { id: 2, name: 'SHO' }, .. ]

doctors
  .map(join('shift', 'shifts'))
  .map(join('speciality', 'specialities'))
  .map(join('grade', 'grades.name'))
  .map(join('hospital', 'hospitals.location'))

// you can use with array map to replace ids with corresponding records
[1, 2, 3]
  .map(join('grades')) // returns [ { .. }, { .. }, { .. } ]

// you can use without ripple, if the table (last parameter) is a string 
// it'll resolve from ripple, but if it's a vanilla array, it'll just use that
[1, 2, 3]
  .map(join(grades)) // returns [ { .. }, { .. }, { .. } ]

If the second parameter is a string, it uses that as the ripple resource to look in. You can also use a primitive array outside of a ripple context.

key

Powerful versatile operator for accessing/setting key(s)

key('name')(d)                        // returns value of property name from object d
key(d => d.first + d.last)(d)         // returns computed value
key('details.profile.name')(d)        // returns deep property
key('details', 'foo')(d)              // set property
key('details.profile.name', 'foo')(d) // set deep property
key(['name', 'city.name'])(d)         // returns object with selected keys (can mix from any level)
key()(d)                              // returns object root if key undefined

Accessing deep keys returns undefined if a link is missing, which prevents doing things like:

(((d || {}).details || {}).profile || {}).name

Setting a deep key will create any missing keys it needs as it traverses the path.

If the second value parameter is a function, it evaluates it with the data before setting.

To make dates in all records human-readable with moment for example:

orders = [ { .. }, { .. } ]
orders
  .map(key('date', mo.format('MMM Do YY')))

keys

Alias for Object.keys

keys({ foo: 1, bar: 2}) // returns ['foo', 'bar']

last

Returns the last element in the array

last(array) // returns array[array.length-1]

link

Links changes in the attribute of one component to the attribute of another

// when a different day is selected in the calendar, the detail page automatically updates
link('events-calendar[selected-day]', 'event-detail[day]')

<events-calendar selected-day="1-1-1970" />
<event-detail day="1-1-1970"/>

lo

Lowercase a string

['A', 'B', 'C'].map(lo) // returns ['a', 'b', 'c']

log

Lightweight scoped version of console.log with a prefix, useful for per module identification

log = log('[module/prefix]')
log('something went wrong!') // [module/prefix] something went wrong

Returns the input, so it is useful with intermediary logging whilst iterating over a list

list
  .map(op1)
  .map(log)
  .map(op2)

lt

Filters array depending if value of a key is less than a threshold.

array.filter(lt(100, 'value'))

Since it uses key internally, you can also filter on a deep key.

mo

Convenience functions working with moment

dates.map(mo)                     // convert to moment object
dates.map(mo.format('Do MMM YY')) // convert to specific format
dates.map(mo.iso)                 // convert to iso date format

noop

Function that does nothing

;(fn || noop)()

not

Negates the result of a function

numbers
  .filter(not(isEven))
  .filter(not(is('5')))

nullify

Converts a truthy/falsy to true/null. This is a useful utility for D3 functions which expect a null value to remove as opposed to just a falsy.

selection
  .attr('disabled', nullify(isDisabled))

once

Function for building entirely data-driven idempotent components/UI with D3.

once(node)                        // limit to this node
  ('div', { results: [1, 2, 3] }) // creates one div (with the specified datum)
    ('li', key('results'))        // creates three li (with datum 1, 2, 3 respectively)
      ('a', inherit)              // creates anchor in each li (with parent datum)
        .text(String)             // sets the text in anchor to the datum

The first time you call once(node | string) it essentially selects that element and limits the scope of subsequent operations to that.

Subsequents calls generate a D3 join using the syntax (selector, data). The selector can be:

• A selector string (foo.bar.baz). Classes are fine too and will be added to the final elements created.
• A real element, which will be replicated.
• A function, which will be given parent data, in case you wish to output different (custom) elements based on data.

The data is the same as what you would normally use to generate a join (array of items, or function), with some convenient defaults: if you pass an object, number or boolean it'll be converted to a single-valued array, meaning "create one element with this as the datum". If you pass in a falsy, it defaults to empty array "meaning removing all elements of this type".

The return value is essentially a D3 join selection (enter/update/exit), so you can continue to customise using .text, .classed, .attr, etc. You can also access the elements added via .enter and removed via .exit.

There are two further optional arguments you can use (selector, data[, key[, before]]). The key function has the exact same meaning as normal (how to key data), which D3 defaults to by index. The before parameter can be used to force the insertion before a specific element à la .insert(something, before) as opposed to just .append(something).

Once will also emitterify elements as well as the selection so you can fluently listen/proxy events. You can create custom events, use namespaces for unique listeners, and listeners on events like "click" will trigger from both real user interaction from the DOM as well as via .emit.

once('list-component', 1)
  .on('selected', d => alert(d))

once('list-component')
  ('.items', [1,2,3])
    .on('click', d => this.parentNode.emit('selected', d))

overwrite

Extends an object with properties from another, overwriting existing properties. See also extend.

to = { foo: 1 }
from = { foo: 2, bar: 3 }
overwrite(to)(from)  // to == { foo: 2, bar: 3 }

owner

Either window or global dependeing on executing context

// browser
owner == window

// server
owner == global

parse

Equivalent to JSON.parse

patch

Updates multiple values at a key, updates the internal log (if versioned), and emits a standardised change event (if emitterified). See also other functional versioned operators.

patch('key', { a, b, c })(object)

pause

Actually pauses a stream so you can build up a pipeline, pass it around, attach more pipes, before starting the flow. Server only.

var stream = pause(browserify)
  .pipe(via(minify))
  .pipe(via(deadcode))

addMorePipes(stream)

function addMorePipes(stream){
  stream
    .on('end', doSomething)
    .pipe(file)
    .flow()
}

perf

Completely unobtrusive way to evaluate how long a function takes in milliseconds.

If you have the following function call:

fn(args)

You wrap a perf around fn to see how long it takes:

perf(fn)(args)  // Evaluates the function, logs the time taken in ms, and returns same value original fn would

You can also add an optional message to the log:

perf(fn, 'foo')(args)  

pop

Pops an element from an array, updates the internal log (if versioned), and emits a standardised change event (if emitterified). See also other functional versioned operators.

pop(users)

prepend

Prepend something to a string

['foo', 'bar']
  .map(prepend('hi-'))  // returns ['hi-foo', 'hi-bar']

promise

Convenience functions for working with (native) Promises

var p = promise()                 // creates promise with resolve/reject attached
p.resolve('result')
p.reject('something went wrong')

promise(5)                        // creates promise and resolves to value
promise.args(1)('foo', 'bar')     // creates promise that resolves to argument given
promise.sync(1)('foo', 'bar')     // creates thenable that immediately invokes then callback
promise.noop()                    // creates empty promise
promise.null()                    // creates promise that resolves to null

proxy

Proxy a function. It is common to use fn.apply(this, arguments) for proxying. This function allows you to do that, but alter the return value and/or context.

proxy(fn, 5)      // returns a function that invokes fn, but then always returns 5
proxy(fn, 5, {})  // same as above, but also changes context variable

This is also useful for functional inheritance:

bill.total = proxy(bill.subtotal, bill.vat)

push

Pushes an element to an array, updates the internal log (if versioned), and emits a standardised change event (if emitterified). See also other functional versioned operators.

push({ firstname: 'foo', lastname: 'bar' })(users)

ready

Calls the function once document.body is ready or immediately if it already is

ready(fn)

raw

Select an element based on a CSS selector, piercing shadow boundaries if the browser supports it.

raw('.foo')

Narrow search space by also passing in a node

raw('li', ul)

rebind

D3 rebind function to rebind accessors. See the docs here.

remove

Removes a key from an object, updates the internal log (if versioned), and emits a standardised change event (if emitterified). See also other functional versioned operators.

remove('key')(object)

replace

Replace a value in a string

['Hi name', 'Bye name']
  .map(replace('name', 'foo'))

resourcify

Returns the specified resource from a ripple instance. Returns an object of resources if multiple specified. Returns undefined if one of the resources not present.

resourcify(ripple)('foo')         // returns body of foo
resourcify(ripple)('foo bar')     // returns { foo: foo-body, bar: bar-body }
resourcify(ripple)('foo bar baz') // returns undefined, since no baz resource

sall

Convenience function for d3.selectAll. If either argument is already a D3 selection, it will not double wrap it.

sall(parent)(selector)

Parent/selector can be selection/string/node. If no parent, selects globally.

sel

Convenience function for d3.select. If the argument is already a D3 selection, it will not double wrap it.

sel(string)

send

Sends a file on an express route. Server only.

app.get('/file', send('./file'))

set

Takes an atomic diff and applies it to an object, updating the internal log, and emitting a standardised change event (if emitterified). An atomic diff is an object in the format { type, key, value, time } where type can be either of add | update | remove.

set({ key, value, type })(object)

If there is no diff, it initialises object with the .log property:

set()(object[[, existing], max])

If an existing object is specified with a .log property, it will branch off that history to create the new .log property. If a max property is specified, there are three options:

• 0: Diffs will be pushed onto the .log

• = 0: The .log property will always be []
• < 0: Nulls will be pushed onto the .log (this is to avoid potentially expensive operations in the critical path, whilst still being able to use .log.length as a revision counter)

Note that this, as will all utilities here, is fully graceful and will work with:

• A vanilla object - just applies diff
• An emitterified object i.e. has a .on/.emit - applies the diff and emits log event
• An emitterified, versioned object (has a .log) - applies diff, emits event, and updates immutable/diff log

See also the more ergonomic functional versioned operators which use this generic operator, but set most of the values for you:

• pop
• push
• patch
• update
• remove

slice

Slice all strings in an array

['Hi name', 'Bye name']
  .map(slice(0, 2)) // ['Hi', 'By']

sort

Sorts elements in an array

once('tr', sort(Boolean))

split

Split items

['a.b', 'b.c'].map(split('.')) // returns [['a','b'], ['b','c']]

str

Coerces anything into a string

str(5)                // returns '5'
str({ foo: 5 })       // returns '{ foo: 5 }'
str(undefined)        // returns ''
str(function(){ .. }) // returns 'function(){ .. }'

stripws

Strips the whitespace between tags (but not between attributes), which is useful to make declarative testing less brittle and less sensitive

stripws`
  <div>
    foo
  </div>
`         
// <div>foo</div>

stripws('<a>\n   <b>')  
// '<a><b>'

tdraw

Simple stub for the draw function to test components. Takes a DOM element, a render function and some initial state. Sets the initial .state, the .draw function and renders the element.

t.plan(2)

const host = tdraw(el('ux-button'), button, { label: 'foo', spinning: true })

host.spin(true)
t.ok(includes(`class="is-spinning"`)(host.outerHTML))

host.spin(false)
t.notOk(includes(`class="is-spinning"`)(host.outerHTML))

th

Invokes a function with the this variable, then the arguments. Useful for using arrow syntax with functions that need the context variable.

// instead of:
// input.on('keyup', log)
// function log(){
//  console.log(this.value)
// }

const log = el => d => console.log(el.value)
input.on('keyup', th(log))

time

Alias for setTimeout with duration as first parameter for better readability

time(10, function(){ .. })
time(20, function(){ .. })
time(30, function(){ .. })

to

to.arr: Converts to a primitive type (only real arrays)

to.arr(NodeList)
to.arr(arguments)

to.obj: Converts an array to an object. Uses id property as key by default if none specified. The key can also be a function which will receive the item and it's index as parameters, and return the computed string to use as the key.

[
  { id: 'foo', value: 1 }
, { id: 'bar', value: 2 }
].reduce(to.obj, {})

/* returns 
{ 
  foo: { id: 'foo', value: 1 }
, bar: { id: 'bar', value: 2 }
} 
*/

// or you can use .reduce(to.obj('prop'), {})

Note: You should always use an initial value with the reduce function. This is because if your array happens to be an array with only one element and there is no initial value, JavaScript will not even call the reduce function.

unique

Filter an array to unique values

[1,1,2,3].filter(unique, 1) // returns [1,2,3]

update

Updates a value at a key, updates the internal log (if versioned), and emits a standardised change event (if emitterified). See also other functional versioned operators.

update('key', value)(object)

values

Converts an object to array

values({ 
  a: { name: 'foo', value: 1 }
, b: { name: 'bar', value: 2 }
})

/* returns
[ 
  { name: 'foo', value: 1 }
, { name: 'bar', value: 2 }
]
*/

via

Buffers output to a stream destination. Useful when you need the whole input rather than chunks. Server only.

stream
  .pipe(via(minify))
  .pipe(via(replace))

wait

Only invoke handler if condition fulfilled. Useful for determining execution based on declarative pattern matching.

o.once(wait(msg => msg.id)(handler))

wrap

Wraps something in a function which returns it when executed

wrapped = wrap(5)
wrapped()          // returns 5