gradle-avrohugger-plugin

Gradle plugin for generating scala case classes from Apache Avro schemas, datafiles and protocols. It is based on avrohugger scala library.

Requirements

Plugin requires Gradle in version 7.0 or higher.

Use plugin in version 0.8.1 for compatibility down to Gradle 5.1
Use plugin in version 0.2.5 for compatibility down to Gradle 4.3

Plugin itself depends on Scala 2.13.

Use plugin in version 0.11.0 if you are using other plugins depending on Scala 2.12

Usage

For minimal usecase it is enough to add following to your build.gradle file:

plugins {
    id 'com.zlad.gradle.avrohugger' version '1.1.0'
}

But usually scala plugin and scala-library dependency should be added to your build.gradle file to enable compilation of generated scala sources:

plugins {
    id 'scala'
    id 'com.zlad.gradle.avrohugger' version '1.1.0'
}

repositories {
   mavenCentral()
}

dependencies {
    implementation 'org.scala-lang:scala-library:2.13.16'
}

Then by default scala classes generation will be triggered during build before compileScala task. It will look into directory src/main/avro for all files ending with .avsc, .avdl, .avro or .avpr. Classes will be generated under ${buildDir}/generated-src/avro and they will be added to main scala source set.

You can always call generateAvroScala task to perform classes generation only

Configuration

You can change default configuration in avrohugger block like in following example:

avrohugger {
    sourceDirectories = files('src-avro')
    destinationDirectory = file('src-scala')
    namespaceMapping = [ 'com.example' : 'com.zlad' ]
    typeMapping {
        protocolType = ScalaADT
    }
    sourceFormat = Standard
}

Description of configuration values

Name	Default	Description
sourceDirectories	src/main/avro	Directories (can be multiple) used to look for .avsc, .avdl, .avro or .avpr files as an input for classes generation
destinationDirectory	${buildDir}/generated-src/avro	Directory where classes will be generated
namespaceMapping	empty map	Map where key is name of namespace and value is it's replacement which will be used as package for resulting classes
typeMapping	default type mapping	Defines how avro types are represented as scala types

Type mapping values

Name (avro type)	Default (scala type)	PossibleValues
intType	ScalaInt	ScalaInt, ScalaLong, ScalaFloat, ScalaDouble
longType	ScalaLong	ScalaInt, ScalaLong, ScalaFloat, ScalaDouble
floatType	ScalaFloat	ScalaInt, ScalaLong, ScalaFloat, ScalaDouble
doubleType	ScalaDouble	ScalaInt, ScalaLong, ScalaFloat, ScalaDouble
recordType	ScalaCaseClass	ScalaCaseClass, ScalaCaseClassWithSchema
enumType	ScalaEnumeration (JavaEnum for SpecificRecord)	ScalaEnumeration, JavaEnum, ScalaCaseObjectEnum, EnumAsScalaString
unionType	OptionEitherShapelessCoproduct	OptionEitherShapelessCoproduct, OptionalShapelessCoproduct, OptionShapelessCoproduct
arrayType	ScalaSeq	ScalaSeq, ScalaArray, ScalaList, ScalaVector
protocolType	NoTypeGenerated	NoTypeGenerated, ScalaADT
decimalType	ScalaBigDecimal	ScalaBigDecimal, ScalaBigDecimalWithPrecision
dateType	JavaTimeLocalDate	JavaTimeLocalDate, JavaSqlDate
timestampMillisType	JavaTimeInstant	JavaTimeInstant, JavaSqlTimestamp
uuidType	JavaUuid	JavaUuid

See avrohugger library for all details about types.

To set rounding mode for decimal types append underscore followed by rounding mode e.g. ScalaBigDecimal_UP or ScalaBigDecimalWithPrecision_DOWN, where rounding mode could be one of UP, DOWN, CEILING, FLOOR, HALF_UP, HALF_DOWN, HALF_EVEN and UNNECESSARY.

Source formats

1. Standard - vanilla case classes

2. SpecificRecord - case classes implementing SpecificRecordBase with mutable fields

   There is need additional dependency on Apache Avro to compile generated classes with SpecificRecord source format.

    dependencies {
       // ... other dependencies here 
       implementation 'org.apache.avro:avro:1.11.4'
    }

See avrohugger library for all details about different formats

How to change default source set

If you do not want generated scala classes be a part of main source set you can change it as in following example to be a part of test source set:

sourceSets {
    test {
        scala {
            srcDir avrohugger.destinationDirectory
        }
    }
}

How to use schema files from dependencies

If you have schema files provided as part of one or more jar files you can include them as dependencies and configure avrohugger to use them:

1. Define new configuration

   configurations {
       avroSchema
   }

2. Declare dependencies

   dependencies {
       avroSchema 'sample:schema-one:0.1.0'
       avroSchema 'sample:schema-two:0.1.0'
   }

3. Configure avrohugger

   avrohugger {
       sourceDirectories {
           from configurations.avroSchema.files.collect { zipTree(it) }
       }
   }

How to use with Kotlin DSL

See following example of configuration made with Kotlin DSL:

import com.zlad.gradle.avrohugger.AvrohuggerExtension

plugins {
    id ("scala") 
    id ("com.zlad.gradle.avrohugger") version "1.1.0"
}

repositories {
   mavenCentral()
}

dependencies {
    implementation ("org.scala-lang:scala-library:2.13.16")
}

avrohugger {
    sourceDirectories.from("src-avro")

    destinationDirectory.set(file("src-scala"))

    namespaceMapping.set(mapOf(
        "example.idl" to "com.zlad"
    ))

    typeMapping {
        protocolType = AvrohuggerExtension.ScalaADT
        enumType = AvrohuggerExtension.ScalaCaseObjectEnum
    }

    sourceFormat.set(AvrohuggerExtension.Standard)
}

Scala 2.10 support

There is a restriction about maximum fields count for case classes in scala 2.10 and older. If some of your records has more then 22 fields and scala 2.10 is used simple serializable class will be generated instead of case class.

Plugin will get your scala version from scala-library version in compile dependencies. If it is not able to found the dependency it will act as the version is 2.13.

Build and Test

Plugin can be built and tested using:

./gradlew clean build

Credits

Core generation logic is implemented by Julian Peeters in scala library called avrohugger. Plugin was inspired by it's sbt version sbt-avrohugger.

Thanks to all contributors.