abstraction

abstraction

A way of modeling your business logic or data structures in a compact and compose-able way.

• Does one job
• Functions, Records, and Arrays, that's it.
• Works for many uses: ODM/ORM, Presenter, Serializer, Deserializer, Decorator, etc.

using

Lets say you have some data from a server that you want to display to the user:

const incomingData = {
  id: "1",
  type: "accounts",
  attributes: {
    name: "Kurtis Rainbolt-Greene",
    age: "27",
    email: "[e-mail]",
    "created-at": "2013-02-04T10:35:24-08:00",
  }
}

The age is a string, the timestamp for created at is also a string, and created at isn't camel case. We can fix this:

const accountResource = abstraction({
  attributes: {
    id: {source: prop("id")},
    type: {source: prop("type")},
    attributes: {
      source: prop("attributes"),
      coerce: accountAttributes,
    }
  }
}

There's a lot happening here so lets break it down:

const accountResource = abstraction({
  attributes: {
    ...
  }
}

Every abstraction has a set of attributes. These are properties that'll be exposed on the returning abstract object.

Lets take the id and type attributes for a moment:

const accountResource = abstraction({
  attributes: {
    id: {source: prop("id")},
    type: {source: prop("type")},
  }
}

For these we're just defining a source. The source is a function that tells us how to get to our data for id and type. This is easy for us, so we'll just use path(). Now it's time to get a little more complicated with attributes.

const accountResource = abstraction({
  attributes: {
    id: {source: prop("id")},
    type: {source: prop("type")},
    attributes: {
      source: prop("attributes"),
      coerce: accountAttributes,
    },
  },
}

So here we're not only sourcing from the raw data, but also using a fun little coerce function. This function should take the data from source and convert it to what you need. In this case we're doing something really cool in that we're actually coercing the data into another abstraction!

const accountAttributes = abstraction({
  attributes: {
    username: {source: pipe(prop("email"), split("@"), head)},
    email: {source: prop("email")},
    name: {source: prop("name")},
    age: {
      source: prop("age"),
      coerce: (value) => parseInt(value, 10),
    },
    createdAt: {
      source: prop("created-at"),
      coerce: moment,
    },
  },
}

There's some interesting things going on here as well, specifically in username's source and createdAt's coerce function. Notice that there is no username on the raw data and is instead really a derived value. In addition, your final property name doesn't have to be the same as your raw property.

Finally we also give a built in validations logic:

const accountResource = abstraction({
  attributes: {
    id: {source: prop("id")},
    type: {source: prop("type")},
    attributes: {
      source: prop("attributes"),
      coerce: accountAttributes,
    }
  },
  validations: {
    id: {
      "Must have an id": propSatisfies(isPresent, "id"),
      "Must be an string": propSatisfies(isType("String"), "id"),
      "Must have characters": propSatisfies(isPopulated, "id"),
    },
    type: {
      "Must have an type": propSatisfies(isPresent, "type"),
      "Must be an string": propSatisfies(isType("String"), "type"),
      "Must have characters": propSatisfies(isPopulated, "type"),
    },
    attributes: {
      "Must have attributes": propSatisfies(isPresent, "attributes"),
      "Must have properties": propSatisfies(isPopulated, "attributes"),
      "Must be an Object": propSatisfies(isType("Object"), "attributes"),
    }
  }
})

Each validation is a group, so you have a group of id validations. The keys can be messages, like above, or slugs (regular property names). You can read more about it in the [validus.js documentation][https://github.com/krainboltgreene/validus.js].

Now time to talk about what you'll get back when you run this:

const account = accountResource(incomingData)

Here's the return value:

{
  id: "1",
  type: "accounts",
  attributes: {
    name: "Kurtis Rainbolt-Greene",
    age: 27,
    email: "[e-mail]",
    username: "kurtis",
    createdAt: 2013-02-04T18:35:24.000Z,
    __abstraction__: {
      ...
    }
  },
  __abstraction__: {
    ...
  }
}

Okay, wait, what's this __abstraction__ nonsense? Well, that's where we keep the validations. Lets say id was missing, here's what you'd see from:

account._errors

It would return:

{
  id: [
    "Must have an id",
    "Must be an string",
    "Must have characters",
  ],
  ...
}

Isn't that great? You can also ask for individual validations using validus.js directly:

import {validate} from "validus"

validate(account._validations)("id")(account)

And that would return:

[
  "Must have an id",
  "Must be an string",
  "Must have characters",
]

Again, assuming id was missing. The initial validations are done when you pass in the data, and the validate is for later validation.