A collection of F(unctional) Util(ities). Resistance is futile.
Mostly, these are generic utilities that could conceivably be part of a library like lodash/fp, but for some reason or other are not.
https://smartprocure.github.io/futil-js/
See our changelog
npm i -S futil-js
This package requires lodash/fp
, so make sure that's available in your app.
import * as f from futil-js
or
import {x,y,z} from futil-js
The syntax: import f from futil-js
is not currently supported.
(fn, a, b) -> fn(a, b)
If fn
is a function, call the function with the passed-in arguments. Otherwise, return false
.
(fn, a, b) -> fn(a, b)
If fn
is a function, call the function with the passed-in arguments. Otherwise, return fn
.
(a, Monoid f) -> f[a] :: f a
Binds a function of an object to it's object.
(f, [g1, g2, ...gn]) -> a -> f([g1(a), g2(a), ...])
http://ramdajs.com/docs/#converge. Note that f
is called on the array of the return values of [g1, g2, ...gn]
rather than applied to it.
(f, g) -> x -> f(g(x))(x)
A combinator that combines compose and apply. f
should be a 2 place curried function. Useful for applying comparisons to pairs defined by some one place function, e.g. var isShorterThanFather = F.comply(isTallerThan, fatherOf)
Implement defer
, ported from bluebird docs and used by debounceAsync
A _.debounce
for async functions that ensure the returned promise is resolved with the result of the execution of the actual call. Using _.debounce
with await
or .then
would result in the earlier calls never returning because they're not executed - the unit tests demonstate it failing with _.debounce
.
(f1, f2, ...fn) -> f1Arg1 -> f1Arg2 -> ...f1ArgN -> fn(f2(f1))
Flurry is combo of flow + curry, preserving the arity of the initial function. See lodash/lodash#3612.
([f1, f2, ...fn]) -> !f1(x) && !f2(x) && ...!fn(x)
Creates a function that checks if none of the array of predicates passed in returns truthy for x
(condition, onTrue, onFalse) -> x -> (T(condition)(x) ? onTrue(x) : onFalse(x))
http://ramdajs.com/docs/#ifElse. The transform function T supports passing a boolean for condition
as well as any valid argument of _.iteratee
, e.g. myBool = applyTest(x); F.ifElse(myBool, doSomething, doSomethingElse);
(condition, onTrue) -> x -> (T(condition)(x) ? onTrue(x) : _.identity(x))
http://ramdajs.com/docs/#when. T
extends _.iteratee
as above.
(condition, onFalse) -> x -> (T(condition)(x) ? _.identity(x) : onFalse(x))
http://ramdajs.com/docs/#unless. T
extends _.iteratee
as above.
when
curried with Boolean
when
curried with exists
[f1, f2, ...fn] -> _.map(_.flow(fn))
Maps a flow of f1, f2, ...fn
over a collection.
f -> x -> f(find(f, x))
A version of find
that also applies the predicate function to the result. Useful when you have an existing function that you want to apply to a member of a collection that you can best find by applying the same function.
(a -> b) -> [a] -> [b]
Maps a function over an iterable. Works by default for Arrays and Plain Objects.
(a -> b) -> [a] -> [b]
Maps a function over a recursive iterable. Works by default for nested Arrays, nested Plain Objects and mixed nested Arrays and Plain Objects. Also works for any other iterable data type as long as two other values are sent: a mapping function, and a type checker (See the unit tests for deepMap).
These are conversions of lodash fp methods.
getIn
, hasIn
, includesIn
, pickIn
lodash/fp is great, but sometimes the curry order isn't exactly what you want.
These methods provide alternative orderings that are sometimes more convenient.
The idea of In
methods is to name them by convention, so when ever you need a method that actually takes the collection first (e.g. a get
where the data is static but the field is dynamic), you can just add In
to the end (such as getIn
which takes the object first)
extendOn
, defaultsOn
, mergeOn
, setOn
, unsetOn
, pullOn
lodash/fp likes to keep things pure, but sometimes JS can get pretty dirty.
These methods are alternatives for working with data that--for whatever the use case is--needs to be mutable
Any methods that interact with mutable data will use the On
convention (as it is some action occuring On
some data)
mapIndexed
, eachIndexed
, reduceIndexed
, mapValuesIndexed
lodash/fp caps iteratees to one argument by default, but sometimes you need the index.
These methods are uncapped versions of lodash's methods.
Any method with uncapped iteratee arguments will use the Indexed
convention.
joinString -> [string1, string2, ...stringN] -> string1 + joinString + string2 + joinString ... + stringN
Joins an array after compacting. Note that due to the underlying behavior of _.curry
no default join
value is supported -- you must pass in some string with which to perform the join.
[string1, string2, ...stringN] -> string1 + '.' + string2 + '.' ... + stringN
Compacts and joins an array with '.'
filterFunction -> [string1, string2, ...stringN] -> string1 + '.' + string2 + '.' ... + stringN
Compacts an array by the provided function, then joins it with '.'
[a] -> [a]
Returns an array of elements that are repeated in the array.
([[], [], []]) -> [[], []]
Takes any number of ranges and return the result of merging them all.
Example: [[0,7], [3,9], [11,15]] -> [[0,9], [11,15]]
insertAtIndex -> (index, val, string) -> string
Insert a string at a specific index.
Example: (1, '123', 'hi') -> 'h123i'
(val, array) -> array
Return array
with val
pushed.
[a, b...] -> a -> b
Creates a function that takes an element of the original array as argument and returns the next element in the array (with wrapping). Note that (1) This will return the first element of the array for any argument not in the array and (2) due to the behavior of _.curry
the created function will return a function equivalent to itself if called with no argument.
(k, v, [a]) -> { k(a): v(a) }
Creates an object from an array by generating a key/value pair for each element in the array using the key and value mapper functions.
A version of _.zipObjectDeep
that supports passing a function to determine values intead of an array, which will be invoked for each key.
[a, b] -> {a:true, b:true}
Converts an array of strings into an object mapping to true. Useful for optimizing includes
.
['a', 'b', 'c'] -> [['a'], ['a', 'b'], ['a', 'b', 'c']]
Returns a list of all prefixes. Works on strings, too. Implementations must guarantee that the orginal argument has a length property.
string -> {encode: array -> string, decode: string -> array}
Creates an object with encode and decode functions for encoding arrays as strings. The input string is used as input for join/split.
{ encode: ['a', 'b'] -> 'a.b', decode: 'a.b' -> ['a', 'b'] }
An encoder using .
as the separator
{ encode: ['a', 'b'] -> 'a/b', decode: 'a/b' -> ['a', 'b'] }
An encoder using /
as the separator
(k, v) -> {k: v}
Creates an object with a key and value.
(v, k) -> {k: v}
Flipped version of singleObject
.
({a, b}) -> [{a}, {b}]
Breaks an object into an array of objects with one key each.
Remove properties with falsey values.
Example: ({ a: 1, b: null, c: false }) -> {a:1}
Check if the variable is an empty object ({}
).
Check if the variable is not an empty object ({}
).
Omit properties whose values are empty objects.
Example: { a:1, b:{}, c:2 } -> {a:1, c:2}
(TODO remame to omitEmptyObjects
)
Checks if an object's property is equal to a value.
Returns true if object keys are only elements from signature list. (but does not require all signature keys to be present)
Similar to _.matches
, except it returns true if 1 or more object properties match instead of all of them. See lodash/lodash#3713.
TODO
sourcePropertyName -> targetPropertyName -> sourceObject -> sourceObject
Rename a property on an object.
Example: renameProperty('a', 'b', { a: 1 }) -> { b: 1 }
'b' -> { a: true, b: [1, 2] } -> { a: true, b: 1 }, { a: true, b: 2}
Just like mongo's $unwind
: produces an array of objects from an object and one of its array-valued properties. Each object is constructed from the original object with the array value replaced by its elements. Unwinding on a nonexistent property returns an empty array.
Flatten an object with the paths for keys.
Example: { a: { b: { c: 1 } } } => { 'a.b.c' : 1 }
.
Unlatten an object with the paths for keys.
Example: { 'a.b.c' : 1 } => { a: { b: { c: 1 } } }
.
Deprecated in favor of lodash update
Applies a map function at a specific path
Example: mapProp(double, 'a', {a: 2, b: 1}) -> {a: 4, b: 1}
.
_.get
that returns the target object if lookup fails
_.get
that returns the prop if lookup fails
Flipped alias
A _.get
that takes an array of paths (or functions to return values) and returns the value at the first path that matches. Similar to _.overSome
, but returns the first result that matches instead of just truthy (and supports a default value)
Flipped cascade
A _.get
that takes an array of paths and returns the first path that matched
A _.get
that takes an array of paths and returns the first value that has an existing path
A _.get
that takes an array of paths and returns the first path that exists
newKey -> {a:x, b:y} -> [{...x, newKey: a}, {...y, newKey: b}]
Opposite of _.keyBy
. Creates an array from an object where the key is merged into the values keyed by newKey
. Example: F.unkeyBy('_key')({ a: { status: true}, b: { status: false }) -> [{ status: true, _key: 'a' }, { status: false, _key: 'b' }]
. Passing a falsy value other than undefined
for newKay
will result in each object key being pushed into its corresponding return array member with itself as value, e.g. F.unkeyBy('')({ a: { status: true}, b: { status: false }) -> [{ status: true, a: 'a' }, { status: false, b: 'b' }]
. Passing undefined
will return another instance of F.unkeyBy.
(from, to) -> simpleDiff
Produces a simple flattened (see flattenObject
) diff between two objects. For each (flattened) key, it produced a from
and a to
value. Note that this will omit any values that aren't present in the deltas object.
(from, to) -> [simpleDiffChanges]
Same as simpleDiff
, but produces an array of { field, from, to }
objects instead of { field: { from, to } }
(from, to) -> diff
Same as simpleDiff
, but also takes in count deleted properties.
Note: We're considering not maintaining this in the long term, so you might probably have more success with any existing library for this purpose.
(from, to) -> [diffChanges]
Same as simpleDiffArray
, but also takes in count deleted properties.
Note: We're considering not maintaining this in the long term, so you might probably have more success with any existing library for this purpose.
A _.pick
that mutates the object
Like _.mergeAll
, but concats arrays instead of replacing. This is basically the example from the lodash mergeAllWith
docs.
{ a: [x, y, z], b: [x] } -> { x: [a, b], y: [a], z: [a] }
Similar to _.invert
, but expands arrays instead of converting them to strings before making them keys.
'asdf' -> '(asdf)'
Wraps a string in parenthesis.
Maps _.trim
through all the strings of a given object or array.
string -> string
Converts strings like variable names to labels (generally) suitable for GUIs, including support for acronyms and numbers. It's basically _.startCase
with acronym and number support.
string -> {value:string, label:string}
Creates a {value, label}
which applies autoLabel
the string parameter on puts it on the label property, with the original on the value property. You can also pass in an object with value or with both value and label.
[string] -> [{value:string, label:string}]
Applies autoLabelOption
to a collection. Useful for working with option lists like generating select tag options from an array of strings.
regex -> string -> bool
Just like ramda test, creates a function to test a regex on a string.
options:string -> string -> regex
A curried implementation of RegExp
construction.
options:string -> string -> (string -> bool)
Makes and tests a RegExp with makeRegex and testRegex.
string -> string -> bool
Returns true if the second string matches any of the words in the first string.
string -> string -> bool
Returns true if the second string matches all of the words in the first string.
regex -> string -> [[number, number]]
Returns an array of postings (position ranges) for a regex and string to test, e.g. F.postings(/a/g, 'vuhfaof') -> [[4, 5]]
start -> end -> regex -> input -> highlightedInput
Wraps the matches for regex
found in input
with the strings start
and end
.
Example: ('<b>', '</b>', /h/, 'hi') -> '<b>h</b>i'
regex -> string -> [{text: string, start: number, end: number}]
Returns an array of matches with start/end data, e.g. F.allMatches(/a/g, 'vuhfaof') -> [ { text: 'a', start: 4, end: 5 } ]
number -> bool
Returns true if number is greater than one.
Language level utilities
Just throws whatever it is passed.
Tap error will run the provided function and then throw the first argument. It's like _.tap
for rethrowing errors.
Negated _.isNil
Returns true if the input has a length
property > 1, such as arrays, strings, or custom objects with a lenth property
A curried, flipped _.add
. The flipping matters for strings, e.g. F.append('a')('b') -> 'ba'
x -> bool
Designed to determine if something has a meaningful value, like a ux version of truthiness. It's true for everything except null, undefined, '', [], and {}. Another way of describing it is that it's the same as falsiness except 0 and false are truthy and {} is falsey. Useful for implementing "required" validation rules.
x -> bool
Opposite of isBlank
f -> x -> bool
Recurses through an object's leaf properties and passes an array of booleans to the combinator, such as _.some
, _.every
, and F.none
A lens is a getter and setter pair, which can be used to interface to some part of an object graph. Methods that operate on lenses can encapsulate common operations independent of knowledge of their surrounding context. Unlike some traditional functional lenses (like Ramda's), the set methods here are generally mutable.
An object lens is simply an object that has a get
and set
function.
An example of this is a mobx boxed observable.
A function lens is a lens expressed as a single function that takes the value to set or returns the current value if nothing is passed. Examples of this in the wild are knockout observables and jquery plugin api style methods.
The utilities in this library expect can accept either kind of lens, and utilities are provided to seamlessly convert between the two.
Lens stubs are primarily a reference implementation, but are useful for testing and mocking purposes
Takes a value and returns a function lens for that value
Takes a value and returns a object lens for that value
Methods to convert between lens types
Converts a function lens an object lens
Converts an object lens to a function lens
This the main way you'll generally interact with the lens API
propertyName -> object -> { get: () -> object.propertyName, set: propertyValue -> object.propertyName }
Creates an object lens for a given property on an object. .get
returns the value at that path and set
places a new value at that path. Supports deep paths like lodash get/set.
Takes an object and returns an object with lenses at the values of each path. Basically mapValues(lensProp)
.
Note: As of version 1.37, any manipulation function that takes a lens can also drop in a key and target object for an implicit lensProp conversion (e.g. you can do view(key, obj)
instead of just view(lens)
)
Lens -> object.propertyName
Gets the value of the lens, regardless of if it's a function or object lens
Lens -> (() -> object.propertyName)
Returns a function that gets the value of the lens, regardless of if it's a function or object lens
propertyValue -> Lens -> object.propertyName
Sets the value of the lens, regardless of if it's a function or object lens
Creates a function that will set a lens with the provided value
Takes a lens and negates its value
Returns a function that will set a lens to true
Returns a function that will set a lens to false
Aspects provide a functional oriented implementation of Aspect Oriented Programming. An aspect wraps a function and allows you run code at various points like before and after execution. Notably, aspects in this library allow you to have a shared state object between aspects and are very useful for automating things like status indicators, etc on functions.
There is a lot of prior art in the javascript world, but most of it assumes a vaguely object oriented context.
The implementation in futil-js
is done in just 20 lines of code and seems to capture all of the use cases of AOP.
Note: To do OO style AOP with this these aspects, just use lodash's
_.update
method and optionallyboundMethod
fromfutil
ifthis
matters
Caveat: While you can and should compose (or
_.flow
) aspects together, don't put non aspects in the middle of the composition. Aspects rely on a.state
property on the wrapped function that they propagate through, but the chain will break if a non-aspect is mixed in between. Additionally, if you need external access to the state, make sure the aspects are the outer most part of the composition so the.state
property will be available on the result of the composition.
{options} -> f -> aspectWrapped(f)
The aspect api takes an options object and returns a function which takes a function to wrap.
The wrapped function will be decorated with a state
object and is equivalent to the original function for all arguments.
Options supports the following parameters:
Name | Description |
---|---|
init: (state) -> () |
A function for setting any inital state requirements. Should mutate the shared state object. |
after: (result, state, params) -> () |
Runs after the wrapped function executes and recieves the shared state and the result of the function. Can be async. |
before: (params, state) -> () |
Runs before the wrapped function executes and receves the shared state and the params passed to the wrapped function. Can be async. |
onError: (error, state, params) -> () |
Runs if the wrapped function throws an error. If you don't throw inside this, it will swallow any errors that happen. |
always: (state, params) -> () |
Runs after the wrapped function whether it throws an error or not, similar to a Promise.catch |
Example Usage:
let exampleAspect = aspect({
before: () => console.log('pre run'),
after: () => console.log('post run')
})
let f = () => console.log('run')
let wrapped = exampleAspect(f)
wrapped()
// Logs to the console:
// pre run
// run
// post run
This is a synchronous version of aspect
, for situations when it's not desirable to await
a method you're adding aspects to. The API is the same, but things like onError
won't work if you pass an async function to the aspect.
There are a few basic aspects included on F.aspects
(E.g. var loggedFunc = F.aspect(F.aspects.logs)(func)
) because they seem to be universally useful.
All of the provided aspects take an extend
function to allow customizing the state mutation method (e.g. in mobx, you'd use extendObservable
).
If null, they default to defaultsOn
from futil-js
- check the unit tests for example usage.
Logs adds a logs
array to the function state and just pushes in results on each run
Captures any exceptions thrown and set it on an error
error it puts on state
Captures any exceptions thrown and pushes them sequentially into an errors
array it puts on state
Adds a status
property that is set to processing
before the wrapped function runs and succeeded
when it's done or failed
if it threw an exception. Also adds shortcuts on state for processing
, succeeded
, and failed
, which are booleans which are based on the value of status
. Also adds a setStatus
method which is used internally to update these properties.
Sets status
to null after provided timeout (default is 500ms) elapses. If a null timeout is passed, it will never set status to null.
Prevents a function from running if it's state has processing
set to true at the time of invocation
Flows together status
, clearStatus
, concurrency
, and error
, taking extend
and timeout
as optional parameters to construct the aspect
Utility for marking functions as deprecated - it's just a before
with a console.warn. Takes the name of thing being deprecated, optionally deprecation version, and optionally an alternative and returns a higher order function which you can wrap deprecated methods in. This is what's used internally to mark deprecations. Includes a partial stack trace as part of the deprecation warning.
All tree functions take a traversal function so that you can customize how to traverse arbitrary nested structures.
Note: Be careful about cyclic structures that can result in infinite loops, such as objects with references to itself. There are cases where you'd intentionally want to visit the same node multiple times, such as traversing a directed acyclic graph (which would work just fine and eventually terminate, but would visit a node once for each parent it has connected to it) - but it's up to the user to be sure you don't create infinite loops.
A default check if something can be traversed - currently it is arrays and plain objects.
The default traversal function used in other tree methods if you don't supply one. It returns false if it's not traversable or empty, and returns the object if it is.
traverse -> (pre, post=_.noop) -> tree -> x
A depth first search which visits every node returned by traverse
recursively. Both pre-order
and post-order
traversals are supported (and can be mixed freely). walk
also supports exiting iteration early by returning a truthy value from either the pre
or post
functions. The returned value is also the return value of walk
. The pre, post, and traversal functions are passed the current node as well as the parent stack (where parents[0] is the direct parent).
traverse -> _iteratee -> tree -> newTree
Structure preserving pre-order depth first traversal which clones, mutates, and then returns a tree. Basically walk
with a _.cloneDeep
first (similar to a tree map because it preserves structure). _iteratee
can be any suitable argument to _.iteratee
https://lodash.com/docs/4.17.5#iteratee
traverse -> (accumulator, initialValue, tree) -> x
Just like _.reduce
, but traverses over the tree with the traversal function in pre-order
.
traverse -> tree -> [treeNode, treeNode, ...]
Flattens the tree nodes into an array, simply recording the node values in pre-order traversal.
traverse -> f -> tree -> [f(treeNode), f(treeNode), ...]
Like treeToArray
, but accepts a customizer to process the tree nodes before putting them in an array. It's _.map
for trees - but it's not called treeMap because it does not preserve the structure as you might expect map
to do.
traverse -> tree -> [treeNodes]
Returns an array of the tree nodes that can't be traversed into in pre-order
.
(traverse, buildIteratee) -> ([_iteratee], tree) -> treeNode
Looks up a node matching a path, which defaults to lodash iteratee
but can be customized with buildIteratee. The _iteratee
members of the array can be any suitable arguments for _.iteratee
https://lodash.com/docs/4.17.5#iteratee
traverse -> transformer -> _iteratee -> tree -> result
Similar to a keyBy (aka groupBy) for trees, but also transforms the grouped values (instead of filtering out tree nodes). The transformer takes three args, the current node, a boolean of if the node matches the current group, and what group is being evaluated for this iteratee. The transformer is called on each node for each grouping. _iteratee
is any suitable argument to _.iteratee
, as above.
traverse -> buildPath -> tree -> result
Creates a flat object with a property for each node, using buildPath
to determine the keys. buildPath
takes the same arguments as a tree walking iteratee. It will default to a dot tree path.
(build, encoder) -> treePathBuilderFunction
Creates a path builder for use in flattenTree
. By default, the builder will look use child indexes and a dotEncoder. Encoder can be an encoding function or a futil encoder
(an object with encode and decode functions)
prop -> treePathBuilderFunction
Creates a path builder for use in flattenTree
, using a slashEncoder and using the specified prop function as an iteratee on each node to determine the keys.
A utility tree iteratee that returns the stack of node indexes
A utility tree iteratee that returns the stack of node values
(traverse, buildIteratee) -> {walk, reduce, transform, toArray, toArrayBy, leaves, lookup, keyByWith, traverse, flatten, flatLeaves }
Takes a traversal function and returns an object with all of the tree methods pre-applied with the traversal. This is useful if you want to use a few of the tree methods with a custom traversal and can provides a slightly nicer api.
Exposes provided traverse
function as traverse