scala-db-codegen

A sbt-plugin and mill-plugin to generate boilerplate code from a database schema. Tested with SQLite and Postgresql. Should work for all databases supported by jdbc.

DbSchema + Template => generate scala code

The plugin can be configured to crawl a database schema to extract all tables/columns/enums. It matches the occurring jdbc/sql types to scala types. You can provide a scalate template to generate scala code out of this information like this:

<%@ val schema: dbcodegen.DataSchema %>

package kicks.db.${schema.name}

#for (table <- schema.tables)
case class ${table.scalaName}(
  #for (column <- table.columns)
  ${column.scalaName}: ${column.scalaType},
  #end
)
#end

Usage

You can use the codegen project as a library or use the sbt and mill plugin. For ease of use, jdbc drivers as well as schemacrawler plugins for postgresql, mysql and sqlite are included.

sbt

In project/plugins.sbt:

addSbtPlugin("com.github.cornerman" % "sbt-db-codegen" % "0.4.0")

In build.sbt:

lazy val db = project
  .enablePlugins(dbcodegen.plugin.DbCodegenPlugin)
  .settings(
    // The template file for the code generator
    dbcodegenTemplateFiles := Seq(file("schema.scala.ssp"))
    // The jdbc URL for the database
    dbcodegenJdbcUrl := "jdbc:...",

    // Optional database username
    // dbcodegenUsername          := None,
    // Optional database password
    // dbcodegenPassword          := None,
    // Map sql types to java/scala types
    // dbcodegenTypeMapping       := (sqlType: SQLType, scalaType: Option[String]) => scalaType,
    // Filter which schema and table should be processed
    // dbcodegenSchemaTableFilter := (schema: String, table: String) => true
    // Whether to run scalafmt on the generated code
    // dbcodegenScalafmt := true
    // Setup task to be executed before the code generation runs against the database
    // dbcodegenSetupTask         := { _ => () },
  )

Setup database before codegen

An example for using the dbcodegenSetupTask to setup an sqlite database with a schema.sql file before the code generation runs:

dbcodegenJdbcUrl := "jdbc:sqlite:file::memory:?cache=shared",
dbcodegenSetupTask := { db =>
  db.executeSqlFile(file("./schema.sql"))
}

The functions executeSql and executeSqlFile are provided for these kind of use-cases and use the provided jdbcUrl, username, and password.

mill

In build.sc:

import mill._, scalalib._
import $ivy.`com.github.cornerman::mill-db-codegen:0.4.0`, dbcodegen.plugin._

object backend extends ScalaModule with DbCodegenModule {
  // sources to trigger reloads when a file changes
  def dbTemplateFile         = T.source(os.pwd / "schema.scala.ssp")
  def dbSchemaFile           = T.source(os.pwd / "schema.sql")

  // The template file for the code generator
  def dbcodegenTemplateFiles = T { Seq(dbTemplateFile()) }
  // The jdbc URL for the database
  def dbcodegenJdbcUrl       = "jdbc:sqlite:file::memory:?cache=shared"
  // Setup task to be executed before the code generation runs against the database
  def dbcodegenSetupTask = T.task { (db: Db) =>
    db.executeSqlFile(dbSchemaFile())
  }

  // Optional database username
  // def dbcodegenUsername = None
  // Optional database password
  // def dbcodegenPassword = None
  // Map sql types to java/scala types
  // def dbcodegenTypeMapping = (sqlType: SQLType, scalaType: Option[String]) => scalaType
  // Filter which schema and table should be processed
  // def dbcodegenSchemaTableFilter = (schema: String, table: String) => true
  // Whether to run scalafmt on the generated code
  // def dbcodegenScalafmt = true
}

Template Examples

Template can be configured by setting dbcodegenTemplateFiles.

We are using scalate for templates, so you can use anything that is supported there (e.g. mustache or ssp) - the converter will be picked according to the file extension of the provided template file. Check the scalate user guide for more details.

The template is called on each database schema, and is passed an instance of dbcodegen.DataSchema (variable name schema) which contains all the extracted information. You can see the declaration in the first line of each ssp template.

Simple

case-classes.scala.ssp:

<%@ val schema: dbcodegen.DataSchema %>

package kicks.db.${schema.name}

#for (enum <- schema.enums)
type ${enum.scalaName} = ${enum.values.map(v => "\"" + v.name + "\"").mkString(" | ")}
#end

#for (table <- schema.tables)

case class ${table.scalaName}(
  #for (column <- table.columns)
  ${column.scalaName}: ${column.scalaType},
  #end
)

#end

with scala 3 enums

#for (enum <- schema.enums)

enum ${enum.scalaName}(val sqlValue: String) {
  #for (enumValue <- enum.values)
  case ${enumValue.scalaName} extends ${enum.scalaName}("${enumValue.name}")
  #end
}
object ${enum.scalaName} {
  def bySqlValue(searchValue: String): Option[${enum.scalaName}] = values.find(_.sqlValue == searchValue)
}

#end

Library: quill

quill-case-classes.scala.ssp:

<%@ val schema: dbcodegen.DataSchema %>

package kicks.db.${schema.scalaName}

import io.getquill.*

#for (enum <- schema.enums)
type ${enum.scalaName} = ${enum.values.map(v => "\"" + v.name + "\"").mkString(" | ")}
#end

#for (table <- schema.tables)

case class ${table.scalaName}(
  #for (column <- table.columns)
  ${column.scalaName}: ${column.scalaType},
  #end
)
object ${table.scalaName} {
  inline def query = querySchema[${table.scalaName}](
    "${table.name}",
    #for (column <- table.columns)
    _.${column.scalaName} -> "${column.name}",
    #end
  )
}

#end

Library: magnum

magnum-case-classes.scala.ssp:

<%@ val schema: dbcodegen.DataSchema %>

package kicks.db.${schema.scalaName}

import com.augustnagro.magnum.*

#for (enum <- schema.enums)
type ${enum.scalaName} = ${enum.values.map(v => "\"" + v.name + "\"").mkString(" | ")}
#end

#for (table <- schema.tables)

@Table(SqliteDbType)
case class ${table.scalaName}(
  #for (column <- table.columns)
  #if (column.db.isPartOfPrimaryKey)
  @Id
  #end
  @SqlName("${column.name}")
  ${column.scalaName}: ${column.scalaType},
  #end
) derives DbCodec
object ${table.scalaName} {
  #{ val primaryKeyColumns = table.columns.filter(_.db.isPartOfPrimaryKey)}#
  type Id = ${if (primaryKeyColumns.isEmpty) "Null" else primaryKeyColumns.map(_.scalaType).mkString("(", ", ", ")")}

  #if (!table.isView)
  case class Creator(
    #for (column <- table.columns if !column.db.isGenerated && !column.db.hasDefaultValue && !column.db.isAutoIncremented)
    ${column.scalaName}: ${column.scalaType},
    #end
  )
  #end
}

#if (table.isView)
val ${table.scalaName}Repo = ImmutableRepo[${table.scalaName}, ${table.scalaName}.Id]
#else
val ${table.scalaName}Repo = Repo[${table.scalaName}.Creator, ${table.scalaName}, ${table.scalaName}.Id]
#end

#end

cornerman / scala-db-codegen 0.5.2