One useful Groovy iterator is the SyncIterator, based on Ruby's SyncEnumerator which can iterate across several data structures at the same time. It makes use of DefaultTypeTransformation.asCollection() to perform the same coercion that Groovy uses by default to implement methods such as each(), collect(), and the like...
import org.codehaus.groovy.runtime.typehandling.DefaultTypeTransformation;
public class SyncIterator implements Iterator, Iterable{
private theobjects
SyncIterator(Object[] objects){
theobjects=objects.collect{
if (it instanceof Iterator) return /*from closure*/ it
else return /*from closure*/ DefaultTypeTransformation.asCollection(it).iterator()
}
}
boolean hasNext(){
return theobjects.any{it.hasNext()}
}
Object next(){
if (!hasNext()) throw new java.util.NoSuchElementException()
return theobjects.collect{
try{
return /*from closure*/ it.next()
}catch(NoSuchElementException e){
return /*from closure*/ null
}
}
}
Iterator iterator(){
return this;
}
void remove(){
throw new UnsupportedOperationException("remove() not supported")
}
}
Another useful iterator is the Generator, which takes a closure and offers up the stuff it yields as an iterator, without having to keep all of the generated data around in memory. (It's based loosely on Ruby 1.9's Generator, which was also the challenge in Ruby Quiz #66)
import java.util.concurrent.*;
import java.lang.ref.*;
import groovy.lang.Closure;
import java.util.*;
public class Generator<T> implements Iterator<T>, Iterable<T>{
Semaphore availSemaphore=new Semaphore(0);
Semaphore emptySemaphore=new Semaphore(1);
T pushedValue=null;
T pulledValue=null;
boolean hasPulledValue=false;
Thread internalThread;
Generator(Closure closure){
internalThread=new GeneratorThread<T>(this,closure);
internalThread.setDaemon(true);
internalThread.start();
}
private void pullValue(){
availSemaphore.acquireUninterruptibly();
pulledValue=pushedValue;
pushedValue=null;
hasPulledValue=true;
emptySemaphore.release();
}
public boolean hasNext(){
if (!hasPulledValue)
pullValue();
return emptySemaphore.availablePermits() != 2;
}
public T next(){
if (!hasNext())
throw new NoSuchElementException("Closure has no more values");
T retval=pulledValue;
hasPulledValue=false;
return retval;
}
public void remove(){
throw new UnsupportedOperationException(
"Remove is not supported on generators");
}
public Iterator<T> iterator(){
return this;
}
public void finalize(){
internalThread.interrupt();
}
static class GeneratorThread<T> extends Thread{
WeakReference<Generator<T>> generatorRef;
Closure closure;
public GeneratorThread(Generator<T> generator, Closure cl){
generatorRef=new WeakReference<Generator<T>>(generator);
closure=cl;
}
public void run(){
closure.call(new SaveClosure<T>(this));
Generator generator=generatorRef.get();
if (generator!=null){
generator.availSemaphore.release();
}
}
}
static class SaveClosure<T> extends Closure{
WeakReference<Generator<T>> generatorRef;
Semaphore emptySemaphore;
Semaphore availSemaphore;
public SaveClosure(GeneratorThread<T> gt){
super(gt,null);
generatorRef=gt.generatorRef;
Generator<T> generator=generatorRef.get();
if (generator!=null){
emptySemaphore=generator.emptySemaphore;
availSemaphore=generator.availSemaphore;
}else{
throw new GeneratorDisposedException();
}
}
public void doCall(T value){
try{
emptySemaphore.acquire();
}catch(InterruptedException e){
throw new GeneratorDisposedException();
}
Generator<T> generator=generatorRef.get();
if (generator!=null){
generator.pushedValue=value;
}else{
throw new GeneratorDisposedException();
}
availSemaphore.release();
}
}
/**
* A GeneratorDisposedException is used to terminate the thread
* that was generating values, once the Generator has been garbage
* collected.
*/
static public class GeneratorDisposedException extends RuntimeException{
}
}
Note that Groovy doesn't have this concept built in. Its MethodClosure to Iterator conversion loads everything into an array all at once, which isn't a particularly good idea you may be running find() on it, and only need to generate the first item to find what you're looking for.
For example, consider the following use of a Generator:
def fibonacci(Closure yield){
def a=0
def b=1
def temp
while(true){
yield(b)
temp=a
a=b
b=a+temp
}
}
println(new Generator(this.&fibonacci).find{ it % 20 == 0})
this.&fibonacci.find{it % 20 == 0}
As the Groovy runtime implements this now, you would exhaust all available ram when converting the MethodClosure to an iterator, before find() was ever called. With a Generator, values are only generated on demand.
NOTE: because of the use of threads, the generator may generate one more value than is actually needed before the garbage collector disposes of the generator.