Micro Transform

Tiny library to transform objects between different states.

Suitable cases: serializing, deserializing, transforming into different shapes.

• Micro in size. No deps. ~500B.
• TS-first. All results of transformations are strictly typed and easily inspectable on hover.
• Versatile. Uses plain functions under the hood, can be composed in any way.

Install

npm install micro-transform

Usage

.setModelConfig

You can use this method to cherry-pick fields from a passed model. Possible values:

1. true — this will include the field as is, without any transformations.
2. false — this will exclude the field, that could be previously added by asterisk operator or previously in the chain.
3. function — we will execute it and pass both the model and the context into this function. If it'll return Promise, we'll await for it (those are executed in parallel).
4. nested transformers! Read more about it in the Nested Transformers section.

import { createTransformer } from "micro-transform";

const userExample = {
  id: "id",
  createdAt: 1602201600000,
  email: "[email protected]",
  password: "secret",
};
type User = typeof userExample;

const userSerializer = createTransformer<User>().setModelConfig({
  id: true,
  email: (user) => user.email.toLowerCase(),
});

const result = await userSerializer.transform(userExample);
// -> { id: "id", email: "[email protected]" }

Asterisk operator

There's a special case for the setModelConfig: you can pass a key "*" to include all fields instead of listing them one by one.

const userSerializer = createTransformer<User>().setModelConfig({
  "*": true,
  password: false,
});
const result = await userSerializer.transform(userExample);
// -> { id: "id", createdAt: 1602201600000, email: "[email protected]" }

Infer the result type

The library ships with a small importable helper that can help you work with the resulting data on the type level:

import type { TransformerResult } from 'micro-transform';

const dateSerializer = createTransformer<User>().setModelConfig({
  createdAt: (user) => formatDate(user.createdAt),
});

type SerializedDate = TransformerResult<typeof dateSerializer>;
// { createdAt: string }

.setCustomConfig

With this you can enchance models with new fields. The key would be the new field name. As of values, it pretty much repeats the model config:

1. function — that is the primary case, basically a computed property.
2. false — that will exclude the custom field previously added in the chain.

const userRoleSerializer = createTransformer<User>().setCustomConfig({
  role: async (user) => {
    return db.fetchRole(user.id);
  },
});

const result = await userRoleSerializer.transform(userExample);
// -> { role: "admin" }

Context

Sometimes you need a bit of extra context to transform data. Say, you need user's locale to pick the correct translation, or user's timezone to localize time fields.

In this case you can define context's structure its structure as the second type argument. You'll then need to pass it to transform function. If you do this right, you will be able to read the value in field-level transformer functions:

const userCreationSerializer = createTransformer<
  User,
  { timezone: string }
>().setModelConfig({
  createdAt: (user, ctx) => formatDate(user.createdAt, ctx.timezone),
});

const result = await userCreationSerializer.transform(userExample, {
  timezone: "Europe/Lisbon",
}); // -> { createdAt: "..." }

Nested Transformers

Since we accept functions on the field-level transformers, you could have made your own solution for nested transformers, but instead we added a built-in solution for that!

You can pass a transformer as field value, and it will:

1. correctly transform both arrays and single entities
2. pass on the context from the root transformer to nested transformers
3. evaluate all promises in parallel

type UserWithFriends = User & { friends: User[], bestFriend: User };

declare const user: UserWithFriends;

const friendSerializer = createTransformer<User>().setModelConfig({
  email: true,
});
const userSerializer = createTransformer<UserWithFriends>().setModelConfig({
  id: true,
  friends: friendSerializer,
  bestFriend: friendSerializer,
});

const serializedUser = await userSerializer.transform(user);
// { id: string, friends: { email: string }[], bestFriend: { email: string } }

Immutability and Chaining

As you might have noticed, the basic API involves chaining model configs and custom configs. And, as you might have guessed by the header, all the configs are merged (not replaced), and all the intermediate transformers are immutable.

You can you this to generate configs that are overall very similar, but differ in small details. For example, if you add or remove fields between different API versions, or user groups (admin, public, etc.).

const adminUserSerializer = createTransformer<User>()
  .setModelConfig({
    "*": true,
  })
  .setCustomConfig({ role: (user) => db.fetchRole(user.id) });

// Hiding password
const moderatorUserSerializer = adminUserSerializer.setModelConfig({
  password: false,
});

// Hiding role
const publicUserSerializer = moderatorUserSerializer.setCustomConfig({
  role: false,
});

Validations

Validations are out of the scope for this library. There's no validation of the shape of the incoming data. If you pass in garbage, the library will crash, and it's intended.

If you have untrusted/unexpected input, use any of the schema validation libraries out there, like zod, yup, valita and numerous others.