Annotation Type AutoClone


@Documented
@Retention(RUNTIME)
@Target(TYPE)
public @interface AutoClone
Class annotation used to assist in the creation of Cloneable classes. The @AutoClone annotation instructs the compiler to execute an AST transformation which adds a public clone() method and adds Cloneable to the list of interfaces which the class implements.

Because the JVM doesn't have a one-size fits all cloning strategy, several customizations exist for the cloning implementation. By default, the clone() method will call super.clone() before calling clone() on each Cloneable property of the class.

Example usage:

 import groovy.transform.AutoClone
 @AutoClone
 class Person {
   String first, last
   List favItems
   Date since
 }
 
Which will create a class equivalent to the following:
 class Person implements Cloneable {
   ...
   public Person clone() throws CloneNotSupportedException {
     Person result = (Person) super.clone()
     result.favItems = favItems instanceof Cloneable ? (List) favItems.clone() : favItems
     result.since = (Date) since.clone()
     return result
   }
   ...
 }
 
Which can be used as follows:
 def p = new Person(first:'John', last:'Smith', favItems:['ipod', 'shiraz'], since:new Date())
 def p2 = p.clone()

 assert p instanceof Cloneable
 assert p.favItems instanceof Cloneable
 assert p.since instanceof Cloneable
 assert !(p.first instanceof Cloneable)

 assert !p.is(p2)
 assert !p.favItems.is(p2.favItems)
 assert !p.since.is(p2.since)
 assert p.first.is(p2.first)
 
In the above example, super.clone() is called which in this case calls clone() from java.lang.Object. This does a bit-wise copy of all the properties (references and primitive values). Properties like first has type String which is not Cloneable so it is left as the bit-wise copy. Both Date and ArrayList are Cloneable so the clone() method on each of those properties will be called. For the list, a shallow copy is made during its clone() method.

If your classes require deep cloning, it is up to you to provide the appropriate deep cloning logic in the respective clone() method for your class.

If one of your properties contains an object that doesn't support cloning or attempts deep copying of a data structure containing an object that doesn't support cloning, then a CloneNotSupportedException may occur at runtime.

Another popular cloning strategy is known as the copy constructor pattern. If any of your fields are final and Cloneable you should set style=COPY_CONSTRUCTOR which will then use the copy constructor pattern. Here is an example making use of the copy constructor pattern:

 import groovy.transform.AutoClone
 import static groovy.transform.AutoCloneStyle.*
 @AutoClone(style=COPY_CONSTRUCTOR)
 class Person {
   final String first, last
   final Date birthday
 }
 @AutoClone(style=COPY_CONSTRUCTOR)
 class Customer extends Person {
   final int numPurchases
   final List favItems
 }
 
Which will create classes equivalent to the following:
 class Person implements Cloneable {
   ...
   protected Person(Person other) throws CloneNotSupportedException {
     first = other.first
     last = other.last
     birthday = (Date) other.birthday.clone()
   }
   public Person clone() throws CloneNotSupportedException {
     return new Person(this)
   }
   ...
 }
 class Customer extends Person {
   ...
   protected Customer(Customer other) throws CloneNotSupportedException {
     super(other)
     numPurchases = other.numPurchases
     favItems = other.favItems instanceof Cloneable ? (List) other.favItems.clone() : other.favItems
   }
   public Customer clone() throws CloneNotSupportedException {
     return new Customer(this)
   }
   ...
 }
 
If you use this style on a child class, the parent class must also have a copy constructor (created using this annotation or by hand). This approach can be slightly slower than the traditional cloning approach but the Cloneable fields of your class can be final. When using the copy constructor style, you can provide your own custom constructor by hand if you wish. If you do so, it is up to you to correctly copy, clone or deep clone the properties of your class.

As a variation of the last two styles, if you set style=SIMPLE then the no-arg constructor will be called followed by setting the individual properties (and/or fields) calling clone() if the property/field implements Cloneable. Here is an example:

 import groovy.transform.AutoClone
 import static groovy.transform.AutoCloneStyle.*
 @AutoClone(style=SIMPLE)
 class Person {
   final String first, last
   final Date birthday
 }
 @AutoClone(style=SIMPLE)
 class Customer {
   final List favItems
 }
 
Which will create classes equivalent to the following:
 class Person implements Cloneable {
   ...
   public Person clone() throws CloneNotSupportedException {
     def result = new Person()
     copyOrCloneMembers(result)
     return result
   }
   protected void copyOrCloneMembers(Person other) {
     other.first = first
     other.last = last
     other.birthday = (Date) birthday.clone()
   }
   ...
 }
 class Customer extends Person {
   ...
   public Customer clone() throws CloneNotSupportedException {
     def result = new Customer()
     copyOrCloneMembers(result)
     return result
   }
   protected void copyOrCloneMembers(Customer other) {
     super.copyOrCloneMembers(other)
     other.favItems = favItems instanceof Cloneable ? (List) favItems.clone() : favItems
   }
   ...
 }
 
You would typically use this style only for base classes where you didn't want the normal Object clone() method to be called and you would typically need to use the SIMPLE style for any child classes.

As a final example, if your class already implements the Serializable or Externalizable interface, you can choose the following cloning style:

 @AutoClone(style=SERIALIZATION)
 class Person implements Serializable {
   String first, last
   Date birthday
 }
 
which outputs a class with the following form:
 class Person implements Cloneable, Serializable {
   ...
   Person clone() throws CloneNotSupportedException {
     def baos = new ByteArrayOutputStream()
     baos.withObjectOutputStream{ it.writeObject(this) }
     def bais = new ByteArrayInputStream(baos.toByteArray())
     bais.withObjectInputStream(getClass().classLoader){ (Person) it.readObject() }
   }
   ...
 }
 
This will output an error if your class doesn't implement one of Serializable or Externalizable, will typically be significantly slower than the other approaches, also doesn't allow fields to be final, will take up more memory as even immutable classes like String will be cloned but does have the advantage that it performs deep cloning automatically.

Further references on cloning:

Since:
1.8.0
See Also:
AutoCloneStyle, ExternalizeMethods
  • Optional Element Summary

    Optional Elements
    Modifier and Type Optional Element Description
    String[] excludes
    Comma separated list of property (and/or field) names to exclude from cloning.
    boolean includeFields
    Include fields as well as properties when cloning.
    AutoCloneStyle style
    Style to use when cloning.
  • Element Details

    • excludes

      String[] excludes
      Comma separated list of property (and/or field) names to exclude from cloning. For convenience, a String with comma separated names can be used in addition to an array (using Groovy's literal list notation) of String values.

      NOTE: When using the CLONE style, property (and/or field) copying might occur as part of calling super.clone() which will ignore this list. You can then use this list to streamline the provided clone() implementation by selecting which Cloneable properties (and/or fields) will have a subsequent call to their clone() method. If you have immutable properties (and/or fields) this can be useful as the extra clone() will not be necessary and cloning will be more efficient.

      NOTE: This doesn't affect property (and/or field) copying that might occur as part of serialization when using the SERIALIZATION style, i.e. this flag is ignored; instead adjust your serialization code to include or exclude the desired properties (and/or fields) which should carry over during cloning.

      Default:
      {}
    • includeFields

      boolean includeFields
      Include fields as well as properties when cloning.

      NOTE: When using the CLONE style, field copying might occur as part of calling super.clone() and might be all you require; if you turn on this flag, the provided clone() implementation will also subsequently call clone() for each Cloneable field which can be useful if you have mutable fields.

      NOTE: This doesn't affect field copying that might occur as part of serialization when using the SERIALIZATION style, i.e. this flag is ignored; instead adjust your serialization code to include or exclude your fields.

      Default:
      false
    • style

      Style to use when cloning.
      Default:
      CLONE