Java time-based map/cache with expiring keys

Apache Commons has decorator for Map to expire entries: PassiveExpiringMap It's more simple than caches from Guava.

P.S. be careful, it's not synchronized.

This is a sample implementation that i did for the same requirement and concurrency works well. Might be useful for someone.

import java.text.SimpleDateFormat;
import java.util.Date;
import java.util.Map;
import java.util.concurrent.ConcurrentHashMap;

/**
 * 
 * @author Vivekananthan M
 *
 * @param <K>
 * @param <V>
 */
public class WeakConcurrentHashMap<K, V> extends ConcurrentHashMap<K, V> {

    private static final long serialVersionUID = 1L;

    private Map<K, Long> timeMap = new ConcurrentHashMap<K, Long>();
    private long expiryInMillis = 1000;
    private static final SimpleDateFormat sdf = new SimpleDateFormat("hh:mm:ss:SSS");

    public WeakConcurrentHashMap() {
        initialize();
    }

    public WeakConcurrentHashMap(long expiryInMillis) {
        this.expiryInMillis = expiryInMillis;
        initialize();
    }

    void initialize() {
        new CleanerThread().start();
    }

    @Override
    public V put(K key, V value) {
        Date date = new Date();
        timeMap.put(key, date.getTime());
        System.out.println("Inserting : " + sdf.format(date) + " : " + key + " : " + value);
        V returnVal = super.put(key, value);
        return returnVal;
    }

    @Override
    public void putAll(Map<? extends K, ? extends V> m) {
        for (K key : m.keySet()) {
            put(key, m.get(key));
        }
    }

    @Override
    public V putIfAbsent(K key, V value) {
        if (!containsKey(key))
            return put(key, value);
        else
            return get(key);
    }

    class CleanerThread extends Thread {
        @Override
        public void run() {
            System.out.println("Initiating Cleaner Thread..");
            while (true) {
                cleanMap();
                try {
                    Thread.sleep(expiryInMillis / 2);
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
            }
        }

        private void cleanMap() {
            long currentTime = new Date().getTime();
            for (K key : timeMap.keySet()) {
                if (currentTime > (timeMap.get(key) + expiryInMillis)) {
                    V value = remove(key);
                    timeMap.remove(key);
                    System.out.println("Removing : " + sdf.format(new Date()) + " : " + key + " : " + value);
                }
            }
        }
    }
}

Git Repo Link (With Listener Implementation)

https://github.com/vivekjustthink/WeakConcurrentHashMap

Cheers!!

Yes. Google Collections, or Guava as it is named now has something called MapMaker which can do exactly that.

ConcurrentMap<Key, Graph> graphs = new MapMaker()
   .concurrencyLevel(4)
   .softKeys()
   .weakValues()
   .maximumSize(10000)
   .expiration(10, TimeUnit.MINUTES)
   .makeComputingMap(
       new Function<Key, Graph>() {
         public Graph apply(Key key) {
           return createExpensiveGraph(key);
         }
       });

Update:

As of guava 10.0 (released September 28, 2011) many of these MapMaker methods have been deprecated in favour of the new CacheBuilder:

LoadingCache<Key, Graph> graphs = CacheBuilder.newBuilder()
    .maximumSize(10000)
    .expireAfterWrite(10, TimeUnit.MINUTES)
    .build(
        new CacheLoader<Key, Graph>() {
          public Graph load(Key key) throws AnyException {
            return createExpensiveGraph(key);
          }
        });