C#中的优先级队列实现

时间:2012-05-16 06:33:48

标签: c# .net data-structures

我正在尝试使用SortedDictionary实现优先级队列机制,我想获得有关当前实现的建议。

我的实施如下:

public class PriorityQueue
{
    private Object lockObj;
    private SortedDictionary<PQMsgPriority, Queue<PQMessage>> messageDictionary; 

    public PriorityQueue()
    {
        lockObj = new object();
        messageDictionary = new SortedDictionary<PQMsgPriority, Queue<PQMessage>>();
    }

    public void Enqueue(PQMessage item)
    {
        lock (lockObj)
        {
            if(item != null && item.MsgPriority == PQMsgPriority.None)
            {
                if (messageDictionary.ContainsKey(item.MsgPriority))
                {
                    Queue<PQMessage> dataList = messageDictionary[item.MsgPriority];
                    dataList.Enqueue(item);
                    messageDictionary[item.MsgPriority] = dataList;
                }
                else
                {
                    Queue<PQMessage> dataList = new Queue<PQMessage>();
                    dataList.Enqueue(item);
                    messageDictionary.Add(item.MsgPriority, dataList);
                }
            }
        }
    }

    public PQMessage Dequeue()
    {
        lock (lockObj)
        {
            PQMessage messageData = null;
            PQMsgPriority deleteKey = PQMsgPriority.None;

            //If no data available, throw an exception
            if (messageDictionary.Count == 0)
                throw new InvalidOperationException();

            foreach (KeyValuePair<PQMsgPriority, Queue<PQMessage>> item in messageDictionary)
            {
                Queue<PQMessage> dataList = item.Value;
                messageData = dataList.Dequeue();
                messageDictionary[item.Key] = dataList;

                //If there is no more elements remaining in the list, set a flag (deleteKey) for deleting the key
                if (dataList.Count == 0) 
                    deleteKey = item.Key;

                break;
            }

            //If the deleteKey flag is set, delete the key from the dictionary
            if (deleteKey != PQMsgPriority.None)
                messageDictionary.Remove(deleteKey);

            return messageData;
        }
    }

    public int Count()
    {
        lock (lockObj)
        {
            return messageDictionary.Count;
        }
    }

    public PQMessage Peek()
    {
        lock (lockObj)
        {
            PQMessage messageData = null;

            //If no data available, throw an exception
            if (messageDictionary.Count == 0)
                throw new InvalidOperationException();

            foreach (KeyValuePair<PQMsgPriority, Queue<PQMessage>> item in messageDictionary)
            {
                Queue<PQMessage> dataList = item.Value;
                messageData = dataList.Peek();
                break;
            }

            return messageData;
        }
    }
}

public enum PQMsgPriority
{
    High = 0,
    Medium = 1,
    Low = 2,
    None = 3
}

public class PQMessage
{
    private PQMsgPriority msgPriority;
    private Object message;

    #region Properties
    public PQMsgPriority MsgPriority
    {
        get { return msgPriority; }
        set { msgPriority = value; }
    }
    public Object Message
    {
        get { return message; }
        set { message = value; }
    }
    #endregion

    public PQMessage(PQMsgPriority msgPriority, Object message)
    {
        this.msgPriority = msgPriority;
        this.message = message;
    }
}

如果有任何其他方法来实现优先级队列,请指出我正确的方向。

2 个答案:

答案 0 :(得分:2)

(我应该放弃这个,因为我假设你是为了学习而不是防弹实现。如果想要一些正常的东西,那么你最好重用现有的实现。)

只是一些一般性评论。在下面的示例中,第三行是不必要的。

Queue<PQMessage> dataList = messageDictionary[item.MsgPriority];
dataList.Enqueue(item);
messageDictionary[item.MsgPriority] = dataList;

dataList返回的messageDictionary是地图中参考的副本。这意味着,当您Enqueue数据时,它正在与之前相同的底层队列(不是队列的副本)上工作,因此无需再将其重新放回,您只需删除该行即可。

在你的dequeue实现中,你有一个循环,你每次都在第一个元素上打破(例如你只绕过循环一次)。也许您可以调查使用LINQ来获取First元素并立即返回? (类似于Peek实施)。

最后,假设PQMessage本身具有优先权,也许您可​​以考虑使用SortedList进行实施? (见here

答案 1 :(得分:1)

如果您需要提高concurency性能,则只能锁定一个队列进行写入。 (整个SortedDictionary的读锁仍然在线)。

不要忘记按正确的顺序设置和释放锁