The DynamoDB Toolbox is a set of tools that makes it easy to work with Amazon DynamoDB and the DocumentClient. It's designed with Single Tables in mind, but works just as well with multiple tables. It lets you define your Entities (with typings and aliases) and map them to your DynamoDB tables. You can then generate the API parameters to put, get, delete, update, query, scan, batchGet, and batchWrite data by passing in JavaScript objects. The DynamoDB Toolbox will map aliases, validate and coerce types, and even write complex UpdateExpressions for you. 😉

Learn more about single table design in Alex Debrie's blog.

Feedback is welcome and much appreciated! (Huge thanks to @ThomasAribart for all his work on this 🙌)

• Full Documentation
• GitHub Discussions

ℹ️ We're using aws-sdk v2 DynamoDB tools, the support for aws-sdk v3 is on its way.
You can read more about the development here.

Using your favorite package manager, install DynamoDB Toolbox and aws-sdk v2 in your project by running one of the following commands:

# npm
npm i dynamodb-toolbox
npm install aws-sdk

# yarn
yarn add dynamodb-toolbox
yarn add aws-sdk

Require or import Table and Entity from dynamodb-toolbox:

import { Table, Entity } from 'dynamodb-toolbox'

Create a Table (with the DocumentClient using aws-sdk v2):

import DynamoDB from 'aws-sdk/clients/dynamodb'

const DocumentClient = new DynamoDB.DocumentClient({
  // Specify your client options as usual
  convertEmptyValues: false
})

// Instantiate a table
const MyTable = new Table({
  // Specify table name (used by DynamoDB)
  name: 'my-table',

  // Define partition and sort keys
  partitionKey: 'pk',
  sortKey: 'sk',

  // Add the DocumentClient
  DocumentClient
})

Create an Entity:

const Customer = new Entity({
  // Specify entity name
  name: 'Customer',

  // Define attributes
  attributes: {
    id: { partitionKey: true }, // flag as partitionKey
    sk: { hidden: true, sortKey: true }, // flag as sortKey and mark hidden
    age: { type: 'number' }, // set the attribute type
    name: { type: 'string', map: 'data' }, // map 'name' to table attribute 'data'
    emailVerified: { type: 'boolean', required: true }, // specify attribute as required
    co: { alias: 'company' }, // alias table attribute 'co' to 'company'
    status: ['sk', 0], // composite key mapping
    date_added: ['sk', 1] // composite key mapping
  },

  // Assign it to our table
  table: MyTable

  // In Typescript, the "as const" statement is needed for type inference
} as const)

Put an item:

// Create an item (using table attribute names or aliases)
const customer = {
  id: 123,
  age: 35,
  name: 'Jane Smith',
  emailVerified: true,
  company: 'ACME',
  status: 'active',
  date_added: '2020-04-24'
}

// Use the 'put' method of Customer:
await Customer.put(customer)

The item will be saved to DynamoDB like this:

{
  "pk": 123,
  "sk": "active#2020-04-24",
  "age": 35,
  "data": "Jane Smith",
  "emailVerified": true,
  "co": "ACME",
  // Attributes auto-generated by DynamoDB-Toolbox
  "_et": "customer", // Entity name (required for parsing)
  "_ct": "2021-01-01T00:00:00.000Z", // Item creation date (optional)
  "_md": "2021-01-01T00:00:00.000Z" // Item last modification date (optional)
}

You can then get the data:

// Specify primary key
const primaryKey = {
  id: 123,
  status: 'active',
  date_added: '2020-04-24'
}

// Use the 'get' method of Customer
const response = await Customer.get(primaryKey)

Since v0.4, the method inputs, options and response types are inferred from the Entity definition:

await Customer.put({
  id: 123,
  // ❌ Sort key is required ("sk" or both "status" and "date_added")
  age: 35,
  name: ['Jane', 'Smith'], // ❌ name should be a string
  emailVerified: undefined, // ❌ attribute is marked as required
  company: 'ACME'
})

const { Item: customer } = await Customer.get({
  id: 123,
  status: 'active',
  date_added: '2020-04-24' // ✅ Valid primary key
})
type Customer = typeof customer
// 🙌 Type is equal to:
type ExpectedCustomer =
  | {
      id: any
      age?: number | undefined
      name?: string | undefined
      emailVerified: boolean
      company?: any
      status: any
      date_added: any
      entity: string
      created: string
      modified: string
    }
  | undefined

See Type Inference in the documentation for more details.

• Table Schemas and DynamoDB Typings: Define your Table and Entity data models using a simple JavaScript object structure, assign DynamoDB data types, and optionally set defaults.
• Magic UpdateExpressions: Writing complex UpdateExpression strings is a major pain, especially if the input data changes the underlying clauses or requires dynamic (or nested) attributes. This library handles everything from simple SET clauses, to complex list and set manipulations, to defaulting values with smartly applied if_not_exists() to avoid overwriting data.
• Bidirectional Mapping and Aliasing: When building a single table design, you can define multiple entities that map to the same table. Each entity can reuse fields (like pk andsk) and map them to different aliases depending on the item type. Your data is automatically mapped correctly when reading and writing data.
• Composite Key Generation and Field Mapping: Doing some fancy data modeling with composite keys? Like setting your sortKey to [country]#[region]#[state]#[county]#[city]#[neighborhood] model hierarchies? DynamoDB Toolbox lets you map data to these composite keys which will both autogenerate the value and parse them into fields for you.
• Type Coercion and Validation: Automatically coerce values to strings, numbers and booleans to ensure consistent data types in your DynamoDB tables. Validate list, map, and set types against your data. Oh yeah, and sets are automatically handled for you. 😉
• Powerful Query Builder: Specify a partitionKey, and then easily configure your sortKey conditions, filters, and attribute projections to query your primary or secondary indexes. This library can even handle pagination with a simple .next() method.
• Simple Table Scans: Scan through your table or secondary indexes and add filters, projections, parallel scans and more. And don't forget the pagination support with .next().
• Filter and Condition Expression Builder: Build complex Filter and Condition expressions using a standardized array and object notation. No more appending strings!
• Projection Builder: Specify which attributes and paths should be returned for each entity type, and automatically filter the results.
• Secondary Index Support: Map your secondary indexes (GSIs and LSIs) to your table, and dynamically link your entity attributes.
• Batch Operations: Full support for batch operations with a simpler interface to work with multiple entities and tables.
• Transactions: Full support for transaction with a simpler interface to work with multiple entities and tables.
• Default Value Dependency Graphs: Create dynamic attribute defaults by chaining other dynamic attribute defaults together.
• TypeScript Support: v0.4 of this library provides strong typing support AND type inference 😍. Inferred type can still overriden with Overlays. Some Utility Types are also exposed. Additional work is still required to support schema validation & typings.

• DocumentClient SDK Reference
• Best Practices for DynamoDB
• DynamoDB, explained.
• The DynamoDB Book

Contributions, ideas and bug reports are welcome and greatly appreciated. Please add issues for suggestions and bug reports or create a pull request. You can also contact me on Twitter: @jeremy_daly.