SimpleDataClassProcessor

This projects provides an annotation processor to create Kotlin like, one-line data classes with the help of Google's AutoValue.

Furthermore it integrates the following autovalue extensions:

How to use

As project currenlty not published to major maven repos please add:

repositories {
   maven { url "https://dl.bintray.com/wgr1984/SimpleDataClasses"}
}

and the following two dependecies:

annotationProcessor "de.wr.simpledataclasses:simpleDataClassesProcessor:0.2"
provided "de.wr.simpledataclasses:libSimpleDataClasses:0.2"
provided "com.google.auto.value:auto-value:1.5.2"
annotationProcessor "com.google.auto.value:auto-value:1.5.2"

(optional in case extensions are used)

annotationProcessor 'com.ryanharter.auto.value:auto-value-parcel:0.2.5'
annotationProcessor 'com.ryanharter.auto.value:auto-value-gson:0.6.0'
provided 'com.ryanharter.auto.value:auto-value-gson-annotations:0.6.0'
compile 'com.google.code.gson:gson:2.8.2'

Motivation

Defining POJOs in java involes quite some boilerplate code

public class SimpleObject {
   private int value1;
   private String value2;
   private List<String> value3;
   
   public int getValue1() {
      return value1;
   }
   
   public String getValue2() {
      return value2;
   }
   
   public List<String> getValue3() {
      return value3;
   }
   
   public void setValue1(int value) {
      value1 = value;
   }
   
   public void setValue2(String value) {
      value2 = value;
   }
   
   public void setValue3(List<String> value) {
      value3 = value;
   }
}

Google's AutoValue provides some ease, as it also provides immutability for free as well as automatically generated toString, equals and hashCode functions.

@AutoValue()
abstract class SimpleObject {

    @com.google.auto.value.AutoValue.Builder()
    abstract static class Builder {

        public abstract Builder value1(int value1);

        public abstract Builder value2(java.lang.String value2);

        public abstract Builder value3(java.util.List<java.lang.String> value3);

        public abstract SimpleObject build();
    }

    static Builder builder() {
        return new AutoValue_SimpleObject.Builder();
    }

    public abstract int value1();
    
    public abstract java.lang.String value2();

    public abstract java.util.List<java.lang.String> value3();
}

But in terms of lines of code it does not even get close to Kotlin's data classes:

data class SimpleObject(var value1: Int, var value2: String, var value3: List<String>)

Challenge

So now the challenge is to combine the relayability of AutoValue with the simplicity of Kotlin like single-line data class definitions. Choosing the easiest way of java code generation - annotion processing - the following idea involed:

Let the annotiation processor generate autoValue classes from one single line

import java.util.List;

import de.wr.libsimpledataclasses.DataClassFactory;

@DataClassFactory
public abstract class SimpleTestDataFactory {
    abstract Void simpleObject(int value1, String value2, List<String> value3);
}

This simple factory will generate one data object per method declared inside itself. Thereby the return type will be ignored, the method name will be used as the name of the data class and each parameter will be transformed into a property

It can then be used like any manually generated autoValue class:

SimpleObject simpleObject = SimpleObject.builder().value1(2).value2("String").value3(Collections.emptyList()).build();
int i = simpleObject.value1();

Furthermore it allows to define additional features such as default values, nullable/nonnull, Parcelable and gson-adapter.

Default Values

@DataClassFactory
public abstract class DefaultValuesTestDataFactory {
    abstract Void defaultValueTestObject(
            @DefaultInt(1) int intV,
            @DefaultLong(2L) long longV,
            @DefaultShort(3) short shortV,
            @DefaultByte(4) byte byteV,
            @DefaultBool(true) boolean boolV,
            @DefaultFloat(5f) float floatV,
            @DefaultDouble(6d) double doubleV,
            @DefaultString("This is a test") String valueS
    );
}

This adds the following default values to the builder function of the generated autoValue class

static Builder builder() {
        return new AutoValue_DefaultValueTestObject.Builder()
                  .intV(1).longV(2).shortV((short) 3).byteV((byte)4)
                  .boolV(true).floatV(5.0f).doubleV(6.0).valueS("This is a test");
    }

Nullability

The user can choose how autovalue handles null values. Per default they are not allowed. To change that behaviour we can just add @Nullable either to the factory itself, infront of each method/class or upfront a parameter that is allowed to contain a null value.

@DataClassFactory
public abstract class NullableTestDataFactory {
    abstract @Nullable Void nullableTestObject0(String s1, @NonNull String s2);
    abstract Void nullableTestObject2(String s1, @Nullable String s2);
    abstract Void nullableTestObject3(String s1, String s2);
    abstract @Nullable Void nullableTestObject4(String s1, String s2);
}

As you can see, using @NonNull allows to exclude certain parameter from allowing to be null in case the factory or method was null annotated.

Parcelable

In order to make a class android.os.Parcelable safe just annotate the corresponding factory or method/class using @Parcable. E.g.:

@Parcelable abstract Void dataObject3(byte by, double d, float f, int i, short s, boolean b, long l, Number number);

Inside the generated code auto-value-parcel will used to automatically generate the needed classes and methodes to implement Parcable Serialization.

GSON

Talking about serialization: The Simple Data Class Processor supports json serialization, too, currently via auto-value-gson. Therefore similar to the Parsable annotation a @Gson annotation is provided.

@DataClassFactory
@Nullable
public abstract class DataFactory {
   @Gson abstract Void dataObject1(@DefaultString(test) String val1, @DefaultInt(2) int number, @DefaultInt(3) int number2);
}

This will also generate the needed TypeAdapterFactory in our case DataFactoryTypeAdapterFactory, which is needed to provide easy non reflection based (de-)serialization:

Gson gson = new GsonBuilder().registerTypeAdapterFactory(DataFactoryTypeAdapterFactory.create()).create();
String json = gson.toJson(o3);
DataObject3 o3New = gson.fromJson(json, DataObject3.class);

Of course it possible to name attributes differently inside the consumed json and the generated data class:

abstract @Gson Void simpleObjectNamed(
            @Named("value_1") int value1,
            @Named("value_2") String value2,
            @Named("value_3") List<String> value3);

This will add internally the well known @SerializedName annotation. In case the default values provided should also be applied inside the gson adapter add the following config inside the build gradle of your project

android {
  // ...
  defaultConfig {
    // ...
    javaCompileOptions {
      annotationProcessorOptions {
        arguments = ['autovaluegson.mutableAdaptersWithDefaultSetters': 'true']
      }
    }
  }
}

and use @Gson(true) instead of the pure annotation.

@Gson(true) abstract Void dataObject1(@DefaultString(test) String val1, int number, @DefaultInt(3) int number2);

Todo:

added named annotation (Gson)
Publish to jcenter
Provide samples and doc
support auto-value-gson default values
support auto-value-moshi
remove data class factories from deployed classes

wgr1984/SimpleDataClassProcessor