Installation
Compiled for Scala 2.12. You can find jars compiled for 2.11 and 2.13 below
For ZIO 1
libraryDependencies += "io.github.karimagnusson" % "kuzminki-zio" % "0.9.2"
For ZIO 2
libraryDependencies += "io.github.karimagnusson" % "kuzminki-zio-2" % "0.9.2"
Example
import zio._
import zio.console._
import zio.blocking._
import kuzminki.api._
object ExampleApp extends zio.App {
class Client extends Model("client") {
val id = column[Int]("id")
val username = column[String]("username")
val age = column[Int]("age")
def all = (id, username, age)
}
val client = Model.get[Client]
val job = for {
_ <- sql
.insert(client)
.cols2(t => (t.username, t.age))
.run(("Joe", 35))
_ <- sql
.update(client)
.set(_.age ==> 24)
.where(_.id === 4)
.run
_ <- sql.delete(client).where(_.id === 7).run
clients <- sql
.select(client)
.cols3(_.all)
.where(_.age > 25)
.limit(5)
.run
_ <- ZIO.foreach(clients) {
case (id, username, age) =>
putStrLn(s"$id $username $age")
}
} yield ()
val dbConfig = DDbConfig.forDb("company").getConfig
val dbLayer = Kuzminki.layer(dbConfig)
override def run(args: List[String]): ZIO[ZEnv, Nothing, ExitCode] = {
job.provideCustomLayer(dbLayer).exitCode
}
}
Results
// query
val stm = sql.select(client).cols3(_.all).where(_.age > 25).limit(5)
stm.run and db.query(stm.render) returns RIO[Has[Kuzminki] with Blocking, List[T]]
stm.runHead and db.queryHead(stm.render) returns RIO[Has[Kuzminki] with Blocking, T]
stm.runHeadOpt and db.queryHeadOpt(stm.render) returns RIO[Has[Kuzminki] with Blocking, Option[T]]
// operation
val stm = sql.update(client).set(_.age ==> 24).where(_.id === 4)
stm.run and db.exec(stm.render) returns RIO[Has[Kuzminki] with Blocking, Unit]
stm.runNum and db.execNum(stm.render) returns RIO[Has[Kuzminki] with Blocking, Int]
Config
val dbConfig = DbConfig
.forDb("{DB-NAME}")
.withPoolSize(10) // default = 10
.withHost("{HOST}") // default = localhost
.withPort("{PORT}") // default = 5432
.withUser("{USER}}")
.withPassword("{PASSWORD}")
.withOptions(Map(...))
.getConfig
Layer
Create a layer to make the driver instance accessable under Has[Kuzminki]
object MyApp extends zio.App {
val job = // ...
val kuzminkiLayer = Kuzminki.layer(dbConfig)
override def run(args: List[String]): ZIO[ZEnv, Nothing, ExitCode] = {
job.provideCustomLayer(kuzminkiLayer).exitCode
}
}
Custom connection
Create an driver instance of the driver
for {
db <- Kuzminki.create(dbConfig)
users <- db.query(fetchSomeUsers)
} yield users
Split connection
If you wish to have two connection pools, one for SELECT and another for INSERT, UPDATE, DELETE, you can use layerSplit. To create an instance rather than a Layer, use createSplit.
// for .query .queryHead .queryHeadOpt
val getConfig = DbConfig.forDb("company").withPoolSize(5).getConfig
// for .exec .execNum
val setConfig = DbConfig.forDb("company").withPoolSize(5).getConfig
val dbLayer = Kuzminki.layerSplit(getConfig, setConfig)
Creating a model
Column types are listed under Data types
import kuzminki.api._
import java.sql.Timestamp
class User extends Model("user_profile") {
val id = column[Int]("id")
val username = column[String]("username")
val email = column[String]("email")
val name = column[String]("name")
val age = column[Int]("age")
val gender = column[String]("gender")
val country = column[String]("country")
val city = column[String]("city")
val discount = column[Int]("discount")
val isActive = column[Boolean]("is_active")
val created = column[Timestamp]("created")
}
Model.register[User]
Custom methods
You can create custom methods to access columns that you regularly use.
class Customer extends Model("customer") {
val id = column[Int]("id")
val userId = column[Int]("user_id")
val spending = column[Int]("spending")
def all = (id, userId, spending)
}
Create a model instance
Model.register[User] creates an instance of the model for later use and makes sure there is only one instance of the model. Model.register[User] gets an existing instance of the model. If it does not exist, it is created.
Model.register[User]
// ...
val user = Model.get[User]
Select query
You select the columns as tuple of model columns. The query will return tuple of the column types. In this case Seq[Tuple2[Int, String]]. If you need more than 22 columns you can use colsVector or colsType. To order by ASC rather than DESC use orderBy(_.age.asc).
import kuzminki.api._
val user = Model.get[User]
sql
.select(user)
.cols2(t => (
t.id,
t.username
))
.where(t => Seq(
t.gender === "f",
t.age > 25
))
.orderBy(_.age.desc)
.limit(10)
.run
SELECT
"id",
"username"
FROM "user_profile"
WHERE "gender" = 'f'
AND "age" > 25
ORDER BY "age" DESC
LIMIT 10
Row as case class
case class ProductItem(id: Int, name: String, price: Float)
class Product extends Model("product") {
val id = column[Int]("id")
val name = column[String]("name")
val price = column[Float]("price")
val item = read[ProductItem](id, name, price)
}
val product = Model.get[Product]
sql
.select(product)
.colsType(_.item)
.where(_.price < 100.0)
.run
// returns List[ProductItem]
Row as Vector
sql
.select(user)
.colsVector(t => Vector(
t.id,
t.username,
t.email
))
.where(_.age < 25)
.run
// returns List[Vector[Any]]
Where
Refer to Operators for a list of operators.
.where(_.id > 100)
.where(t => Seq(
t.gender === "f",
t.age > 25
))
AND / OR
import kuzminki.fn._
.where(t => Seq(
t.age > 25,
Or(
t.country === "RU",
t.country === "FR"
)
))
// WHERE "age" > 25 AND ("country" == 'RU' OR "country" == 'FR')
.where(t => Or(
And(
t.country === "RU",
t.city === "Moscow"
),
And(
t.country === "FR",
t.city === "Paris"
)
))
// WHERE ("country" == 'RU' AND "city" == 'Moscow') OR ("country" == 'FR' AND "city" == 'Paris')
Optional conditions
Optional conditions for example from http GET request.
.whereOpt(_.id > Some(100))
.whereOpts(t => Seq(
t.gender === None,
t.age > Some(25)
))
// WHERE "age" > 25
.whereOpts(t => Seq(
t.age > Some(25),
Or.opts(
t.country === Some("RU"),
t.country === Some("FR")
)
))
// WHERE "age" > 25 AND ("country" == 'RU' OR "country" == 'FR')
Null values
If you have null values in a column you can use one of the folloing ways to turn the result into Option[T]
Convert the column in the model from T to Option[T]
val city = column[String]("city").asOpt
Convert the column in the query from T to Option[T]
cols1(_.city.asOpt)
Use Posgres's coalesce function to return a default value in case of null
import kuzminki.fn._
cols1(t => Fn.coalesce(t.city, "No city"))
Nested query
class Newsletter extends Model("newsletter") {
val email = column[String]("email")
val isSubscribed = column[Boolean]("is_subscribed")
}
val newsletter = Model.get[Newsletter]
sql
.select(user)
.cols1(_.username)
.where(_.email.in(
sql
.select(newsletter)
.cols1(_.email)
.where(_.isSubscribed === true)
))
.run
SELECT "username"
FROM "user_profile"
WHERE "email" = ANY(
SELECT "email"
FROM "newsletter"
WHERE "is_subscribed" = true
)
Cache
For increased performance you can cache your queries. The SQL string will be created only once and you get the same performance as you get with raw queries.
val stm = sql
.select(user)
.cols1(_.username)
.all
.orderBy(_.age.asc)
.cacheWhere2(t => (
t.country.cacheEq,
t.age.cacheGt
))
stm.run(("CN", 25))
SELECT "username"
FROM "user_profile"
WHERE "country" = 'CN'
AND "age" > 25
ORDER BY "age" ASC
Cached with WHERE
You can use normal WHERE conditions with cached queries.
val stm = sql
.select(user)
.cols1(_.username)
.where(_.age > 25)
.orderBy(_.age.asc)
.cacheWhere1(_.country.cacheEq)
stm.run("CN")
SELECT "username"
FROM "user_profile"
WHERE "age" > 25
AND "country" = 'CN'
ORDER BY "age" ASC
Select join
To do Joins you just put two model instances as arguments and the models will be accessable under a and b
sql
.select(user, customer)
.cols3(t => (
t.a.id,
t.a.username,
t.b.spending
))
.joinOn(_.id, _.userId)
.where(t => Seq(
t.a.age > 25,
t.b.spending > 1000
))
.orderBy(_.b.spending.desc)
.limit(10)
.run
// returns List[Tuple3[Int, String, Int]]
SELECT
"a"."id",
"a"."username",
"b"."spending"
FROM "user_profile" "a"
INNER JOIN "customer" "b"
ON "a"."id" = "b"."user_id"
WHERE "a"."age" > 25
AND "b"."spending" > 1000
ORDER BY "b"."spending" DESC
LIMIT 10
Join result types
You can have a join query return a case class.
case class UserSpending(userId: Int, username: String, spending: Int)
class UserCustomer extends JoinRead[User, Customer] {
val userSpending = read[UserSpending](a.id, a.username, b.spending)
}
val userCustomer = new UserCustomer
implicit val userCustomerConv = userCustomer.convert
sql
.select(userCustomer)
.colsType(_.userSpending)
.joinOn(_.id, _.userId)
.where(_.a.age > 25)
.orderBy(_.b.spending.desc)
.limit(10)
.run
// returns List[UserCustomer]
You can also define results as tuples or a Vector
class UserCustomer extends JoinRead[User, Customer] {
val userSpending = read[UserSpending](a.id, a.username, b.spending)
// colsType(_.userSpending)
val userSpendingTuple = (a.id, a.username, b.spending)
// cols3(_.userSpendingTuple)
val userSpendingVector = Vector(a.id, a.username, b.spending)
// colsVector(_.userSpendingVector)
}
SELECT
"a"."id",
"a"."username",
"b"."spending"
FROM "user_profile" "a"
INNER JOIN "customer" "b"
ON "a"."id" = "b"."user_id"
WHERE "a"."age" > 25
ORDER BY "b"."spending" DESC
LIMIT 10
Join types
The following joins are available. Refer to the section Null values to avoid problems that may come up with joins.
.joinOn(_.id, _.userId) // INNER JOIN
.innerJoinOn(_.id, _.userId) // INNER JOIN
.leftJoinOn(_.id, _.userId) // LEFT JOIN
.leftOuterJoinOn(_.id, _.userId) // LEFT OUTER JOIN
.rightJoinOn(_.id, _.userId) // RIGHT JOIN
.rightOuterJoinOn(_.id, _.userId) // RIGHT OUTER JOIN
.fullOuterJoinOn(_.id, _.userId) // FULL OUTER JOIN
.crossJoin // CROSS JOIN
Basic insert
sql
.insert(user)
.cols2(t => (
t.username,
t.email
))
.run(("bob", "bob@mail.com"))
INSERT INTO "user_profile" ("username", "email") VALUES ('bob', 'bob@mail.com')
Insert many
sql
.insert(user)
.cols2(t => (
t.username,
t.email
))
.runSeq(Seq(
("bob", "bob@mail.com"),
("jane", "jane@mail.com"),
("jack", "jack@mail.com")
))
INSERT INTO "user_profile" ("username", "email")
VALUES ('bob', 'bob@mail.com'),
('jane', 'jane@mail.com'),
('jack', 'jack@mail.com')
Insert case class
case class AddProduct(name: String, price: Float)
class Product extends Model("product") {
val id = column[Int]("id")
val name = column[String]("name")
val price = column[Float]("price")
val add = write[AddProduct](name, price)
}
val product = Model.get[Product]
sql
.insert(product)
.colsType(_.add)
.run(AddProduct("Banana", 12.5))
INSERT INTO "product" ("name", "price") VALUES ('Banana', 12.5)
Cache insert queries
val stm = sql
.insert(user)
.cols2(t => (
t.username,
t.email
))
.cache
stm.run(("bob", "bob@mail.com"))
INSERT INTO "user_profile" ("username", "email") VALUES ('bob', 'bob@mail.com')
Insert key/value
If you need to insert columns that exceed the limits of a tuple, larger than 22.
sql
.insert(user)
.data(t => Seq(
t.username ==> "bob",
t.email ==> "bob@mail.com"
))
.run
INSERT INTO "user_profile" ("username", "email") VALUES ('bob', 'bob@mail.com')
Insert returning
sql
.insert(user)
.cols2(t => (
t.username,
t.email
))
.returning3(t => (
t.id,
t.username,
t.email
))
.runHead(("bob", "bob@mail.com"))
INSERT INTO "user_profile"
("username", "email")
VALUES ('bob', 'bob@mail.com')
RETURNING
"id",
"username",
"email"
Insert on conflict do nothing
You can take advantage of ON CONFLICT DO NOTHING to avoid errors on columns with UNIQUE constraint.
sql
.insert(user)
.cols2(t => (
t.username,
t.email
))
.onConflictDoNothing
.run(("bob", "bob@mail.com"))
INSERT INTO "user_profile"
("username", "email")
VALUES ('bob', 'bob@mail.com')
ON CONFLICT DO NOTHING
Upsert
The updated column has to be one of the columns you intend to insert.
sql
.insert(user)
.cols2(t => (
t.username,
t.email
))
.onConflictOnColumn(_.username)
.doUpdate(_.email)
.run(("bob", "bob@hotmail.com"))
INSERT INTO "user_profile"
("username", "email")
VALUES ('bob', 'bob@mail.com')
ON CONFLICT ("username")
DO UPDATE SET "email" = 'bob@mail.com'
Insert where not exists
If you need to avoid duplication on a column that does not have a unique constraint you can use whereNotExists. Also, if you are makeing multible insert statements concurrently, from a stream for example, you will run into problems using onConflictDoNothing.
sql
.insert(user)
.cols2(t => (
t.username,
t.email
))
.whereNotExists(_.username)
.run(("bob", "bob@mail.com"))
INSERT INTO "user_profile"
("username", "email")
SELECT 'bob', 'bob@mail.com'
WHERE NOT EXISTS (
SELECT 1
FROM "user_profile"
WHERE "username" = 'bob'
)
Insert from select
sql
.insert(newsletter)
.cols1(_.email)
.fromSelect(sql
.select(user)
.cols1(_.email)
.where(_.isActive === true)
)
.run
INSERT INTO "newsletter" ("email")
SELECT "email"
FROM "user_profile"
WHERE "is_active" = true