Slick (Scala language-integrated connection kit)是scala的一个FRM(Functional Relational Mapper),即函数式的关系数据库编程工具库。Slick的主要目的是使关系数据库能更容易、更自然的融入函数式编程模式,它可以使使用者像对待scala集合一样来处理关系数据库表。也就是说可以用scala集合的那些丰富的操作函数来处理库表数据。Slick?把数据库编程融入到scala编程中,编程人员可以不需要编写SQL代码。我把Slick官方网站上Slick3.1.1文档的Slick介绍章节中的一些描述和例子拿过来帮助介绍Slick的功能。下面是Slick数据库和类对象关系对应的一个例子:
1 import slick.driver.H2Driver.api._ 2 object slickIntro { 3 case class Coffee(id: Int, 4 name: String, 5 supID: Int = 0, 6 price: Double , 7 sales: Int = 0, 8 total: Int = 0) 9 10 class Coffees(tag: Tag) extends Table[Coffee](tag, "COFFEES") { 11 def id = column[Int]("COF_ID", O.PrimaryKey, O.AutoInc) 12 def name = column[String]("COF_NAME") 13 def supID = column[Int]("SUP_ID") 14 def price = column[Double]("PRICE") 15 def sales = column[Int]("SALES", O.Default(0)) 16 def total = column[Int]("TOTAL", O.Default(0)) 17 def * = (id, name, supID, price, sales, total) <> (Coffee.tupled, Coffee.unapply) 18 } 19 val coffees = TableQuery[Coffees] 20 //> coffees : slick.lifted.TableQuery[worksheets.slickIntro.Coffees] = Rep(TableExpansion) 21 }
我们把数据库中的COFFEES表与Coffees类做了对应,包括字段、索引、默认值、返回结果集字段等。现在这个coffees就是scala里的一个对象,但它代表了数据库表。现在我们可以用scala语言来编写数据存取程序了:
1 val limit = 10.0 //> limit : Double = 10.0 2 // // 写Query时就像下面这样: 3 ( for( c <- coffees; if c.price < limit ) yield c.name ).result 4 //> res0: slick.driver.H2Driver.StreamingDriverAction[Seq[String],String,slick.dbio.Effect.Read] = slick.dri[email protected]46cdf8bd 5 // 相当于 SQL: select COF_NAME from COFFEES where PRICE < 10.0
或者下面这些不同的Query:
1 // 返回"name"字段的Query 2 // 相当于 SQL: select NAME from COFFEES 3 coffees.map(_.name) 4 //> res1: slick.lifted.Query[slick.lifted.Rep[String],String,Seq] = Rep(Bind) 5 // 选择 price < 10.0 的所有记录Query 6 // 相当于 SQL: select * from COFFEES where PRICE < 10.0 7 coffees.filter(_.price < 10.0) 8 //> res2: slick.lifted.Query[worksheets.slickIntro.Coffees,worksheets.slickIntro.Coffees#TableElementType,Seq] = Rep(Filter @1946988038)
我们可以这样表述:coffees.map(_.name) >>> coffees.map{row=>row.name}, coffees.filter(_.price<10.0) >>> coffees.filter{row=>row.price<10.0),都是函数式集合操作语法。
Slick把Query编写与scala语言集成,这使编程人员可以用熟悉惯用的scala来表述SQL Query,直接的好处是scalac在编译时就能够发现Query错误:
1 //coffees.map(_.prices) 2 //编译错误:value prices is not a member of worksheets.slickIntro.Coffees
当然,嵌入scala的Query还可以获得运行效率的提升,因为在编译时可以进行前期优化。
最新版本的Slick最大的特点是采用了Functional I/O技术,从而实现了安全的多线程无阻碍I/O操作。再就是实现了Query的函数组合(functional composition),使Query编程更贴近函数式编程模式。通过函数组合实现代码重复利用,提高编程工作效率。具体实现方式是利用freemonad(DBIOAction类型就是个freemonad)的延迟运算模式,将DBIOAction的编程和实际运算分离,在DBIOAction编程过程中不会产生副作用(side-effect),从而实现纯代码的函数组合。我们来看看Query函数组合和DBIOAction运算示范:
1 import scala.concurrent.ExecutionContext.Implicits.global 2 val qDelete = coffees.filter(_.price > 0.0).delete 3 //> qDelete : slick.driver.H2Driver.DriverAction[Int,slick.dbio.NoStream,slick.dbio.Effect.Write] ... 4 val qAdd1 = (coffees returning coffees.map(_.id)) += Coffee(name="Columbia",price=128.0) 5 //> qAdd1 : slick.profile.FixedSqlAction[Int,slick.dbio.NoStream,slick.dbio.Effect.Write]... 6 val qAdd2 = (coffees returning coffees.map(_.id)) += Coffee(name="Blue Mountain",price=828.0) 7 //> qAdd2 : slick.profile.FixedSqlAction[Int,slick.dbio.NoStream,slick.dbio.Effect.Write]... 8 def getNameAndPrice(n: Int) = coffees.filter(_.id === n) 9 .map(r => (r.name,r.price)).result.head 10 //> getNameAndPrice: (n: Int)slick.profile.SqlAction[(String, Double),slick.dbio.NoStream,slick.dbio.Effect.Read] 11 12 val actions = for { 13 _ <- coffees.schema.create 14 _ <- qDelete 15 c1 <- qAdd1 16 c2 <- qAdd2 17 (n1,p1) <- getNameAndPrice(c1) 18 (n2,p2) <- getNameAndPrice(c2) 19 } yield (n1,p1,n2,p2) 20 //> actions : slick.dbio.DBIOAction[(String, Double, String, Double),..
我们可以放心的来组合这个actions,不用担心有任何副作用。actions的类型是:DBAction[String,Double,String,Double]。我们必须用Database.Run来真正开始运算,产生副作用:
1 import java.sql.SQLException 2 import scala.concurrent.Await 3 import scala.concurrent.duration._ 4 val db = Database.forURL("jdbc:h2:mem:demo", driver="org.h2.Driver") 5 //> db : slick.driver.H2Driver.backend.DatabaseDef = [email protected] 6 Await.result( 7 db.run(actions.transactionally).map { res => 8 println(s"Add coffee: ${res._1},${res._2} and ${res._3},${res._4}") 9 }.recover { 10 case e: SQLException => println("Caught exception: " + e.getMessage) 11 }, Duration.Inf) //> Add coffee: Columbia,128.0 and Blue Mountain,828.0
在特殊的情况下我们也可以引用纯SQL语句:Slick提供了Plain SQL API, 如下:
1 val limit = 10.0 2 sql"select COF_NAME from COFFEES where PRICE < $limit".as[String] 3 // 用$来绑定变量: // select COF_NAME from COFFEES where PRICE < ?
下面是这篇讨论的示范代码:
1 package worksheets 2 import slick.driver.H2Driver.api._ 3 object slickIntro { 4 case class Coffee(id: Int = 0, 5 name: String, 6 supID: Int = 0, 7 price: Double, 8 sales: Int = 0, 9 total: Int = 0) 10 11 class Coffees(tag: Tag) extends Table[Coffee](tag, "COFFEES") { 12 def id = column[Int]("COF_ID", O.PrimaryKey, O.AutoInc) 13 def name = column[String]("COF_NAME") 14 def supID = column[Int]("SUP_ID") 15 def price = column[Double]("PRICE") 16 def sales = column[Int]("SALES", O.Default(0)) 17 def total = column[Int]("TOTAL", O.Default(0)) 18 def * = (id, name, supID, price, sales, total) <> (Coffee.tupled, Coffee.unapply) 19 } 20 val coffees = TableQuery[Coffees] 21 22 val limit = 10.0 23 // // 写Query时就像下面这样: 24 ( for( c <- coffees; if c.price < limit ) yield c.name ).result 25 // 相当于 SQL: select COF_NAME from COFFEES where PRICE < 10.0 26 27 // 返回"name"字段的Query 28 // 相当于 SQL: select NAME from COFFEES 29 coffees.map(_.name) 30 // 选择 price < 10.0 的所有记录Query 31 // 相当于 SQL: select * from COFFEES where PRICE < 10.0 32 coffees.filter(_.price < 10.0) 33 //coffees.map(_.prices) 34 //编译错误:value prices is not a member of worksheets.slickIntro.Coffees 35 36 37 import scala.concurrent.ExecutionContext.Implicits.global 38 val qDelete = coffees.filter(_.price > 0.0).delete 39 val qAdd1 = (coffees returning coffees.map(_.id)) += Coffee(name="Columbia",price=128.0) 40 val qAdd2 = (coffees returning coffees.map(_.id)) += Coffee(name="Blue Mountain",price=828.0) 41 def getNameAndPrice(n: Int) = coffees.filter(_.id === n) 42 .map(r => (r.name,r.price)).result.head 43 44 val actions = for { 45 _ <- coffees.schema.create 46 _ <- qDelete 47 c1 <- qAdd1 48 c2 <- qAdd2 49 (n1,p1) <- getNameAndPrice(c1) 50 (n2,p2) <- getNameAndPrice(c2) 51 } yield (n1,p1,n2,p2) 52 import java.sql.SQLException 53 import scala.concurrent.Await 54 import scala.concurrent.duration._ 55 val db = Database.forURL("jdbc:h2:mem:demo", driver="org.h2.Driver") 56 Await.result( 57 db.run(actions.transactionally).map { res => 58 println(s"Add coffee: ${res._1},${res._2} and ${res._3},${res._4}") 59 }.recover { 60 case e: SQLException => println("Caught exception: " + e.getMessage) 61 }, Duration.Inf) 62 63 }