this post was originally published on my Github Pages site on September 13th, 2017
This might sound a bit outlandish or ridiculous, but I seldom write loops nowadays. What I have found is that just about every programming language includes a set of methods or applicable functions that can replace just about every loop that I was previously writing. These higher-order functions are called map, filter, and fold.
map function takes a function you provide and iterates over each item in your list, applying the function you provide and putting the result in the same spot in the new array. Since
map takes an array and returns an array, you can chain together your
map calls and transform your data incrementally.
We'll start off with some easy examples:
That didn't hurt too much, did it? In the first
map, we essentially applied
f(x) = 2x to each element in the sequence. Likewise, the second
f(x) = xÂ² to each element. Easy cheesy.
Learning to crawl
Here's an example of mapping over an array of objects:
In the first
map we are pulling out each person's name, their age in the second
map, and their second hobby in the third
map. If what is happening here isn't immediately apparent to you, here is the same thing in an imperative style:
Crazy, right? What we can do with a single expression with
map, takes FOUR lines with imperative code. Oh, and did you notice the fact that we did not mutate the value of the original array in the imperative example? This is also true for
map, which is important since we might need to do other things to that original array.
Up and running!
Alright, "this is child's play", you say. Where are we REALLY gonna use
map? Well, buckle up! Check out this real world example:
This example is straight outta polysvg, albeit cut down a bit for brevity. This
mapchain takes an array of six zeroes and performs the following steps:
- Numbers each spot according to its index
- Multiplies each position by 60 to get the angles of each vertex from the centroid of a hexagon
- Converts each angle to radians
- Converts from polar to Cartesian coordinates
- Rounds these nasty floats
- Applies an offset to each point equal to the radius
Without annotations, this is about nineteen lines of code.. Most importantly, it works and you can try it out yourself.
When should I use
map has a 1:1 relationship between the number of things you put in and the number of things you get out, you should use
map when you want to transform
x amount of things into
x amount of other things. If you need to turn
x amount of things into
x - 5 amount of things,
map might not be the ideal solution right away. You may need to segregate your things into subsets, and then
map each subset separately.
Like, all the good ones. Though the names might be a bit different. In an effort to avoid plagiarism and only write what I really know about, I'll list out a few equivalent methods/functions here.
|C#||IEnumerable.Select<T> (as part of System.Linq)|
|MongoDB||$project (as part of an aggregation pipeline)|
Alright, I'm convinced. When do I start?
Right now! Go!
map all the things!
The best way to get familiar with
map is to just start using it.
Top comments (6)
Will you talk about Filter and Fold too?
Not sure if you caught it but filter and fold are published now!
Yes. I should have filter and fold out within the next few days.
Shouldn't it be person.hobbies?
Map is really cool though. This is really helping with my understanding.
It depends on what you are after. In the given example, I wanted to pull out the second item in the hobbies array from each person.
I'm glad I could help. When I first bumped into map it freaked me out quite a bit.
Scala and Java have map too. But it's overrated.