Type Inference

DynamoDB-Toolbox exposes several generic types to infer custom types from your schemas.

Which one you should use depends on your usage context, for instance, whether it’s within a write or a read operation.

Writes

For write operations, DynamoDB-Toolbox exposes the following generic types:

• ValidValue: A valid schema item
• InputValue: Similar to ValidValue, but with defaulted and linked attributes optional
• TransformedValue: A valid schema item after transformation

import type {
  InputValue,
  ValidValue,
  TransformedValue
} from 'dynamodb-toolbox/schema'

type Input = InputValue<typeof pokemonSchema>
type Valid = ValidValue<typeof pokemonSchema>
type Transformed = TransformedValue<typeof pokemonSchema>

By default, those generics use the put write mode, but you can switch to the key or update modes with the mode option. This impacts which the presence and requiredness of attributes:

type ValidKey = ValidValue<
  typeof pokemonSchema,
  { mode: 'key' }
>
type ValidUpdate = ValidValue<
  typeof pokemonSchema,
  { mode: 'update' }
>

Example

Here are step-by-step examples:

☝️ Schema

const now = () => new Date().toISOString()

const pokemonSchema = schema({
  // key attributes
  pokemonClass: string()
    .key()
    .transform(prefix('POKEMON'))
    .savedAs('partitionKey'),
  pokemonId: string().key().savedAs('sortKey'),

  // timestamps
  created: string().default(now),
  updated: string()
    .required('always')
    .putDefault(now)
    .updateDefault(now),

  // other attributes
  name: string().optional(),
  level: number().default(1)
}).and(prevSchema => ({
  levelPlusOne: number().link<typeof prevSchema>(
    ({ level }) => level + 1
  )
}))

🔎 'put' mode

InputValue

{
  "pokemonClass": "pikachu",
  "pokemonId": "123",
  "name": "Pikachu"
}

ValidValue

{
  "pokemonClass": "pikachu",
  "pokemonId": "123",
+ "created": "2022-01-01T00:00:00.000Z",
+ "modified": "2022-01-01T00:00:00.000Z",
  "name": "Pikachu",
+ "level": 1,
+ "levelPlusOne": 2,
}

TransformedValue

{
- "pokemonClass": "pikachu",
+ "partitionKey": "POKEMON#pikachu",
- "pokemonId": "123",
+ "sortKey": "123",
  "created": "2022-01-01T00:00:00.000Z",
  "modified": "2022-01-01T00:00:00.000Z",
  "name": "Pikachu",
  "level": 1,
  "levelPlusOne": 2,
}

🔎 'key' mode

InputValue

{
  "pokemonClass": "pikachu",
  "pokemonId": "123",
}
+ (Only key attributes are required)

ValidValue

{
  "pokemonClass": "pikachu",
  "pokemonId": "123",
}

TransformedValue

{
- "pokemonClass": "pikachu",
+ "partitionKey": "POKEMON#pikachu",
- "pokemonId": "123",
+ "sortKey": "123",
}

🔎 'update' mode

InputValue

{
  "pokemonClass": "bulbasaur",
  "pokemonId": "123",
  "name": "PlantyDino",
}

ValidValue

{
  "pokemonClass": "bulbasaur",
  "pokemonId": "123",
+ "modified": "2022-01-01T00:00:00.000Z",
  "name": "PlantyDino",
}

TransformedValue

{
- "pokemonClass": "bulbasaur",
+ "partitionKey": "POKEMON#bulbasaur",
- "pokemonId": "123",
+ "sortKey": "123",
  "modified": "2022-01-01T00:00:00.000Z",
  "name": "PlantyDino",
}

Reads

For read operations, DynamoDB-Toolbox exposes the following generic types:

• ReadValue: A valid schema item (differs from ValidValue as options are different, see below)
• FormattedValue: Similar to ReadValue, but with hidden attributes omitted

import type {
  ReadValue,
  FormattedValue
} from 'dynamodb-toolbox/schema'

type Read = ReadValue<typeof pokemonSchema>
type Formatted = FormattedValue<typeof pokemonSchema>

By default, those generics return complete items, but you can filter attributes and/or apply Partial (deeply) with the attributes and partial options:

type Filtered = FormattedValue<
  typeof pokemonSchema,
  { attributes: 'level' | 'name' | 'deep.attr[0].path' }
>
type Partial = FormattedValue<
  typeof pokemonSchema,
  { partial: true }
>