groovy.transform

Annotation Type EqualsAndHashCode

@Documented
 @Retention(value=RUNTIME)
 @Target(value=TYPE)
public @interface EqualsAndHashCode

Class annotation used to assist in creating appropriate equals() and hashCode() methods.

It allows you to write classes in this shortened form:

 import groovy.transform.EqualsAndHashCode
 @EqualsAndHashCode
 class Person {
     String first, last
     int age
 }
 def p1 = new Person(first:'John', last:'Smith', age:21)
 def p2 = new Person(first:'John', last:'Smith', age:21)
 assert p1 == p2
 def map = [:]
 map[p1] = 45
 assert map[p2] == 45
 

The @EqualsAndHashCode annotation instructs the compiler to execute an AST transformation which adds the necessary equals and hashCode methods to the class.

The hashCode() method is calculated using Groovy's HashCodeHelper class which implements an algorithm similar to the one outlined in the book Effective Java.

The equals() method compares the values of the individual properties (and optionally fields) of the class. It can also optionally call equals on the super class. Two different equals method implementations are supported both of which support the equals contract outlined in the javadoc for java.lang.Object

To illustrate the 'canEqual' implementation style (see http://www.artima.com/lejava/articles/equality.html for further details), consider this class:

  @EqualsAndHashCode
 class IntPair {
     int x, y
 }
 

The generated equals and canEqual methods will be something like below:

 public boolean equals(java.lang.Object other)
     if (other == null) return false
     if (this.is(other)) return true
     if (!(other instanceof IntPair)) return false
     if (!other.canEqual(this)) return false
     if (x != other.x) return false
     if (y != other.y) return false
     return true
 }

 public boolean canEqual(java.lang.Object other) {
     return other instanceof IntPair
 }
 

If no further options are specified, this is the default style for @Canonical and @EqualsAndHashCode annotated classes. The advantage of this style is that it allows inheritance to be used in limited cases where its purpose is for overriding implementation details rather than creating a derived type with different behavior. This is useful when using JPA Proxies for example or as shown in the following examples:

 import groovy.transform.*
 @Canonical class IntPair { int x, y }
 def p1 = new IntPair(1, 2)

 // overridden getter but deemed an IntPair as far as domain is concerned
 def p2 = new IntPair(1, 1) { int getY() { 2 } }

 // additional helper method added through inheritance but
 // deemed an IntPair as far as our domain is concerned
 @InheritConstructors class IntPairWithSum extends IntPair {
     def sum() { x + y }
 }

 def p3 = new IntPairWithSum(1, 2)

 assert p1 == p2 && p2 == p1
 assert p1 == p3 && p3 == p1
 assert p3 == p2 && p2 == p3
 

Note that if you create any domain classes which don't have exactly the same contract as IntPair then you should provide an appropriate equals and canEqual method. The easiest way to achieve this would be to use the @Canonical or @EqualsAndHashCode annotations as shown below:

  @EqualsAndHashCode
  @TupleConstructor(includeSuperProperties=true)
 class IntTriple extends IntPair { int z }
 def t1 = new IntTriple(1, 2, 3)
 assert p1 != t1  && p2 != t1  && t1 != p3
 

The alternative supported style regards any kind of inheritance as creation of a new type and is illustrated in the following example:

  @EqualsAndHashCode(useCanEqual=false)
 class IntPair {
     int x, y
 }
 

The generated equals method will be something like below:

 public boolean equals(java.lang.Object other)
     if (other == null) return false
     if (this.is(other)) return true
     if (IntPair != other.getClass()) return false
     if (x != other.x) return false
     if (y != other.y) return false
     return true
 }
 

This style is appropriate for final classes (where inheritance is not allowed) which have only java.lang.Object as a super class. Most @Immutable classes fall in to this category. For such classes, there is no need to introduce the canEqual() method.

Note that if you explicitly set useCanEqual=false for child nodes in a class hierarchy but have it true for parent nodes and you also have callSuper=true in the child, then your generated equals methods will not strictly follow the equals contract.

Note that when used in the recommended fashion, the two implementations supported adhere to the equals contract. You can provide your own equivalence relationships if you need, e.g. for comparing instances of the IntPair and IntTriple classes discussed earlier, you could provide the following method in IntPair:

 boolean hasEqualXY(other) { other.x == getX()  && other.y == getY() }
 

Then for the objects defined earlier, the following would be true:

 assert p1.hasEqualXY(t1)  && t1.hasEqualXY(p1)
 assert p2.hasEqualXY(t1)  && t1.hasEqualXY(p2)
 assert p3.hasEqualXY(t1)  && t1.hasEqualXY(p3)
 

There is also support for including or excluding fields/properties by name when constructing the equals and hashCode methods as shown here:

 import groovy.transform.*
 @EqualsAndHashCode(excludes="z")
 @TupleConstructor
 public class Point2D {
     int x, y, z
 }

 assert  new Point2D(1, 1, 1).equals(new Point2D(1, 1, 2))
 assert !new Point2D(1, 1, 1).equals(new Point2D(2, 1, 1))

 @EqualsAndHashCode(excludes=["y", "z"])
 @TupleConstructor
 public class Point1D {
     int x, y, z
 }

 assert  new Point1D(1, 1, 1).equals(new Point1D(1, 1, 2))
 assert  new Point1D(1, 1, 1).equals(new Point1D(1, 2, 1))
 assert !new Point1D(1, 1, 1).equals(new Point1D(2, 1, 1))
 

Note: @EqualsAndHashCode is a transform to help reduce boilerplate in the common cases. It provides a useful implementation of equals() and hashCode() but not necessarily the most appropriate or efficient one for every use case. You should write custom versions if your scenario demands it. In particular, if you have mutually-referential classes, the implementations provided may not be suitable and may recurse infinitely (leading to a StackOverflowError). In such cases, you need to remove the infinite loop from your data structures or write your own custom methods. If your data structures are self-referencing, the code generated by this transform will try to avoid infinite recursion but the algorithm used may not suit your scenario, so use with caution if you have such structures. A future version of this transform may better handle some additional recursive scenarios.

More examples:

 import groovy.transform.EqualsAndHashCode

 @EqualsAndHashCode(includeFields=true)
 class User {
     String name
     boolean active
     List likes
     private int age = 37
 }

 def user = new User(name: 'mrhaki', active: false, likes: ['Groovy', 'Java'])
 def mrhaki = new User(name: 'mrhaki', likes: ['Groovy', 'Java'])
 def hubert = new User(name: 'Hubert Klein Ikkink', likes: ['Groovy', 'Java'])

 assert user == mrhaki
 assert mrhaki != hubert

 Set users = new HashSet()
 users.add user
 users.add mrhaki
 users.add hubert
 assert users.size() == 2
 

Since:

1.8.0

See Also:

HashCodeHelper

Optional Element Summary

Optional Elements

Modifier and Type	Optional Element and Description
boolean	allNames Whether to include all fields and/or properties in equals and hashCode calculations, including those with names that are considered internal.
boolean	allProperties Whether to include all properties (as per the JavaBean spec) in the generated constructor.
boolean	cache Whether to cache hashCode calculations.
boolean	callSuper Whether to include super in equals and hashCode calculations.
String[]	excludes List of property names (and field names if includeFields is true) to exclude from the equals and hashCode calculations.
boolean	includeFields Include fields as well as properties in equals and hashCode calculations.
String[]	includes List of property names (and field names if includeFields is true) to include within the equals and hashCode calculations.
boolean	useCanEqual Generate a canEqual method to be used by equals.

Element Detail

excludes

public abstract String[] excludes

List of property names (and field names if includeFields is true) to exclude from the equals and hashCode calculations. Must not be used if 'includes' is used. 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.

Default:

{}

includes

public abstract String[] includes

List of property names (and field names if includeFields is true) to include within the equals and hashCode calculations. Must not be used if 'excludes' is used. 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. The default value is a special marker value indicating that no includes are defined; all fields and/or properties are included if 'includes' remains undefined and 'excludes' is explicitly or implicitly an empty list.

Default:

"<DummyUndefinedMarkerString-DoNotUse>"

cache

public abstract boolean cache

Whether to cache hashCode calculations. You should only set this to true if you know the object is immutable (or technically mutable but never changed).

Default:

false

callSuper

public abstract boolean callSuper

Whether to include super in equals and hashCode calculations.

Default:

false

includeFields

public abstract boolean includeFields

Include fields as well as properties in equals and hashCode calculations.

Default:

false

useCanEqual

public abstract boolean useCanEqual

Generate a canEqual method to be used by equals.

Default:

true

allProperties

public abstract boolean allProperties

Whether to include all properties (as per the JavaBean spec) in the generated constructor. When true, Groovy treats any explicitly created setXxx() methods as property setters as per the JavaBean specification. JavaBean properties come after any Groovy properties but before any fields for a given class (unless 'includes' is used to determine the order).

Since:

2.5.0

Default:

false

allNames

public abstract boolean allNames

Whether to include all fields and/or properties in equals and hashCode calculations, including those with names that are considered internal.

Since:

2.5.0

Default:

false