Skip to content

Object mapping, and more, for Redis and Node.js. Written in TypeScript.


Notifications You must be signed in to change notification settings


Repository files navigation

Redis OM

Object mapping, and more, for Redis and Node.js. Written in TypeScript.

Discord Twitch YouTube Twitter

NPM Build License

Redis OM for Node.js makes it easy to model Redis data in your Node.js applications.

Redis OM .NET | Redis OM Node.js | Redis OM Python | Redis OM Spring

Table of contents

Redis OM for Node.js

Redis OM (pronounced REDiss OHM) makes it easy to add Redis to your Node.js application by mapping the Redis data structures you know and love to simple JavaScript objects. No more pesky, low-level commands, just pure code with a fluent interface.

Define a schema:

const schema = new Schema('album', {
  artist: { type: 'string' },
  title: { type: 'text' },
  year: { type: 'number' }

Create a JavaScript object and save it:

const album = {
  artist: "Mushroomhead",
  title: "The Righteous & The Butterfly",
  year: 2014


Search for matching entities:

const albums = await

Pretty cool, right? Read on for details.

⚠️ Warning: This Version Has Breaking Changes from 0.3.6

Redis OM 0.4 is new, improved, and includes breaking changes. If you're trying it for the first time, no worries. Just follow what's in this README and you'll be fine.

However, you might be a user of Redis OM already. If that is the case, you'll want to review this document to understand those changes.

Of course, you don't have to upgrade. If this is you, you'll want to check out the README for that version over on NPM.

However, I hope you choose to try the new version. It has many changes that have been frequently requested that are documented in the CHANGELOG. And more, non-breaking changes will follow these.

Getting Started

First things first, get yourself a Node.js project. There are lots of ways to do this, but I'm gonna go with a classic:

$ npm init

Once you have that sweet, sweet package.json, let's add our newest favorite package to it:

$ npm install redis-om

Redis OM for Node.js uses Node Redis. So you should install that too:

$ npm install redis

And, of course, you'll need some Redis, preferably Redis Stack as it comes with RediSearch and RedisJSON ready to go. The easiest way to do this is to set up a free Redis Cloud instance. But, you can also use Docker:

$ docker run -d --name redis-stack -p 6379:6379 -p 8001:8001 redis/redis-stack:latest

Excellent. Setup done. Let's write some code!

Connect to Redis with Node Redis

Before you can use Redis OM, you need to connect to Redis with Node Redis. Here's how you do that, stolen straight from the top of the Node Redis README:

import { createClient } from 'redis'

const redis = createClient()
redis.on('error', (err) => console.log('Redis Client Error', err));
await redis.connect()

Node Redis is a powerful piece of software with lots and lots of capabilities. Its details are way beyond the scope of this README. But, if you're curious—or if you need that power—you can find all the info in the Node Redis documentation.

Regardless, once you have a connection to Redis you can use it to execute Redis commands:

const aString = await // 'PONG'
const aNumber = await redis.hSet('foo', 'alfa', '42', 'bravo', '23') // 2
const aHash = await redis.hGetAll('foo') // { alfa: '42', bravo: '23' }

You might not need to do this, but it's always handy to have the option. When you're done with a Redis connection, you can let the server know by calling .quit:

await redis.quit()

Redis Connection Strings

By default, Node Redis connects to localhost on port 6379. This is, of course, configurable. Just pass in a url with the hostname and port that you want to use:

const redis = createClient({ url: 'redis://alice:foobared@awesome.redis.server:6380' })

The basic format for this URL is:


This will probably cover most scenarios, but if you want something more, the full specification for the URL is defined with the IANA. And yes, there is a TLS version as well.

Node Redis has lots of other ways you can create a connection. You can use discrete parameters, UNIX sockets, and all sorts of cool things. Details can be found in the client configuration guide for Node Redis and the clusterting guide.

Entities and Schemas

Redis OM is all about saving, reading, and deleting entities. An Entity is just data in a JavaScript object that you want to save or retrieve from Redis. Almost any JavaScript object is a valid Entity.

Schemas define fields that might be on an Entity. It includes a field's type, how it is stored internally in Redis, and how to search on it if you are using RediSearch. By default, they are mapped to JSON documents using RedisJSON, but you can change it to use Hashes if want (more on that later).

Ok. Let's start doing some object mapping and create a Schema:

import { Schema } from 'redis-om'

const albumSchema = new Schema('album', {
  artist: { type: 'string' },
  title: { type: 'text' },
  year: { type: 'number' },
  genres: { type: 'string[]' },
  songDurations: { type: 'number[]' },
  outOfPublication: { type: 'boolean' }

const studioSchema = new Schema('studio', {
  name: { type: 'string' },
  city: { type: 'string' },
  state: { type: 'string' },
  location: { type: 'point' },
  established: { type: 'date' }

The first argument is the Schema name. It defines the key name prefix that entities stored in Redis will have. It should be unique for your particular instance of Redis and probably meaningful to what you're doing. Here we have selected album for our album data and studio for data on recording studios. Imaginative, I know.

The second argument defines fields that might be stored in that key. The property name is the name of the field that you'll be referencing in your Redis OM queries. The type property tells Redis OM what sort of data is in that field. Valid types are: string, number, boolean, string[], number[], date, point, and text.

The first three types do exactly what you think—they define a field that is a String, a Number, or a Boolean. string[] and number[] do what you'd think as well, specifically describing an Array of Strings or Numbers respectively.

date is a little different, but still more or less what you'd expect. It describes a property that contains a Date and can be set using not only a Date but also a String containing an ISO 8601 date or a number with the UNIX epoch time in seconds (NOTE: the JavaScript Date object is specified in milliseconds).

A point defines a point somewhere on the globe as a longitude and a latitude. It is expressed as a simple object with longitude and latitude properties. Like this:

const point = { longitude: 12.34, latitude: 56.78 }

A text field is a lot like a string. If you're just reading and writing objects, they are identical. But if you want to search on them, they are very, very different. I'll cover that in detail when I talk about searching but the tl;dr is that string fields can only be matched on their exact value and are best for keys and discrete data—like postal codes or status indicators—while text fields have full-text search enabled on them, are optimized for human-readable text, and can take advantage of stemming and stop words.

JSON and Hashes

As I mentioned earlier, by default Redis OM stores your entities in JSON documents using RedisJSON. You can make this explicit in code if you like:

const albumSchema = new Schema('album', {
  artist: { type: 'string' },
  title: { type: 'string' },
  year: { type: 'number' },
  genres: { type: 'string[]' },
  songDurations: { type: 'number[]' },
  outOfPublication: { type: 'boolean' }
}, {
  dataStructure: 'JSON'

But you can also store your entities as Hashes instead. Just change the dataStructure property to reflect it:

const albumSchema = new Schema('album', {
  artist: { type: 'string' },
  title: { type: 'string' },
  year: { type: 'number' },
  genres: { type: 'string[]' },
  outOfPublication: { type: 'boolean' }
}, {
  dataStructure: 'HASH'

And that's it.

Of course, Hashes and JSON are somewhat different data structures. Hashes are flat with fields containing values. JSON documents, however, are trees and can have depth and—most excitingly—can be nested. This difference is reflected in how Redis OM maps data to entities and how you configure your Schema.

Note that I have not included the songDurations in the Hash. This is because number[] is only possible when working with JSON, it will generate an error if you try to use it with Hashes.

Configuring JSON

When you store your entities as JSON, the path to the properties in your JSON document and your JavaScript object default to the name of your property in the schema. In the above example, this would result in a document that looks like this:

  "artist": "Mushroomhead",
  "title": "The Righteous & The Butterfly",
  "year": 2014,
  "genres": [ "metal" ],
  "songDurations": [ 204, 290, 196, 210, 211, 105, 244, 245, 209, 252, 259, 200, 215, 219 ],
  "outOfPublication": true

However, you might not want your JavaScript object and your JSON to map this way. So, you can provide a path option in your schema that contains a JSONPath pointing to where that field actually exists in the JSON and your entity. For example, we might want to store some of the album's data inside of an album property like this:

  "album": {
    "artist": "Mushroomhead",
    "title": "The Righteous & The Butterfly",
    "year": 2014,
    "genres": [ "metal" ],
    "songDurations": [ 204, 290, 196, 210, 211, 105, 244, 245, 209, 252, 259, 200, 215, 219 ]
  "outOfPublication": true

To do this, we'll need to specify the path property for the nested fields in the schema:

const albumSchema = new Schema('album', {
  artist: { type: 'string', path: '$.album.artist' },
  title: { type: 'string', path: '$.album.title' },
  year: { type: 'number', path: '$.album.year' },
  genres: { type: 'string[]', path: '$.album.genres[*]' },
  songDurations: { type: 'number[]', path: '$.album.songDurations[*]' },
  outOfPublication: { type: 'boolean' }

There are two things to note here:

  1. We haven't specified a path for outOfPublication as it's still in the root of the document. It defaults to $.outOfPublication.
  2. Our genres field points to a string[]. When using a string[] the JSONPath must return an array. If it doesn't, an error will be generated.
  3. Same for our songDurations.

Configuring Hashes

When you store your entities as Hashes there is no nesting—all the entities are flat. In Redis, the properties on your entity are stored in fields inside a Hash. The default name for each field is the name of the property in your schema and this is the name that will be used in your entities. So, for the following schema:

const albumSchema = new Schema('album', {
  artist: { type: 'string' },
  title: { type: 'string' },
  year: { type: 'number' },
  genres: { type: 'string[]' },
  outOfPublication: { type: 'boolean' }
}, {
  dataStructure: 'HASH'

In your code, your entities would look like this:

  artist: 'Mushroomhead',
  title: 'The Righteous & The Butterfly',
  year: 2014,
  genres: [ 'metal' ],
  outOfPublication: true

Inside Redis, your Hash would be stored like this:

Field Value
artist Mushroomhead
title The Righteous & The Butterfly
year 2014
genres metal
outOfPublication 1

However, you might not want the names of your fields and the names of the properties on your entity to be exactly the same. Maybe you've got some existing data with existing names or something.

Fear not! You can change the name of the field used by Redis with the field property:

const albumSchema = new Schema('album', {
  artist: { type: 'string', field: 'album_artist' },
  title: { type: 'string', field: 'album_title' },
  year: { type: 'number', field: 'album_year' },
  genres: { type: 'string[]' },
  outOfPublication: { type: 'boolean' }
}, {
  dataStructure: 'HASH'

With this configuration, your entities will remain unchanged and will still have properties for artist, title, year, genres, and outOfPublication. But inside Redis, the field will have changed:

Field Value
album_artist Mushroomhead
album_title The Righteous & The Butterfly
album_year 2014
genres metal
outOfPublication 1

Reading, Writing, and Removing with Repository

Now that we have a client and a schema, we have what we need to make a repository. A repository provides the means to write, read, and remove entities. Creating a repository is pretty straightforward—just instantiate one with a schema and a client:

import { Repository } from 'redis-om'

const albumRepository = new Repository(albumSchema, redis)
const studioRepository = new Repository(studioSchema, redis)

Once we have a repository, we can use .save to, well, save entities:

let album = {
  artist: "Mushroomhead",
  title: "The Righteous & The Butterfly",
  year: 2014,
  genres: [ 'metal' ],
  songDurations: [ 204, 290, 196, 210, 211, 105, 244, 245, 209, 252, 259, 200, 215, 219 ],
  outOfPublication: true

album = await

This saves your entity and returns a copy, a copy with some additional properties. The primary property we care about right now is the entity ID, which Redis OM will generate for you. However, this isn't stored and accessed like a typical property. After all, you might have a property in your data with a name that conflicts with the name Redis OM uses and that would create all sorts of problems.

So, Redis OM uses a Symbol to access it instead. You'll need to import this symbol from Redis OM:

import { EntityId } from 'redis-om'

Then you can access the entity ID using that symbol:

album = await
album[EntityId] // '01FJYWEYRHYFT8YTEGQBABJ43J'

The entity ID that Redis OM generates is a ULID and is a unique id representing that object. If you don't like using generated IDs for some reason and instead want to provide your own, you can totally do that:

album = await'BWOMP', album)

Regardless, once you have an object's entity ID you can .fetch with it:

const album = await albumRepository.fetch('01FJYWEYRHYFT8YTEGQBABJ43J')
album.artist // "Mushroomhead"
album.title // "The Righteous & The Butterfly"
album.year // 2014
album.genres // [ 'metal' ]
album.songDurations // [ 204, 290, 196, 210, 211, 105, 244, 245, 209, 252, 259, 200, 215, 219 ]
album.outOfPublication // true

If you call .save with an entity that already has an entity ID, probably because you fetched it, .save will update it instead of creating a new Entity:

let album = await albumRepository.fetch('01FJYWEYRHYFT8YTEGQBABJ43J')
album.genres = [ 'metal', 'nu metal', 'avantgarde' ]
album.outOfPublication = false

album = await

You can even use .save to clone an Entity. Just pass in a new entity ID to .save and it'll save the data to that entity ID:

const album = await albumRepository.fetch('01FJYWEYRHYFT8YTEGQBABJ43J')
album.genres = [ 'metal', 'nu metal', 'avantgarde' ]
album.outOfPublication = false

const clonedEntity = await'BWOMP', album)

And, of course, you need to be able to delete things. Use .remove to do that:

await albumRepository.remove('01FJYWEYRHYFT8YTEGQBABJ43J')

You can also set an entity to expire after a certain number of seconds. Redis will automatically remove that entity when the time's up. Use the .expire method to do this:

const ttlInSeconds = 12 * 60 * 60  // 12 hours
await albumRepository.expire('01FJYWEYRHYFT8YTEGQBABJ43J', ttlInSeconds)

Missing Entities and Null Values

Redis, and by extension Redis OM, doesn't differentiate between missing and null—particularly for Hashes. Missing fields in Redis Hashes are returned as null. Missing keys also return null. So, if you fetch an entity that doesn't exist, it will happily return you an empty entity, complete with the provided entity ID:

const album = await albumRepository.fetch('TOTALLY_BOGUS')
album[EntityId] // 'TOTALLY_BOGUS'
album.artist // undefined
album.title // undefined
album.year // undefined
album.genres // undefined
album.outOfPublication // undefined

Conversely, if you remove all the properties on an entity and then save it, it will remove the entity from Redis:

const album = await albumRepository.fetch('01FJYWEYRHYFT8YTEGQBABJ43J')
delete album.artist
delete album.title
delete album.year
delete album.genres
delete album.outOfPublication

const entityId = await

const exists = await redis.exists('album:01FJYWEYRHYFT8YTEGQBABJ43J') // 0

It does this because Redis doesn't distinguish between missing and null. You could have an entity that is empty. Or you could not have an entity at all. Redis doesn't know which is your intention, and so always returns something when you call .fetch.


Using RediSearch with Redis OM is where the power of this fully armed and operational battle station starts to become apparent. If you have RediSearch installed on your Redis server you can use the search capabilities of Redis OM. This enables commands like:

const albums = await

Let's explore this in full.

Build the Index

To use search you have to build an index. If you don't, you'll get errors. To build an index, just call .createIndex on your repository:

await albumRepository.createIndex();

If you change your schema, no worries. Redis OM will automatically rebuild the index for you. Just call .createIndex again. And don't worry if you call .createIndex when your schema hasn't changed. Redis OM will only rebuild your index if the schema has changed. So, you can safely use it in your startup code.

However, if you have a lot of data, rebuilding an index can take some time. So, you might want to explicitly manage the building and rebuilding of your indices in some sort of deployment code script thing. To support those devops sorts of things, Redis OM includes a .dropIndex method to explicitly remove an index without rebuilding it:

await albumRepository.dropIndex();

You probably won't use this in your application, but if you come up with a cool use for it, I'd love to hear about it!

Finding All The Things (and Returning Them)

Once you have an index created (or recreated) you can search. The most basic search is to just return all the things. This will return all of the albums that you've put in Redis:

const albums = await


It's possible you have a lot of albums; I know I do. In that case, you can page through the results. Just pass in the zero-based offset and the number of results you want:

const offset = 100
const count = 25
const albums = await, count)

Don't worry if your offset is greater than the number of entities. If it is, you just get an empty array back. No harm, no foul.

First Things First

Sometimes you only have one album. Or maybe you only care about the first album you find. You can easily grab the first result of your search with .first:

const firstAlbum = await;

Note: If you have no albums, this will return null. And I feel sorry for you.


Sometimes you just want to know how many albums you have. For that, you can call .count:

const count = await

Finding Specific Things

It's fine and dandy to return all the things. But that's not what you usually want to do. You want to find specific things. Redis OM will let you find those specific things by strings, numbers, and booleans. You can also search for strings that are in an array, perform full-text search within strings, search by date, and search for points on the globe within a particular area.

And it does it with a fluent interface that allows—but does not demand—code that reads like a sentence. See below for exhaustive examples of all the syntax available to you.

Searching on Strings

When you set the field type in your schema to string, you can search for a particular value in that string. You can also search for partial strings (no shorter than two characters) that occur at the beginning, middle, or end of a string. If you need to search strings in a more sophisticated manner, you'll want to look at the text type and search it using the Full-Text Search syntax.

let albums

// find all albums where the artist is 'Mushroomhead'
albums = await'artist').eq('Mushroomhead').return.all()

// find all albums where the artist is *not* 'Mushroomhead'
albums = await'artist').not.eq('Mushroomhead').return.all()

// find all albums using wildcards
albums = await'artist').eq('Mush*').return.all()
albums = await'artist').eq('*head').return.all()
albums = await'artist').eq('*room*').return.all()

// fluent alternatives that do the same thing
albums = await'artist').equals('Mushroomhead').return.all()
albums = await'artist').does.equal('Mushroomhead').return.all()
albums = await'artist').is.equalTo('Mushroomhead').return.all()
albums = await'artist').does.not.equal('Mushroomhead').return.all()
albums = await'artist').is.not.equalTo('Mushroomhead').return.all()

Searching on Numbers

When you set the field type in your schema to number, you can store both integers and floating-point numbers. And you can search against it with all the comparisons you'd expect to see:

let albums

// find all albums where the year is ===, >, >=, <, and <= 1984
albums = await'year').eq(1984).return.all()
albums = await'year').gt(1984).return.all()
albums = await'year').gte(1984).return.all()
albums = await'year').lt(1984).return.all()
albums = await'year').lte(1984).return.all()

// find all albums where the year is between 1980 and 1989 inclusive
albums = await'year').between(1980, 1989).return.all()

// find all albums where the year is *not* ===, >, >=, <, and <= 1984
albums = await'year').not.eq(1984).return.all()
albums = await'year')
albums = await'year').not.gte(1984).return.all()
albums = await'year')
albums = await'year').not.lte(1984).return.all()

// find all albums where year is *not* between 1980 and 1989 inclusive
albums = await'year').not.between(1980, 1989);

// fluent alternatives that do the same thing
albums = await'year').equals(1984).return.all()
albums = await'year').does.equal(1984).return.all()
albums = await'year').does.not.equal(1984).return.all()
albums = await'year').is.equalTo(1984).return.all()
albums = await'year').is.not.equalTo(1984).return.all()

albums = await'year').greaterThan(1984).return.all()
albums = await'year').is.greaterThan(1984).return.all()
albums = await'year').is.not.greaterThan(1984).return.all()

albums = await'year').greaterThanOrEqualTo(1984).return.all()
albums = await'year').is.greaterThanOrEqualTo(1984).return.all()
albums = await'year').is.not.greaterThanOrEqualTo(1984).return.all()

albums = await'year').lessThan(1984).return.all()
albums = await'year').is.lessThan(1984).return.all()
albums = await'year').is.not.lessThan(1984).return.all()

albums = await'year').lessThanOrEqualTo(1984).return.all()
albums = await'year').is.lessThanOrEqualTo(1984).return.all()
albums = await'year').is.not.lessThanOrEqualTo(1984).return.all()

albums = await'year').is.between(1980, 1989).return.all()
albums = await'year').is.not.between(1980, 1989).return.all()

Searching on Booleans

You can search against fields that contain booleans if you defined a field type of boolean in your schema:

let albums

// find all albums where outOfPublication is true
albums = await'outOfPublication').true().return.all()

// find all albums where outOfPublication is false
albums = await'outOfPublication').false().return.all()

You can negate boolean searches. This might seem odd, but if your field is null, then it would match on a .not query:

// find all albums where outOfPublication is false or null
albums = await'outOfPublication').not.true().return.all()

// find all albums where outOfPublication is true or null
albums = await'outOfPublication').not.false().return.all()

And, of course, there's lots of syntactic sugar to make this fluent:

albums = await'outOfPublication').eq(true).return.all()
albums = await'outOfPublication').equals(true).return.all()
albums = await'outOfPublication').does.equal(true).return.all()
albums = await'outOfPublication').is.equalTo(true).return.all()

albums = await'outOfPublication').true().return.all()
albums = await'outOfPublication').false().return.all()
albums = await'outOfPublication').is.true().return.all()
albums = await'outOfPublication').is.false().return.all()

albums = await'outOfPublication').not.eq(true).return.all()
albums = await'outOfPublication').does.not.equal(true).return.all()
albums = await'outOfPublication').is.not.equalTo(true).return.all()
albums = await'outOfPublication').is.not.true().return.all()
albums = await'outOfPublication').is.not.false().return.all()

Searching on Dates

If you have a field type of date in your schema, you can search on it using Dates, ISO 8601 formatted strings, or the UNIX epoch time in seconds:

studios = await'established').on(new Date('2010-12-27')).return.all()
studios = await'established').on('2010-12-27').return.all()
studios = await'established').on(1293408000).return.all()

There are several date comparison methods to use. And they can be negated:

const date = new Date('2010-12-27')
const laterDate = new Date('2020-12-27')

studios = await'established').on(date).return.all()
studios = await'established').not.on(date).return.all()
studios = await'established').before(date).return.all()
studios = await'established').not.before(date).return.all()
studios = await'established').after(date).return.all()
studios = await'established').not.after(date).return.all()
studios = await'established').onOrBefore(date).return.all()
studios = await'established').not.onOrBefore(date).return.all()
studios = await'established').onOrAfter(date).return.all()
studios = await'established').not.onOrAfter(date).return.all()
studios = await'established').between(date, laterDate).return.all()
studios = await'established').not.between(date, laterDate).return.all()

More fluent variations work too:

const date = new Date('2010-12-27')
const laterDate = new Date('2020-12-27')

studios = await'established').is.on(date).return.all()
studios = await'established').is.not.on(date).return.all()

studios = await'established').is.before(date).return.all()
studios = await'established').is.not.before(date).return.all()

studios = await'established').is.onOrBefore(date).return.all()
studios = await'established').is.not.onOrBefore(date).return.all()

studios = await'established').is.after(date).return.all()
studios = await'established').is.not.after(date).return.all()

studios = await'established').is.onOrAfter(date).return.all()
studios = await'established').is.not.onOrAfter(date).return.all()

studios = await'established').is.between(date, laterDate).return.all()
studios = await'established').is.not.between(date, laterDate).return.all()

And, since dates are really just numbers, all the numeric comparisons work too:

const date = new Date('2010-12-27')
const laterDate = new Date('2020-12-27')

studios = await'established').eq(date).return.all()
studios = await'established').not.eq(date).return.all()
studios = await'established').equals(date).return.all()
studios = await'established').does.equal(date).return.all()
studios = await'established').does.not.equal(date).return.all()
studios = await'established').is.equalTo(date).return.all()
studios = await'established').is.not.equalTo(date).return.all()

studios = await'established').gt(date).return.all()
studios = await'established')
studios = await'established').greaterThan(date).return.all()
studios = await'established').is.greaterThan(date).return.all()
studios = await'established').is.not.greaterThan(date).return.all()

studios = await'established').gte(date).return.all()
studios = await'established').not.gte(date).return.all()
studios = await'established').greaterThanOrEqualTo(date).return.all()
studios = await'established').is.greaterThanOrEqualTo(date).return.all()
studios = await'established').is.not.greaterThanOrEqualTo(date).return.all()

studios = await'established').lt(date).return.all()
studios = await'established')
studios = await'established').lessThan(date).return.all()
studios = await'established').is.lessThan(date).return.all()
studios = await'established').is.not.lessThan(date).return.all()

studios = await'established').lte(date).return.all()
studios = await'established').not.lte(date).return.all()
studios = await'established').lessThanOrEqualTo(date).return.all()
studios = await'established').is.lessThanOrEqualTo(date).return.all()
studios = await'established').is.not.lessThanOrEqualTo(date).return.all()

Searching String Arrays

If you have a field type of string[] you can search for whole strings that are in that array:

let albums

// find all albums where genres contains the string 'rock'
albums = await'genres').contain('rock').return.all()

// find all albums where genres contains the string 'rock', 'metal', or 'blues'
albums = await'genres').containOneOf('rock', 'metal', 'blues').return.all()

// find all albums where genres does *not* contain the string 'rock'
albums = await'genres').not.contain('rock').return.all()

// find all albums where genres does *not* contain the string 'rock', 'metal', and 'blues'
albums = await'genres').not.containOneOf('rock', 'metal', 'blues').return.all()

// alternative syntaxes
albums = await'genres').contains('rock').return.all()
albums = await'genres').containsOneOf('rock', 'metal', 'blues').return.all()
albums = await'genres').does.contain('rock').return.all()
albums = await'genres').does.not.contain('rock').return.all()
albums = await'genres').does.containOneOf('rock', 'metal', 'blues').return.all()
albums = await'genres').does.not.containOneOf('rock', 'metal', 'blues').return.all()

Wildcards work here too:

albums = await'genres').contain('*rock*').return.all()

Searching Arrays of Numbers

If you have a field of type number[], you can search on it just like a number. If any number in the array matches your criteria, then it'll match and the document will be returned.

let albums

// find all albums where at least one song is at least 3 minutes long
albums = await'songDuration').gte(180).return.all()

// find all albums where at least one song is at exactly 3 minutes long
albums = await'songDuration').eq(180).return.all()

// find all albums where at least one song is between 3 and 4 minutes long
albums = await'songDuration').between(180, 240).return.all()

I'm not going to include all the examples again. Just go check out the section on searching on numbers.

Full-Text Search

If you've defined a field with a type of text in your schema, you can store text in it and perform full-text searches against it. Full-text search is different from how a string is searched. With full-text search, you can look for words, partial words, fuzzy matches, and exact phrases within a body of text.

Full-text search is optimized for human-readable text and it's pretty clever. It understands that certain words (like a, an, or the) are common and ignores them. It understands how words relate to each other and so if you search for give, it matches gives, given, giving, and gave too. It ignores punctuation and whitespace.

Here are some examples of doing full-text search against some album titles:

let albums

// finds all albums where the title contains the word 'butterfly'
albums = await'title').match('butterfly').return.all()

// finds all albums using fuzzy matching where the title contains a word which is within 3 Levenshtein distance of the word 'buterfly'
albums = await'title').match('buterfly', { fuzzyMatching: true, levenshteinDistance: 3 }).return.all()

// finds all albums where the title contains the words 'beautiful' and 'children'
albums = await'title').match('beautiful children').return.all()

// finds all albums where the title contains the exact phrase 'beautiful stories'
albums = await'title').matchExact('beautiful stories').return.all()

If you want to search for a part of a word. To do it, just tack a * on the beginning or end (or both) of your partial word and it'll match accordingly:

// finds all albums where the title contains a word that contains 'right'
albums = await'title').match('*right*').return.all()

Do not combine partial-word searches or fuzzy matches with exact matches. Partial-word searches and fuzzy matches with exact matches are not compatible in RediSearch. If you try to exactly match a partial-word search or fuzzy match a partial-word search, you'll get an error.

albums = await'title').matchExact('beautiful sto*').return.all()
albums = await'title').matchExact('*buterfly', { fuzzyMatching: true, levenshteinDistance: 3 }).return.all()

As always, there are several alternatives to make this a bit more fluent and, of course, negation is available:

albums = await'title').not.match('butterfly').return.all()
albums = await'title').matches('butterfly').return.all()
albums = await'title').does.match('butterfly').return.all()
albums = await'title').does.not.match('butterfly').return.all()

albums = await'title').exact.match('beautiful stories').return.all()
albums = await'title').not.exact.match('beautiful stories').return.all()
albums = await'title').exactly.matches('beautiful stories').return.all()
albums = await'title').does.exactly.match('beautiful stories').return.all()
albums = await'title').does.not.exactly.match('beautiful stories').return.all()

albums = await'title').not.matchExact('beautiful stories').return.all()
albums = await'title').matchesExactly('beautiful stories').return.all()
albums = await'title').does.matchExactly('beautiful stories').return.all()
albums = await'title').does.not.matchExactly('beautiful stories').return.all()

Searching on Points

RediSearch, and therefore Redis OM, both support searching by geographic location. You specify a point in the globe and a radius and it'll gleefully return all the entities within that radius:

let studios

// finds all the studios with 50 miles of downtown Cleveland
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).miles).return.all()

Note that coordinates are specified with the longitude first, and then the latitude. This might be the opposite of what you expect but is consistent with how Redis implements coordinates in RediSearch and with GeoSets.

If you don't want to rely on argument order, you can also specify longitude and latitude more explicitly:

// finds all the studios within 50 miles of downtown Cleveland using a point
studios = await'location').inRadius(
  circle => circle.origin({ longitude: -81.7758995, latitude: 41.4976393 }).radius(50).miles).return.all()

// finds all the studios within 50 miles of downtown Cleveland using longitude and latitude
studios = await'location').inRadius(
  circle => circle.longitude(-81.7758995).latitude(41.4976393).radius(50).miles).return.all()

Radius can be in miles, feet, kilometers, and meters in all the spelling variations you could ever want:

// finds all the studios within 50 miles
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).miles).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).mile).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).mi).return.all()

// finds all the studios within 50 feet
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).feet).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).foot).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).ft).return.all()

// finds all the studios within 50 kilometers
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).kilometers).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).kilometer).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).km).return.all()

// finds all the studios within 50 meters
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).meters).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).meter).return.all()

studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50).m).return.all()

If you don't specify the origin, Redis OM will use a longitude 0.0 and a latitude 0.0, also known as Null Island:

// finds all the studios within 50 miles of Null Island (probably ain't much there)
studios = await'location').inRadius(
  circle => circle.radius(50).miles).return.all()

If you don't specify the radius, it defaults to 1 and if you don't provide units, it defaults to meters:

// finds all the studios within 1 meter of downtown Cleveland
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393)).return.all()

// finds all the studios within 1 kilometer of downtown Cleveland
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).kilometers).return.all()

// finds all the studios within 50 meters of downtown Cleveland
studios = await'location').inRadius(
  circle => circle.origin(-81.7758995, 41.4976393).radius(50)).return.all()

And there are plenty of fluent variations to help make your code pretty:

studios = await'location').not.inRadius(
  circle => circle.longitude(-81.7758995).latitude(41.4976393).radius(50).miles).return.all()

studios = await'location').is.inRadius(
  circle => circle.longitude(-81.7758995).latitude(41.4976393).radius(50).miles).return.all()

studios = await'location').is.not.inRadius(
  circle => circle.longitude(-81.7758995).latitude(41.4976393).radius(50).miles).return.all()

studios = await'location').not.inCircle(
  circle => circle.longitude(-81.7758995).latitude(41.4976393).radius(50).miles).return.all()

studios = await'location').is.inCircle(
  circle => circle.longitude(-81.7758995).latitude(41.4976393).radius(50).miles).return.all()

studios = await'location').is.not.inCircle(
  circle => circle.longitude(-81.7758995).latitude(41.4976393).radius(50).miles).return.all()

Chaining Searches

So far we've been doing searches that match on a single field. However, we often want to query on multiple fields. Not a problem:

const albums = await

These are executed in order from left to right, and ignore any order of operations. So this query will match an artist of "Mushroomhead" OR a title matching "butterfly" before it goes on to match that the year is greater than 1990.

If you'd like to change this you can nest your queries:

const albums = await
  .or(search => search

This query finds all Mushroomhead albums after 1990 or albums that have "butterfly" in the title.

Running Raw Searches

The fluent search interface is nice, but sometimes you need to do something just a bit more. If you want, you can execute a search against your entities using the native RediSearch query syntax. I'm not going to explain the syntax here as it's a bit involved, but you can read it for yourself in the RediSearch documentation.

To execute a raw search, just call .searchRaw on the repository with your query:

// finds all the Mushroomhead albums with the word 'beautiful' in the title from 1990 and beyond
const query = "@artist:{Mushroomhead} @title:beautiful @year:[1990 +inf]"
const albums = albumRepository.searchRaw(query).return.all();

The nice thing here is that it returns the same entities that you've been using for everything else. It's just a lower-level way of executing a query for when you need that extra bit of power.

Sorting Search Results

RediSearch provides a basic mechanism for sorting your search results and Redis OM exposes it. You can sort on a single field and can sort on the following types: string, number, boolean, date, and text. To sort, simply call .sortBy, .sortAscending, or .sortDescending:

const albumsByYear = await

const albumsByTitle = await
    .sortBy('title', 'DESC').return.all()

You can also tell RediSearch to preload the sorting index to improve performance when you sort. This doesn't work with all of the types that you can sort by, but it's still pretty useful. To preload the index, mark the field in the Schema with the sortable property:

const albumSchema = new Schema(Album, {
  artist: { type: 'string' },
  title: { type: 'text', sortable: true },
  year: { type: 'number', sortable: true },
  genres: { type: 'string[]' },
  outOfPublication: { type: 'boolean' }

If your schema is for a JSON data structure (the default), you can mark number, date, and text fields as sortable. You can also mark string and boolean fields as sortable, but this will have no effect and will generate a warning.

If your schema is for a Hash, you can mark string, number, boolean, date, and text fields as sortable.

Fields of the types point and string[] are never sortable.

If this seems like a confusing flowchart to parse, don't worry. If you call .sortBy on a field in the Schema that's not marked as sortable and it could be, Redis OM will log a warning to let you know.

Advanced Stuff

This is a bit of a catch-all for some of the more advanced stuff you can do with Redis OM.

Schema Options

Additional field options can be set depending on the field type. These correspond to the Field Options available when creating a RediSearch full-text index. Other than the separator option, these only affect how content is indexed and searched.

schema type RediSearch type indexed sortable normalized stemming matcher weight separator caseSensitive
string TAG yes HASH Only HASH Only - - - yes yes
number NUMERIC yes yes - - - - - -
boolean TAG yes HASH Only - - - - - -
string[] TAG yes HASH Only HASH Only - - - yes yes
number[] NUMERIC yes yes - - - - - -
date NUMERIC yes yes - - - - -
point GEO yes - - - - - -
text TEXT yes yes yes yes yes yes - -
  • indexed: true | false, whether this field is indexed by RediSearch (default true)
  • sortable: true | false, whether to create an additional index to optimize sorting (default false)
  • normalized: true | false, whether to apply normalization for sorting (default true)
  • matcher: string defining phonetic matcher which can be one of: 'dm:en' for English, 'dm:fr' for French, 'dm:pt' for Portugese, 'dm:es' for Spanish (default none)
  • stemming: true | false, whether word-stemming is applied to text fields (default true)
  • weight: number, the importance weighting to use when ranking results (default 1)
  • separator: string, the character to delimit multiple tags (default '|')
  • caseSensitive: true | false, whether original letter casing is kept for search (default false)

Example showing additional options:

const commentSchema = new Schema(Comment, {
  name: { type: 'text', stemming: false, matcher: 'dm:en' },
  email: { type: 'string', normalized: false, },
  posted: { type: 'date', sortable: true },
  title: { type: 'text', weight: 2 },
  comment: { type: 'text', weight: 1 },
  approved: { type: 'boolean', indexed: false },