Redis For Kotlin

Overview

Inspired by Kotlin Exposed, this library aims to provide a type-safe DSL for interacting with Redis Graph to Kotlin.

Kotlin Redis Graph

• Construct schemas for nodes
• Create, read, update and delete nodes and relationships
• Perform path queries

Setup

Gradle

Step 1. Add the JitPack repository to your build.gradle

allprojects {
    repositories {
        /* ... */
        maven { url "https://jitpack.io" }
    }
}

Step 2. Add the dependency

dependencies {
  // Core DSL
    implementation "com.github.mnbjhu.KotlinRedisGraph:core:$kotlinRedisVersion"
  // Annotations
    kapt "com.github.mnbjhu.KotlinRedisGraph:annotations:$kotlinRedisVersion"
}

Gradle Kotlin DSL

Step 1. Add the JitPack repository to your build.gradle.kts

repositories {
    mavenCentral()
    maven("https://jitpack.io")
}

Step 2. Add the dependency

dependencies {
  // Core DSL
  implementation("com.github.mnbjhu.KotlinRedisGraph:core:$kotlinRedisVersion")
  // Annotations
  implementation("com.github.mnbjhu.KotlinRedisGraph:annotations:$kotlinRedisVersion")
}

Basic Usage

Connect To Redis

To start using Redis Graph you need to create a new instance of the RedisGraph class with a graph name, host address and an option port (default is 6379).

import uk.gibby.redis.core.RedisGraph

val moviesGraph = RedisGraph(
    name = "movies",
    host = "raspberrypi.local",
)

Define A Schema

Node types and their relationships are defined by a schema. To create a node type, create a class which:

import uk.gibby.redis.core.UnitNode
import uk.gibby.redis.core.UnitRelation

class ActorNode: UnitNode(){
    val name by string()
    val actorId by long()
    val actedIn = relates(ActedInRelation::class)
}
class MovieNode : UnitNode(){
    val title by string()
    val releaseYear by long()
    val movieId by long()
}
class ActedInRelation: UnitRelation<ActorNode, MovieNode>(){
    val role by string()
}

Experimental CodeGen

This will generate the same as the above but will instead create instances of RedisRelation<ActedIn>, RedisNode<Movie > and RedisNode<Actor> respectively (as appose to UnitNode). This will allow you to return the nodes them self as the defined data classes.

import uk.gibby.redis.annotation.RedisType
import uk.gibby.redis.annotation.Node
import uk.gibby.redis.annotation.RelatesTo

@RedisType
data class ActedIn(val role: String)

@Node
data class Movie(val title: String, val releaseYear: Long)

@Node
@Relates(to = Movie::class, by = "actedIn", data = ActedIn::class)
data class Actor(val name: String)

Attributes can be defined on both RedisClass and RedisRelation. While in either scope, you'll have access to functions for setting the values of the attributes.

Currently Supported Types Are:

Primitives

Type	Function
String	string()
Long	long()
Double	double()
Boolean	boolean()

Arrays

class MyNode: UnitNode(){
    /* ND arrays of any result type are supported */
    val myArray by array(string())
    val my2DArray by array(array(string()))
}

Structured Types

import ...
/* Defines a type which can be returned */
class Vector3Result : StructResult<Vector3>() {
    val x: DoubleResult by double()
    val y: DoubleResult by double()
    val z: DoubleResult by double()
    
    fun ResultScope.getResult(): Vector3 = Vector3(!x, !y, !z)
    
    fun ParamMap.setResult(value: Vector3){
        x[value.x]
        y[value.y]
        z[value.z]
    }
}
/* Defining the attribute will allow you to store the type on a node or relation */
class Vector3Attribute : Vector3Result(), Attribute<Vector3>

class MyNode: UnitNode(){
    val myVector by ::Vector3Attribute
}

Experimental CodeGen Annotation

@RedisType
data class Vector(val x: Double, val y: Double, val z: Double)

Create Nodes

After a node type has been defined as a RedisClass, you can create a single instance like so:

moviesGraph.create(::MovieNode) {
    title["Star Wars: Episode V - The Empire Strikes Back"]
    releaseYear[1980]
    movieId[1]
}

Generated Cypher

CREATE (:MovieNode{title:'Star Wars: Episode V - The Empire Strikes Back', release_year:1980, movie_id:1})

(If an attribute is defined in the type but not set on creation, and exception will be thrown)

You can also create multiple instances by mapping elements from a list.

var index = 1L
val actors = listOf(
    "Mark Hamill",
    "Harrison Ford",
    "Carrie Fisher"
)
moviesGraph.create(::ActorNode, actors) {
    name[it]
    actorId[index++]
}

Generated Cypher

CREATE (:ActorNode{name:'Mark Hamill', actor_id:1}), (:ActorNode{name:'Harrison Ford', actor_id:2}), (:ActorNode{name:'Carrie Fisher', actor_id:3})

Query Scope

Currently, all other functionality is performed with the query function which takes a lambda returning a result value.

Example:

Returns a list of all the movie names in the graph.

moviesGraph.query {
    match(::MovieNode)
    movie.title
}

Create Relationships

moviesGraph.query {
    val (actor, movie) = match(::ActorNode, ::MovieNode)
    where ((actor.actorId eq 1) and (movie.movieId eq 1))
    create(actor - { actedIn { role["Luke Skywalker"] } } - movie)
}
moviesGraph.query {
    val (actor, movie) = match(::ActorNode, ::MovieNode)
    where ((actor.actorId eq 2) and (movie.movieId eq 1))
    create(actor - { actedIn{role["Han Solo"]} } - movie)
}
moviesGraph.query {
    val (actor, movie) = match(::ActorNode, ::MovieNode)
    where ( (actor.actorId eq 3) and (movie.movieId eq 1) )
    create(actor - { actedIn{role["Princess Leia"]} } - movie)
}

Generated Cypher

MATCH (actor:ActorNode), (movie:MovieNode)
WHERE (actor.actor_id = 1) AND (movie.movie_id = 1)
CREATE (actor)-[r:ACTED_IN {role:'Luke Skywalker'}]->(movie)

Alternatively you can create the node and edges as a single query:

graph.query {
    val movie = create(::MovieNode {
        it[title] = "Star Wars: Episode V - The Empire Strikes Back"
        it[releaseYear] = 1980
    })
    create(::ActorNode{ it[name] = "Mark Hamill" } - { actedIn { it[role] = "Luke Skywalker" } } - movie)
    create(::ActorNode{ it[name] = "Harrison Ford" } - { actedIn{ it[role] = "Han Solo" } } - movie)
    create(::ActorNode{ it[name] = "Carrie Fisher" } - { actedIn{ it[role] = "Princess Leia" } } - movie)
}

Constructing Queries

In this example we search for all movies and return the movie 'title'.

The same however we also return the 'releaseYear' and the 'movieId':

val movies = moviesGraph.query{
    val movie = match(::MovieNode)
    movie.title
}
movies `should contain` "Star Wars: Episode V - The Empire Strikes Back"

Generated Cypher

MATCH (movie:MovieNode)
RETURN movie.title

When using the @Node annotation we can return the movie node itself

@Node
data class Movie(val title: String, val releaseYear: Long, val movieId: Long)

val (title, releaseYear, id) = moviesGraph.query {
    val movie = match(::MovieNode)
    movie
}.first()

title as String `should be equal to` "Star Wars: Episode V - The Empire Strikes Back"
releaseYear as Long `should be equal to` 1980
id as Long `should be equal to` 1

Generated Cypher

MATCH (movie:MovieNode)
RETURN movie.title, movie.release_year, movie.movie_id

Here we:

• Search for an actor and a movie where the actor acted in the movie.
• Filter by movieId = 1
• And return the actor name and movie title

val actedInMovies = moviesGraph.query {
    val (actor, _, movie) = match(::ActorNode - { actedIn }  - ::MovieNode)
    where (movie.movieId eq 1)
    orderBy(actor.actorId)
    result(actor.name, movie.title)
}

actedInMovies.size `should be equal to` 3

val (actorName, movieName) = actedInMovieNodes.last()

actorName `should be equal to` "Carrie Fisher"
movieName `should be equal to` "Star Wars: Episode V - The Empire Strikes Back"

Generated Cypher

MATCH (actor:ActorNode)-[movieRelation:ACTED_IN]-(movie:MovieNode)
WHERE movie.movie_id = 1
RETURN actor.name, movie.title
ORDER BY actor.actor_id

Delete Nodes And Relationships

Any nodes or relationships referenced in the Query block can be deleted calling them in the (vararg) delete function:

val removedActor = moviesGraph.query {
    val (actor, relationship) = match(::ActorNode - { actedIn }  - ::MovieNode)
    where (actor.actorId eq 1)
    delete(relationship)
}

Generated Cypher

MATCH (actor:ActorNode)-[movieRelation:ACTED_IN]-(movie:MovieNode)
WHERE actor.actor_id = 1
RETURN movieRelation.role