Skip to content

ArsenyYankovsky/nova-odm

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Nova ODM / Amazon DynamoDB DataMapper For JavaScript

GitHub GitHub Workflow Status (with branch) npm npm bundle size (scoped)

A schema-based object to document mapper for Amazon DynamoDB.

This project is a fork and a drop-in replacement of the original dynamodb-data-mapper-js. The goal of this project is to continue maintaining the project. One major step on this way is already done: this project is migrated to use AWS SDK v3. Read the migration guide here for more details.

Table of Contents

Getting Started

Installation

npm install @nova-odm/mapper @nova-odm/annotations reflect-metadata
yarn add @nova-odm/mapper @nova-odm/annotations reflect-metadata

Import reflect-metadata at the top level of your application:

import 'reflect-metadata'

Defining a Model

Start by defining a model using decorators from the @nova-odm/annotations package.

import {
  attribute,
  hashKey,
  rangeKey,
  table,
} from '@nova-odm/annotations';

@table('table_name')
class MyDomainObject {
  public constructor(partial?: Partial<MyDomainObject>) {
    Object.assign(this, partial)
  }
  
  @hashKey()
  id: string;

  @rangeKey({defaultProvider: () => new Date()})
  createdAt: Date;

  @attribute()
  completed?: boolean;
}

We also highly recommend adding a partial constructor to your domain classes as shown above.

Simple Usage

With domain classes defined, you can interact with records in DynamoDB via an instance of DataMapper:

import { DataMapper } from '@nova-odm/mapper';
import { DynamoDBClient } from '@aws-sdk/client-dynamodb';

const mapper = new DataMapper({
    client: new DynamoDBClient({ region: 'us-west-2' }), // the SDK client used to execute operations
    tableNamePrefix: 'dev_' // optionally, you can provide a table prefix to keep your dev and prod tables separate
});

put

Creates (or overwrites) an item in the table

const savedObject = await mapper.put(new MyDomainObject({id: 'foo'}));

get

Retrieves an item from DynamoDB

const item = await mapper.get(new MyDomainObject({ id: 'foo', createdAt: new Date(946684800000) }))

NB: The promise returned by the mapper will be rejected with an ItemNotFoundException if the item sought is not found.

update

Updates an item in the table

const item = await mapper.get(new MyDomainObject({ id: 'foo', createdAt: new Date(946684800000) }));

item.completed = true;

await mapper.update(item);

Also supports partial updates:

await this.mapper.update({
  item: new MyDomainObject({
    id,
    completed: true,
  }),
  onMissing: 'skip',
})

This will not update a the createdAt property of the item in this example.

delete

Removes an item from the table

await mapper.delete(new MyDomainObject({ id: 'foo', createdAt: new Date(946684800000) }));

scan

Lists the items in a table or index

for await (const item of mapper.scan(MyDomainObject)) {
    // individual items will be yielded as the scan is performed
}

// Optionally, scan an index instead of the table:
for await (const item of mapper.scan(MyDomainObject, { indexName: 'myIndex' })) {
    // individual items will be yielded as the scan is performed
}

You can also use the pages() method to read items in pages and get the last evaluated key:

const paginator = mapper.scan(MyDomainObject, {
  startKey: {
    id: 'foo',
    createdAt: new Date(946684800000),
  },
}).pages()

const domainObjects = []

for await (const page of paginator) {
  // page will be an array of items yielded from the scan
  // Note: the last evaluated key is automatically passed to the next scan operation
  domainObjects.push(...page)
  
  // You can also access the last evaluated key as an object of the shape of your model:
  console.log(paginator.lastEvaluatedKey)
}

query

Finds a specific item (or range of items) in a table or index

for await (const foo of mapper.query(MyDomainObject, { id: 'foo' })) {
    // individual items with a hash key of "foo" will be yielded as the query is performed
}

You can also use the pages() method to read items in pages and get the last evaluated key:

const paginator = mapper.query(MyDomainObject, {
  type: 'Equals',
  subject: 'id',
  object: 'foo',
}, {
  startKey: {
    id: 'foo',
    createdAt: new Date(946684800000),
  },
}).pages()

const domainObjects = []

for await (const page of paginator) {
  // page will be an array of items yielded from the scan
  // Note: the last evaluated key is automatically passed to the next scan operation
  domainObjects.push(...page)
  
  // You can also access the last evaluated key as an object of the shape of your model:
  console.log(paginator.lastEvaluatedKey)
}

Batch Operations

The mapper also supports batch operations. Under the hood, the batch will automatically be split into chunks that fall within DynamoDB's limits (25 for batchPut and batchDelete, 100 for batchGet). The items can belong to any number of tables, and exponential backoff for unprocessed items is handled automatically.

batchPut

Creates (or overwrites) multiple items in the table

const toSave = [
    new MyDomainObject({id: 'foo', completed: false}),
    new MyDomainObject({id: 'bar', completed: false}),
];
for await (const persisted of mapper.batchPut(toSave)) {
    // items will be yielded as they are successfully written
}

batchGet

Fetches multiple items from the table

const toGet = [
    new MyDomainObject({id: 'foo', createdAt: new Date(946684800000)}),
    new MyDomainObject({id: 'bar', createdAt: new Date(946684800001)}),
];
for await (const found of mapper.batchGet(toGet)) {
    // items will be yielded as they are successfully retrieved
}

NB: Only items that exist in the table will be retrieved. If a key is not found, it will be omitted from the result.

batchDelete

Removes multiple items from the table

const toRemove = [
    new MyDomainObject({ id: 'foo', createdAt: new Date(946684800000) }),
    new MyDomainObject({ id: 'bar', createdAt: new Date(946684800001) }),
];
for await (const found of mapper.batchDelete(toRemove)) {
    // items will be yielded as they are successfully removed
}

Operations with Expressions

Application example

The following example shows how to use the mapper with expressions. It will only insert a new record if the email is not already in use.

import {
    AttributePath,
    FunctionExpression,
    UpdateExpression,
} from '@nova-odm/expressions';

// given the entity below
@table('tableName')
class MyRecord {
    public constructor(partial?: Partial<MyRecord>) {
        Object.assign(this, partial)
    }
  
    @hashKey()
    email?: string;

    @attribute()
    passwordHash?: string;

    @attribute()
    passwordSalt?: string;

    @attribute()
    verified?: boolean;

    @attribute()
    verifyToken?: string;
}

// you make a mapper operation as follows
const aRecord = new MyRecord({
    email,
    passwordHash: password,
    passwordSalt: salt,
    verified: false,
    verifyToken: token,
});

const result = await mapper.put(aRecord, { 
    condition: new FunctionExpression('attribute_not_exists', new AttributePath('email')) 
})

Table lifecycle operations

createTable

Creates a table for the mapped class and waits for it to be initialized:

await mapper.createTable(MyDomainObject, {readCapacityUnits: 5, writeCapacityUnits: 5})

ensureTableExists

Like createTable, but only creates the table if it doesn't already exist:

await mapper.ensureTableExists(MyDomainObject, {readCapacityUnits: 5, writeCapacityUnits: 5})

deleteTable

Deletes the table for the mapped class and waits for it to be removed:

await mapper.deleteTable(MyDomainObject)

ensureTableNotExists

Like deleteTable, but only deletes the table if it exists:

await mapper.ensureTableNotExists(MyDomainObject)

Advanced Usage

Optimistic Locking

The Nova ODM supports optimistic locking via the versionAttribute decorator.

import { attribute, hashKey, rangeKey, table, versionAttribute } from '@nova-odm/annotations'
import { State, JourneyModel } from '../journey'

@table('posts')
export class Post {
  public constructor(partial?: Partial<Post>) {
    Object.assign(this, partial)
  }

  @hashKey()
  public id: string

  @attribute()
  public title: string

  @attribute()
  public text: string

  @versionAttribute()
  public version: number
}

Every time an item is saved, the version attribute will be incremented. If the version attribute is not present on the item, it will be set to 1. If the version attribute is present but does not match the version of the item in the table, the save will fail with a ConditionalCheckFailedException.

You can also skip the version check/increment by passing skipVersionCheck: true to the put or update methods.

await mapper.update({
  item: new Post({
    id: '123',
    title: 'My Post',
  }),
  onMissing: 'skip',
  skipVersionCheck: true,
})

Using with esbuild

Nova ODM is written in TypeScript and can be used with esbuild to bundle your application. However, esbuild does not emit decorator metadata, so we recommend using a esbuild-plugin-tsc plugin.

First, install the plugin and typescript:

npm install --save-dev esbuild-plugin-tsc typescript

or

yarn add -D esbuild-plugin-tsc typescript

Then, add the plugin to your esbuild configuration:

Javascript:

+const esbuildPluginTsc = require('esbuild-plugin-tsc');
 ...
 esbuild.build({
   ...
   plugins: [
+    esbuildPluginTsc(),
   ],
 })

Typescript:

+import esbuildPluginTsc from 'esbuild-plugin-tsc';
 ...
 esbuild.build({
   ...
   plugins: [
+    esbuildPluginTsc(),
   ],
 })

Migrating from the original dynamodb-data-mapper-js

This project provides drop-in replacement packages for the original packages. Replace your dependencies / imports with the following respective packages:

dynamodb-data-mapper-js Nova ODM
@aws/dynamodb-data-mapper @nova-odm/mapper
@aws/dynamodb-query-iterator @nova-odm/query-iterator
@aws/dynamodb-data-marshaller @nova-odm/marshaller
@aws/dynamodb-expressions @nova-odm/expressions
@aws/dynamodb-batch-iterator @nova-odm/batch-iterator
@aws/dynamodb-auto-marshaller @nova-odm/auto-marshaller
@aws/dynamodb-data-mapper-annotations @nova-odm/annotations

Constituent packages

The Nova ODM is developed as a monorepo using lerna. More detailed documentation about the ODM's constituent packages is available by viewing those packages directly.

About

A schema-based data mapper for Amazon DynamoDB.

Resources

License

Stars

Watchers

Forks

Packages

No packages published