FutureCache.java

/*
 * Copyright (C) 2009 eXo Platform SAS.
 *
 * This is free software; you can redistribute it and/or modify it
 * under the terms of the GNU Lesser General Public License as
 * published by the Free Software Foundation; either version 2.1 of
 * the License, or (at your option) any later version.
 *
 * This software is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this software; if not, write to the Free
 * Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
 * 02110-1301 USA, or see the FSF site: http://www.fsf.org.
 */

package org.exoplatform.commons.cache.future;

import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ConcurrentMap;
import java.util.concurrent.ExecutionException;

import org.gatein.common.logging.Logger;
import org.gatein.common.logging.LoggerFactory;

/**
 * <p>
 * A future cache that prevents the loading of the same resource twice. This should be used when the resource to load is very
 * expensive or cannot be concurrently retrieved (like a classloading).
 * </p>
 *
 * <p>
 * The future cache should be used with the {@link #get(Object, Object)} method, that retrieves an object from the cache. When
 * the object is not found, then the {@link Loader#retrieve(Object, Object)} method is used to retrieve the data and then this
 * data is inserted in the cache.
 * </p>
 *
 * <p>
 * The class is abstract and does not implement a cache technology by itself, the cache implementation is delegated to the
 * contractual methods {@link #get(Object)} and {@link #put(Object, Object)}. Those methods are intended to be used by the
 * future cache only.
 * </p>
 *
 * <p>
 * The {@link Loader} interface provides a source to retrieve objects to put in the cache. The goal to maintain this interface
 * is to decouple the cache from the object source.
 * </p>
 *
 * @author <a href="mailto:julien.viet@exoplatform.com">Julien Viet</a>
 * @version $Revision$
 * @param <K> the key type parameter
 * @param <V> the value type parameter
 * @param <C> the context type parameter
 */
public abstract class FutureCache<K, V, C> {

    /** . */
    final Loader<K, V, C> loader;

    /** . */
    private final ConcurrentMap<K, Retrieval<K, V, C>> futureEntries;

    /** . */
    private final Logger log = LoggerFactory.getLogger(FutureCache.class);

    public FutureCache(Loader<K, V, C> loader) {
        this.loader = loader;
        this.futureEntries = new ConcurrentHashMap<K, Retrieval<K, V, C>>();
    }

    /**
     * Retrieves the cached value corresponding to the specified key from the cache, it must returns null when the key does not
     * exist. This method is intended for internal use by the future cache only.
     *
     * @param key the key
     * @return the cache value
     */
    protected abstract V get(K key);

    /**
     * Updates the cache with a new key/value pair. This method is intended for internal use by the future cache only.
     *
     * @param key the key
     * @param value the cache value
     */
    protected abstract void put(K key, V value);

    /**
     * Perform a cache lookup for the specified key within the specified context. When the value cannot be loaded (because it
     * does not exist or it failed or anything else that does not come to my mind), the value null is returned.
     *
     * @param context the context in which the resource is accessed
     * @param key the key identifying the resource
     * @return the value
     */
    public final V get(final C context, final K key) {
        // First we try a simple cache get
        V value = get(key);

        // If it does not succeed then we go through a process that will avoid to load
        // the same resource concurrently
        if (value == null) {
            // Create our future
            Retrieval<K, V, C> retrieval = new Retrieval<K, V, C>(context, key, this);

            // This boolean means we inserted in the local
            boolean inserted = true;

            //
            try {
                Retrieval<K, V, C> phantom = futureEntries.putIfAbsent(key, retrieval);

                // Use the value that could have been inserted by another thread
                if (phantom != null) {
                    retrieval = phantom;
                    inserted = false;
                } else {
                    try {
                        retrieval.current = Thread.currentThread();
                        retrieval.future.run();
                    } catch (Exception e) {
                        log.error("Retrieval of resource " + key + " threw an exception", e);
                    } finally {
                        retrieval.current = null;
                    }
                }

                // Returns the value
                if (retrieval.current == Thread.currentThread()) {
                    throw new IllegalStateException("Reentrancy detected when obtaining key " + key + " with context "
                            + context + " detected");
                } else {
                    try {
                        value = retrieval.future.get();
                    } catch (ExecutionException e) {
                        log.error("Computing of resource " + key + " threw an exception", e.getCause());
                    } catch (InterruptedException e) {
                        // We should handle interruped exception in some manner
                        log.error("Retrieval of resource " + key + " threw an exception", e);
                    }
                }
            } finally {
                // Clean up the per key map but only if our insertion succeeded and with our future
                if (inserted) {
                    futureEntries.remove(key, retrieval);
                }
            }
        }

        //
        return value;
    }
}