Skip to content
/ oaty Public

Object Array Transposer(y) - JS objects with multiple key/value structures

License

Notifications You must be signed in to change notification settings

jmsv/oaty

Repository files navigation

oaty

Object Array Transposer(y) - JS objects with multiple key/value structures

Why?

The idea (j-m's) was to find a way of being able to get values from an array of objects in O(1) time, in an environment where memory isn't an issue.

For example, I could have the following JS object:

const food = [
  {
    id: 1,
    name: "apple",
    type: "fruit",
  },
  {
    id: 2,
    name: "orange",
    type: "fruit",
  },
  {
    id: 3,
    name: "broccoli",
    type: "vegetable",
  },
];

As-is, to get the item where the id is 3, I'd use food.find(x => x.id === 3) or similar.

If the array was transposed to use id as object keys, the resultant object would look like the following:

const foodById = {
  1: {
    id: 1,
    name: "apple",
    type: "fruit",
  },
  2: {
    id: 2,
    name: "orange",
    type: "fruit",
  },
  3: {
    id: 3,
    name: "broccoli",
    type: "vegetable",
  },
};

This way, we can get the food with the id of 3 with foodById[3].

However, we don't know that object keys will all be unique, so objects should be placed in arrays. For example:

const foodByType = {
  fruit: [
    {
      id: 1,
      name: "apple",
      type: "fruit",
    },
    {
      id: 2,
      name: "orange",
      type: "fruit",
    },
  ],
  vegetable: [
    {
      id: 3,
      name: "broccoli",
      type: "vegetable",
    },
  ],
};

Now, to get an array of fruit, rather than using food.filter(f => f.type === 'fruit') we can just use foodByType['fruit'].


For data that changes frequently, this is a bad approach since transposing the data to use different key values is expensive.

However, for data that is assigned once (e.g. when a server first starts running) or assigned relatively infrequently (e.g. polling a database) this idea should be far less CPU-intensive than frequently searching the array using filter, find, or manually.

Getting Started

This library's default export is OatyArray. Initialise it as such:

const oatyFood = new OatyArray(food, { keys: ["id", "type"] });

In the above case, food is the initial array of items and ['id', 'type'] is the array of keys you want to be able to query.

The OatyArray constructor generates the transposed caches.

To query data, use the get method:

const fruitArray = oatyFood.get("type", "fruit");

The above is effectively the same as foodByType['fruit'] in the above examples

Development

oaty is built with Typescript

To get started, run npm install to install dependencies.

  • Run tests: npm test
  • Build: npm run build

Feel free to open Issues/PRs with suggestions/problems/improvements.

Library is versioned as-per the semver standard.

Maintainers

  • James Vickery - jmsv
  • Jonathan Marsh - j-m

Changelog

1.0.0

  • Added extra type inference and enforcement to Oaty, this means that an array can be initialised new OatyArray([{ myKey: "myValue" }]) without needing to provide a type.
  • keys input is now enforced to a subset of the keys of the given type
  • keys input will now be used to enforce that only valid objects are transposed by Oaty
  • .keys will now always return either an array containing all the auto transposed keys, or an array of the configured keys, and is now type correct.
  • Oaty will now enforce the correct keys are the objects of .push meaning that an error will be thrown at compile time if an invalid object is being passed to Oaty.
  • Updated overloads on .get to return the correct types depending on usage
  • Updated to typescript@3.8
  • Added tsd tests

0.4.0

  • Oaty can now be initialised with new OatyArray<T>() so that .get , .data, and .transposed return data of type T. This is completely optional as T defaults to any.
  • Removed missingKeyReturns and noResultsReturns options.
  • Exported a new type for the _transposed property.
  • Expanded the ternary conditions to improve readability.
  • Made the data constructor param optional and default to [].
  • When a key is not transposed and .get searches for that key, it will now throw a ReferenceError.
  • Added NYC and increased tests to 100% coverage.

0.3.0

  • Added gets for data, keys, and transposed
  • Made keys optional - will transpose every (root) key in an object
  • Added two options to change the return value for when they key or value of .get() is undefined
  • Renamed original to data
  • .get() can now retrieve all objects that have a key, like .get('fruit')
  • Added simple benchmark
  • Added more tests

0.2.1

  • Fixed .get() method type (object[])

0.2.0

  • Renamed OatyObject to OatyArray, since it's intended as an array alternative, rather than an object alternative
  • Named export, rather than default export

0.1.0

  • Added .push function

0.0.0

  • Initial proof of concept