dynamodb-cursor-based-pagination v0.4.3

DynamoDB Cursor-Based Pagination

Introduction

DynamoDB

From Amazon DynamoDB page:

Amazon DynamoDB is a key-value and document database that delivers single-digit millisecond performance at any scale. It's a fully managed, multiregion, multimaster, durable database with built-in security, backup and restore, and in-memory caching for internet-scale applications. DynamoDB can handle more than 10 trillion requests per day and can support peaks of more than 20 million requests per second.

To achieve this hyper performance and scalability , it lacks common RDBMS and some NoSQL databases features. For that reason, DynamoDB provides only features that are scalable and its query is one of them.

Querying on DynamoDB is not powerful compared to other databases. It must satisfy some requirements:

1. The table or the secondary index must have a composite primary key (a partition/hash key and a sort/range key).
2. The partition key must be defined and the sort key is not required. If defined, it's used to sort the items.

For instance, you cannot query items whose hash key is different. Because of that you have to design your table and indexes to support all queries you need to perform.

Cursor-Based Pagination

If you're familiarized with GraphQL, you may have seen from its documentation that cursor-based pagination is more powerful than others pagination designs. Also, if we do some search, we can find comparisons among pagination designs (here) and cursor-based pagination is the winner.

Considering the advantages of cursor-based pagination, this package proposes a design to allow us perform cursor-based pagination in a DBB table.

Table Design

There aren't much requirements to achieve to be able to perform cursor-based pagination. In resume, we need:

1. A table or a secondary index with a composite primary key.
2. Items must be saved in such way that the range key ordination must represent the ordination of the pagination.

Why do we need the second requirement? First we need to understand what cursor is. Cursor is like an edge identifier, with cursor we must be able to retrieve and locate that edge on your backend. In a DBB table with composite primary key, the sort key is a good choice to be our cursor.

Installation

npm install -S dynamodb-cursor-based-pagination

or

yarn add dynamodb-cursor-based-pagination

You also need install aws-sdk in your project because it is a peer dependency of this project.

How to Use

import { paginate } from 'dynamodb-cursor-based-pagination';

paginate is a method whose signature is:

type paginate<T = any> = ({
  credentials,
  region,
  tableName,
  hashKeyName,
  hashKeyValue,
  rangeKeyName,
  indexName,
  projectionExpression,
  filterExpression,
  filterAttributeNames,
  filterAttributeValues,
  beginsWith,
  sort,
  after,
  first,
  before,
  last,
}: {
  credentials?: Credentials | undefined;
  region: string;
  tableName: string;
  hashKeyName: string;
  hashKeyValue: string;
  rangeKeyName: string;
  beginsWith?: string | undefined;
  indexName?: string | undefined;
  projectionExpression?: string | undefined;
  filterExpression?: string | undefined;
  filterAttributeNames?:
    | {
        [key: string]: string;
      }
    | undefined;
  filterAttributeValues?:
    | {
        [key: string]: any;
      }
    | undefined;
  sort?: 'ASC' | 'DESC' | undefined;
  after?: string | undefined;
  before?: string | undefined;
  first?: number | undefined;
  last?: number | undefined;
}) => Promise<{
  edges: {
    cursor: string;
    node: T;
  }[];
  pageInfo: {
    hasPreviousPage: boolean;
    hasNextPage: boolean;
    startCursor?: string | undefined;
    endCursor?: string | undefined;
  };
  consumedCapacity: number | undefined;
  count: number | undefined;
  scannedCount: number | undefined;
  lastEvaluatedKey: string | undefined;
}>;

Credentials

You must have AWS credentials in your environment to use this package. The only permission needed is dynamodb:Query.

If you don't have credentials in your environment, you may want provide them passing a Credentials object to credentials:

import { Credentials } from 'aws-sdk';
import { paginate } from 'dynamodb-cursor-based-pagination';

const credentials = new Credentials({
  accessKeyId: ...,
  secretAccessKey: ...,
  sessionToken: ...
})

paginate({
  credentials,
  ...
})
...

DynamoDB Table Parameters

The parameters region, tableName, hashKeyName, hashKeyValue, rangeKeyName, indexName are used to identify your DynamoDB table and the partition.

Cursor Parameters

• first and after: forward pagination arguments.
• last and before: backward pagination arguments.

Sorting

• sort: 'ASC' | 'DESC' (default 'DESC')

Querying on DynamoBD is related to the sorting of the items in function of their sort key value. Because of this, the parameter sort defines the items order before perform pagination. ASC is for ascending sorting (a, b, ..., z) and DESC, for descending (z, y, ..., a).

Begins With

• beginsWith: string | undefined

Your DynamoDB table may have an architecture that made the items have a beginsWith property. If you want to paginate over items that have such property, just add beginsWith to paginate method.

Projection Expression

• projectionExpression: string | undefined

DynamoDB projection expression reference.

Filtering

• filterExpression?: string | undefined
• filterAttributeNames: { [key: string]: string; } | undefined
• filterAttributeValues: { [key: string]: string; } | undefined

DynamoDB filtering reference.

// Example

paginate({
  ... // oher params,
  filterExpression: '#parity = :parity',
  filterAttributeNames: {
    '#parity': 'parity',
  },
  filterAttributeValues: {
    ':parity': 'EVEN',
  },
});

Examples

Let's check some examples to understand better these requirements and the design.

Test

To test against a real table:

yarn run test

To test against a real table, provide these values in your environment - you may want provide them in a .env file:

AWS_ACCESS_KEY_ID
AWS_SECRET_ACCESS_KEY
AWS_SESSION_TOKEN
HASH_KEY_NAME
HASH_KEY_VALUE
RANGE_KEY_NAME
TABLE_NAME
REGION
INDEX_NAME
BEGINS_WITH

Also, you need to populate the table with data to be tested. Please, refer to this section to add data to the table.

Author

• Pedro Arantes

License

MIT

Bootstrapped with tsdx

Made with ❤️