groovy.lang
@Documented @Retention(value=SOURCE) @Target(value=TYPE) public @interface Category
ExpandoMetaClass
mechanism which avoids
the category definition altogether. If you already have an investment in
categories or like some of the other features which categories currently give you,
then read on.
The mechanics: during compilation, all methods are transformed to static ones with an additional
self parameter of the type you supply as the annotation parameter (the default type
for the self parameters is Object
which might be more broad reaching than
you like so it is usually wise to specify a type).
Properties invoked using 'this' references are transformed so that
they are instead invoked on the additional self parameter and not on
the Category instance. (Remember that once the category is applied, the reverse
will occur and we will be back to conceptually having methods on the this
references again!)
Classes conforming to the conventional Groovy category conventions can be used
within use
statements or mixed in at compile time with the @Mixin
transformation or at runtime with the mixin
method on classes.
An example showing a use
statement (allowing fine-grained application of
the category methods):
@Category
(Integer)
class IntegerOps {
def triple() {
this * 3
}
}
use (IntegerOps) {
assert 25.triple() == 75
}
Or, using the @Mixin
flavor for compile-time "mixing in" of the methods:
Or, instead of using@Category
(List) class Shuffler { def shuffle() { def result = new ArrayList(this) Collections.shuffle(result) result } }@Mixin
(Shuffler) class Sentence extends ArrayList { Sentence(Collection initial) { super(initial) } } def words = ["The", "quick", "brown", "fox"] println new Sentence(words).shuffle() // => [quick, fox, The, brown] (order will vary)
@Mixin
, try "mixing in" your methods at runtime:
// ... as before ... class Sentence extends ArrayList { Sentence(Collection initial) { super(initial) } } Sentence.mixin Shuffler // ... as before ...
public abstract Class value