Virgil is a functional Cassandra client built using ZIO 2.x, Cats Effect 3.x, Magnolia and the Datastax 4.x Java drivers

Javadoc

libraryDependencies += "io.kaizen-solutions" %% "virgil-zio" % "<see badge for latest version>"

libraryDependencies += "io.kaizen-solutions" %% "virgil-cats-effect" % "<see badge for latest version>"

If you want to integrate another effect system (or runtime), depend on the core module and reference the implementations for ZIO & Cats Effect for inspiration:

libraryDependencies += "io.kaizen-solutions" %% "virgil-core" % "<see badge for latest version>"

Please note that Virgil is built for Scala 2.12.x, 2.13.x and 3.3.x but fully-automatic derivation is not present for 3.3.x.

You can follow along by checking out this repository and running docker-compose up which will bring up Datastax Enterprise Cassandra node along with Datastax Studio which provides a nice UI to interact with Cassandra. You have to create a new connection in Datastax Studio where you point to the container (since this is running in the same Docker network, we utilize the Docker DNS to resolve the hostname to the container IP and the hostname of the Cassandra cluster is datastax-enterprise):

Please keep reading if you want to follow along 👇

Given the following Cassandra keyspace:

CREATE KEYSPACE IF NOT EXISTS virgil
  WITH REPLICATION = {
    'class': 'SimpleStrategy',
    'replication_factor': 1
}

And the following Casandra table along with its User Defined Types (UDTs) (Make sure we are using the keyspace with USE virgil):

CREATE TYPE info (
  favorite BOOLEAN,
  comment TEXT
);

CREATE TYPE address (
  street TEXT,
  city TEXT,
  state TEXT,
  zip INT,
  data frozen<list<info>>
);

CREATE TABLE IF NOT EXISTS persons (
  id TEXT,
  name TEXT,
  age INT,
  past_addresses frozen<set<address>>,
  PRIMARY KEY ((id), age)
);

If we want to read and write data to this table, we create case classes that mirror the table and UDTs in Scala:

import io.kaizensolutions.virgil.annotations.CqlColumn

final case class Info(favorite: Boolean, comment: String)
final case class Address(street: String, city: String, state: String, zip: Int, data: List[Info])
final case class Person(
  id: String, 
  name: String, 
  age: Int, 
  @CqlColumn("past_addresses") addresses: Set[Address]
)

Note that the CqlColumn annotation can be used if the column/field name in the Cassandra table is different from the Scala representation. This can also be used inside User Defined Types as well.

If you are using Scala 3.3.x, you will need to use semi-automatic derivation as I have not yet figured out how to enable fully automatic derivation like Scala 2.x has.

final case class Info(favorite: Boolean, comment: String)
object Info:
    given cqlUdtValueEncoderForInfo: CqlUdtValueEncoder.Object[Info] = CqlUdtValueEncoder.derive[Info]
    given cqlUdtValueDecoderForInfo: CqlUdtValueDecoder.Object[Info] = CqlUdtValueDecoder.derive[Info]

final case class Address(street: String, city: String, state: String, zip: Int, data: List[Info])
object Address:
    given cqlUdtValueEncoderForAddress: CqlUdtValueEncoder.Object[Address] = CqlUdtValueEncoder.derive[Address]
    given cqlUdtValueDecoderForAddress: CqlUdtValueDecoder.Object[Address] = CqlUdtValueDecoder.derive[Address]

final case class Person(
    id: String,
    name: String,
    age: Int,
    @CqlColumn("past_addresses") addresses: Set[Address]
)
object Person:
    given cqlRowDecoderForPersonForPerson: CqlRowDecoder.Object[Person] = CqlRowDecoder.derive[Person]

Now that all the data-types are in place, we can write some data:

import io.kaizensolutions.virgil._
import io.kaizensolutions.virgil.dsl._

def insert(p: Person): CQL[MutationResult] =
  InsertBuilder("persons")
    .value("id", p.id)
    .value("name", p.name)
    .value("age", p.age)
    .value("past_addresses", p.addresses)
    .build

def setAddress(personId: String, personAge: Int, address: Address): CQL[MutationResult] =
  UpdateBuilder("persons")
    .set("past_addresses" := Set(address))
    .where("id" === personId)
    .and("age" === personAge)
    .build

We can also read data:

def select(personId: String, personAge: Int): CQL[Person] =
  SelectBuilder
    .from("persons")
    .columns("id", "name", "age", "past_addresses")
    .where("id" === personId)
    .and("age" === personAge)
    .build[Person]
    .take(1)

If you find that you have a complex query that cannot be expressed with the DSL yet, then you can use the lower level cql interpolator to express your query or mutation:

import io.kaizensolutions.virgil.cql._
def selectAll: CQL[Person] =
  cql"SELECT id, name, age, addresses FROM persons".query[Person]
  
def insertLowLevel(p: Person): CQL[MutationResult] =
  cql"INSERT INTO persons (id, name, age, addresses) VALUES (${p.id}, ${p.name}, ${p.age}, ${p.addresses}) USING TTL 10".mutation

Note that the lower-level API will turn the CQL into a string along with bind markers for each parameter and use bound statements under the hood, so you do not have to worry about CQL injection attacks.

If you want to string interpolate some part of the query because you may not know your table name up front (i.e. its passed through configuration, then you can use s"I am a String ${forExample}".appendCql(cql"continuing the cassandra query") or cql"SELECT * FROM ".appendString(s"$myTable")). Doing interpolation in cql is different from string interpolation as it will cause bind markers to be created.

You can also batch (i.e. Cassandra's definition of the word) mutations together by using +:

val batch: CQL[MutationResult]         = insert(p1) + update(p2.id, newPInfo) + insert(p3)
val unloggedBatch: CQL[MutationResult] = CQL.unlogged(batch)

Note: You cannot batch together queries and mutations as this is not allowed by Cassandra.

Now that we have built our CQL queries and mutations, we can execute them:

import zio._
import zio.stream._

// A single element stream is returned
val insertResult: ZStream[Has[CQLExecutor], Throwable, MutationResult] = insert(person).execute

// A stream of results is returned
val queryResult: ZStream[Has[CQLExecutor], Throwable, Person] = selectAll.execute

Running CQL queries and mutations is done through the CQLExecutor, which produces a ZStream that contains the results. You can obtain a CQLExecutor layer provided you have a CqlSessionBuilder from the Datastax Java Driver:

val dependencies: ULayer[Has[CQLExecutor]] = {
  val cqlSessionBuilderLayer: ULayer[Has[CqlSessionBuilder]] =
    ZLayer.succeed(
      CqlSession
        .builder()
        .withKeyspace("virgil")
        .withLocalDatacenter("dc1")
        .addContactPoint(InetSocketAddress.createUnresolved("localhost", 9042))
        .withApplicationName("virgil-tester")
  )
  val executor: ZLayer[Any, Throwable, Has[CQLExecutor]] = cqlSessionBuilderLayer >>> CQLExecutor.live
  executor.orDie
}

val insertResultReady: Stream[Throwable, MutationResult] = insertResult.provideLayer(dependencies)

Virgil provides all the default primitive data-types supported by the Datastax Java Driver. However, you can add support for your own primitive data-types. For example, if you want to add support for java.time.LocalDateTime, you can do so in the following manner (make sure to be extra careful when managing timezones):

import io.kaizensolutions.virgil.codecs._
import java.time.{LocalDateTime, ZoneOffset}

implicit val javaTimeInstantEncoder: CqlPrimitiveEncoder[LocalDateTime] =
  CqlPrimitiveEncoder[java.time.Instant].contramap[java.time.LocalDateTime](_.toInstant(ZoneOffset.UTC))

implicit val javaTimeInstantDecoder: CqlPrimitiveDecoder[LocalDateTime] =
  CqlPrimitiveDecoder[java.time.Instant].map[LocalDateTime](instant =>
    LocalDateTime.ofInstant(instant, ZoneOffset.UTC)
  )

This library is built on the Datastax Java Driver, please see the Datastax Java Driver documentation if you would like to configure the driver.

Virgil was an ancient Roman poet who composed an epic poem about Cassandra and so we thought it would be appropriate.

We were heavily inspired by Doobie, Cassandra4IO and Quill and wanted a more native ZIO solution for Cassandra focused on ergonomics, ease of use and performance (compile-time and runtime).

Special thanks to John De Goes, Francis Toth and Nigel Benns for their help, mentorship, and guidance. Shout out to Samuel Gómez for his significant contribution to the Cats Effect 3.x module.

We stand on the shoulders of giants, this work would not be possible without the effect systems and libraries that were used to build Virgil.

Virgil uses the excellent sbt-ci-release plugin to automate releases and leverages this plugin to publish artifacts to Sonatype & Maven Central.

As a fallback, you can also download Virgil from JitPack.

Add the JitPack resolver:

resolvers += "jitpack" at "https://jitpack.io"

Add the dependency:

libraryDependencies += "com.github.kaizen-solutions.virgil" %% "virgil-zio" % "<see JitPack for release>"
libraryDependencies += "com.github.kaizen-solutions.virgil" %% "virgil-cats-effect" % "<see JitPack for release>"

Virgil is used in production at the following companies:

• Caesars Digital
• Kaizen Solutions

Please let us know if you are using Virgil in production and would like to be added to this list.