Build a business layer with a Repository

A Repository is the next layer above the data stores. When setting up a repository, you can create DataSets that can rely on a previously configured physical store. The difference is that while PhysicalStore focuses on the data, DataSet focuses on business logic. You can authorize, check permissions, subscribe to entity changes, etc...

23 Jun 2021•3 min read

Build a business layer with a Repository

Disclaimer

If you are using a repository, use the repository all the time. - Anonymous

As callbacks, authorization and data manipulation works through the Repository layer, you should always use Repository if you have dependant business logic through it.

Simplifying a complex idea

Let’s say you have an entity type. You have to be notified each and every time somebody creates a new instance. You also have to take care about that new entities should have some validation.

class MyEntity {
  public key!: number
  public value!: string
}

const myInjector = new Injector()
  .setupStores((sm) => sm.addStore(new InMemoryStore({ model: MyEntity, primaryKey: 'key' })))
  .setupRepository((repo) =>
    repo.createDataSet(MyEntity, 'key', {
      authorizeAdd: async ({ entity }) => {
        if (entity.value && entity.value.length > 2)
          return {
            isAllowed: true,
          }
        return {
          isAllowed: false,
          message: `Failed to create entity. The value length should be greater than 2 but was ${
            entity.value?.length
          }. Entity: ${JSON.stringify(entity)}`,
        }
      },
    }),
  )

const dataSet = myInjector.getDataSetFor(MyEntity, 'key')

dataSet.onEntityAdded.subscribe(({ entity }) => {
  console.log('Hey, a new entity has been added', entity)
})

dataSet.add(myInjector, { key: 1, value: 'a' }) // Will fail
dataSet.add(myInjector, { key: 1, value: 'asd' }) // Will pass and will be logged to the console

Event subscriptions

You can subscribe to the following events in your Repository:

onEntityAdded will notify you when a new entity has been created
onEntityUpdated will fire on updates
onEntityRemoved will notify about the deletions

Authorizing operations

You can define authorization / validation during you create the DataSet. These callbacks are the followings:

authorizeAdd will verify entities before adding / inserting them into the stores
authorizeGet / authorizeGetEntity will check that you can get that single entity (also authorizeGetEntity won’t check collections)
authorizeUpdate / authorizeUpdateEntity will verify before update. authorizeUpdate won’t load the entity from the Store (you won’t have as a parameter) while authorizeUpdateEntity does
authorizeRemove / authorizeRemoveEntity will fire before remove. Again, authorizeRemoveEntity loads the whole entity from the store before deleting it.
with addFilter, you can append conditions to your filter expressions to narrow collection queries and optimize performance

Manipulating Data

There are also two callbacks that allows you to modify data: with modifyOnAdd and modifyOnUpdate, you can update data in some fields like e.g. createdByUser or lastModificationDate

Getting the Context

All methods above has an injector instance on the call parameter - you can use that injector to get service instances from the right caller context. It means that you can e.g.: get the current user in the following way.

      authorizeAdd: async ({ injector }) => {
        const currentUser = await injector.getCurrentUser()
        if (currentUser.roles.find((role) => role === 'CanAddMyEntity'))
          return {
            isAllowed: true,
          }
        return {
          isAllowed: false,
          message: "The user doesn't have the role 'CanAddMyEntity'",
        }
      },