.NET深入了解哈希表和Dictionary

引子

问题:给定一串数字{1,2,5,7,15,24,33,52},如何在时间复杂度为O(1)下,对数据进行CURD?

数组:我创建一个Length为53的数组,将元素插入相同下标处,是不是就可以实现查找复杂度O(1)了?但是添加修改元素时间复杂度为O(n)了。

链表:添加删除复杂度为O(1),但是查找时间复杂度为O(n)了。

身为.NETer肯定熟练使用Dictionary和HashSet,这两个容器的底层就是HashTable,所以带着对技术浓重的兴趣(面试),所以就从头到尾梳理一下!

理论

链地址法(拉链法)

回到问题本身,我们用数组可以实现查找复杂度为O(1),链表实现添加删除复杂度为O(1),如果我们将两个合起来,不就可以实现增删查都为O(1)了么?如何结合呢?

我们先定义一个数组,长度为7(敲黑板,思考下为什么选7?),将所有元素对7取余,这样所有元素都可以放在数组上了,如下图所示:

《.NET深入了解哈希表和Dictionary》

如上图,如果我们将数组中每个下标位置都放成一个链条,这样,复杂度不久降下去了么?

有问题么?没问题。真没问题么?有问题……

注意

  1. 插入元素是{0,7,14,21,28}怎么办?这样都落在下标为0的链条里,时间复杂度不又上去了?针对这种情况,隔壁Java将链表优化成了红黑书,我们.NET呢?往下看。

  2. 如果我的数组长度不是7,是2怎么办?所有数对2取余,不是1就是0,时间复杂度不又上去了?所以我们对数组长度应该取素数。

  3. 如果元素超级多或者特别少,我们的数组长度要固定么?就要动态长度

上边这种方法学名就叫拉链法!

开放地址法

上边我们聊过拉链法(为什么老想着裤子拉链……),拉链法是向下开辟新的空间,如果我们横向开辟空间呢?还是刚才的例子,我们这样搞一下试试。

线性探测法

《.NET深入了解哈希表和Dictionary》

我们插完7以后,在插24时,发现下标为2的地方有元素了,于是向后移动一位,发现有空位,于是就插进去了。

上边这种方法就是线性探测法!

二次聚集(堆积)

聪明的老鸟们,肯定疑惑啦,如果我们继续添加元素{x%11=4},{y%11=5},此时x,y元素都要往下标6插数据。这样就导致了原始哈希地址不同的元素要插入同一个地址。即添加同义词的冲突过程中又添加了非同义词的冲突。这就是二次聚集

二次探测法

如果在线性探测法中,我们不依次寻找下一个呢?我们针对”下一个”采取{1 ^ 2,-1 ^ 2,2 ^ 2,-2 ^ 2….}(垃圾编辑器,次方样式乱了)这样的步长呢?真聪明!你已经知道二次探测法了!

这……这还能用么?不都乱了么?下标和元素对不上了呀!怎么去查找元素呢?

别急呀,家人们呐,我们按照这个思路查询就好了:

查找算法步骤

1. 给定待查找的关键字key,获取原始应该插入的下标index
2. 如果原始下标index处,元素为空,则所查找的元素不存在
3. 如果index处的元素等于key,则查找成功
4. 否则重复下述解决冲突的过程
  * 按照处理冲突的方法,计算下一个地址nextIndex
  * 若nextIndex为空,则查找元素不存在
  * 若nextIndex等于关键词key,则查找成功

还有要注意的点么?必须有!

注意(敲重点啦)

  1. 数组长度必须大于给定元素的长度!
  2. 当数组元素快装满时,时间复杂度也是O(n)!
  3. 如果都装满了,就会一直循环找空位,我们应该进行扩容!

理论小结

《.NET深入了解哈希表和Dictionary》

接口设计

干活啦,干活啦,领导嫌查询效率太低,让设计一种CURD时间复杂度都为O(n)的数据结构。给了接口。接口如下:

internal interface IDictionary<TK, TV> : IEnumerable<KeyValuePair<TK, TV>>
{
    TV this[TK key] { get; set; }

    int Count { get; }
    /// <summary>
    /// 根据key判断元素是否存在
    /// </summary>
    /// <param name="key"></param>
    /// <returns></returns>
    bool ContainsKey(TK key);
    /// <summary>
    /// 添加元素
    /// </summary>
    /// <param name="key"></param>
    /// <param name="value"></param>
    void Add(TK key, TV value);
    /// <summary>
    /// 根据key移除元素
    /// </summary>
    /// <param name="key"></param>
    void Remove(TK key);
    /// <summary>
    /// 清除
    /// </summary>
    void Clear();
}

.NET实现线性探测法

实现过程

1. 先来个对象,存储key和value

对象:KeyValuePair

internal class DictionaryKeyValuePair<TK, TV>
{
    internal TK Key;
    internal TV Value;

    internal DictionaryKeyValuePair(TK key, TV value)
    {
        Key = key;
        Value = value;
    }
}

2. 来个类OpenAddressDictionary,继承IDictionary接口,就是我们的实现类

实现类:OpenAddressDictionary

/// <summary>
/// 使用线性探测法实现哈希表
/// </summary>
/// <typeparam name="TK"></typeparam>
/// <typeparam name="TV"></typeparam>
public class OpenAddressDictionary<TK, TV> : IDictionary<TK, TV>
{
    //创建一个数组,用来存储元素
    private DictionaryKeyValuePair<TK, TV>[] hashArray;

    //记录已插入元素的数量
    public int Count { get; private set; }

    public OpenAddressDictionary(int capacity)
    {
        if (capacity < 0)
            throw new ArgumentOutOfRangeException("初始值容量不能小于0");
        hashArray = new DictionaryKeyValuePair<TK, TV>[capacity];
    }
    public TV this[TK key] {
        get => throw new System.NotImplementedException();
        set => throw new System.NotImplementedException();
    }

    public void Add(TK key, TV value)
    {
        throw new System.NotImplementedException();
    }

    public void Clear()
    {
        throw new System.NotImplementedException();
    }

    public System.Boolean ContainsKey(TK key)
    {
        throw new System.NotImplementedException();
    }

    public IEnumerator<KeyValuePair<TK, TV>> GetEnumerator()
    {
        throw new System.NotImplementedException();
    }

    public void Remove(TK key)
    {
        throw new System.NotImplementedException();
    }

    IEnumerator IEnumerable.GetEnumerator()
    {
        throw new System.NotImplementedException();
    }
}

3.如何实现查找?跟着上文查找步骤就行

线性探测:查找

/// <summary>
/// 查找,按照上文线性探测查找步骤
/// </summary>
/// <param name="key"></param>
/// <returns></returns>
public bool ContainsKey(TK key)
{
    //1.给定待查找的关键字key,获取原始应该插入的下标index
    var hashCode = GetHash(key);
    var index = hashCode % hashArray.Length;

    //2.如果原始下标index处,元素为空,则所查找的元素不存在
    if (hashArray[index] == null) return false;

    var current = hashArray[index];//当前元素

    /*这个点用来判断是否走了一整圈*/
    var hitKey = current.Key;

    //4.否则重复下述解决冲突的过程           
    while (current != null)
    {
        //3.如果index处的元素等于key,则查找成功
        if (current.Key.Equals(key)) return true;

        /*这个地方来修改获取下一个元素位置*/
        index++;

        /*到尾了,但是没有走完一圈*/
        if (index == hashArray.Length)
            index = 0;

        current = hashArray[index];

        /*走完一圈了,没找到*/
        if (current != null && current.Key.Equals(hitKey)) break;
    }

    return false;
}

4. 添加

线性探测:添加

/// <summary>
/// 添加元素
/// </summary>
/// <param name="key"></param>
/// <param name="value"></param>
/// <exception cref="Exception"></exception>
public void Add(TK key, TV value)
{
    Grow();
    //1.获取原始插入位置
    var hashCode = GetHash(key);
    var index = hashCode % hashArray.Length;

    //2.此位置为空,直接插入
    if (hashArray[index] == null)
    {
        hashArray[index] = new DictionaryKeyValuePair<TK, TV>(key, value);
    }
    //3.坑被占了,去看看下一个
    else
    {
        var current = hashArray[index];
        /*这个点用来判断是否走了一整圈*/
        var hitKey = current.Key;
        while (current != null)
        {
            if (current.Key.Equals(key)) throw new Exception("重复key");

            /*这个地方来修改获取下一个元素位置*/
            index++;

            /*到尾了,但是没有走完一圈*/
            if (index == hashArray.Length)
                index = 0;

            current = hashArray[index];

            /*走完一圈了,没找到空位*/
            if (current != null && current.Key.Equals(hitKey)) throw new Exception("容器满了");
        }
        hashArray[index] = new DictionaryKeyValuePair<TK, TV>(key, value);
    }
    Count++;
}
/// <summary>
/// 扩容
/// </summary>
private void Grow()
{
    /*这个地方判断使用多少扩容*/
    if (hashArray.Length * 0.7 <= Count)
    {
        var orghashArray = hashArray.Length;
        var currentArray = hashArray;

        /*这个地方改变扩容大小的规则*/
        hashArray = new DictionaryKeyValuePair<TK, TV>[hashArray.Length * 2];
        for (var i = 0; i < orghashArray; i++)
        {
            var current = currentArray[i];

            /*旧数组中存在元素,添加到新数组中,Add方法会对Count++,所以加入后要Count--*/
            if (current != null)
            {
                Add(current.Key, current.Value);
                Count--;
            }
        }
        currentArray = null;
    }
}

5. 删除

线性探测:删除

/// <summary>
/// 删除元素key
/// </summary>
/// <param name="key"></param>
/// <exception cref="Exception"></exception>
public void Remove(TK key)
{
    //1.获取原始插入位置
    var hashCode = GetHash(key);
    var curIndex = hashCode % hashArray.Length;

    //2.此位置为空,无法删除
    if (hashArray[curIndex] == null) throw new Exception("未找到元素key");

    var current = hashArray[curIndex];

    /*这个点用来判断是否走了一整圈*/
    var hitKey = current.Key;

    #region 找到待删除元素
    DictionaryKeyValuePair<TK, TV> target = null;

    while (current != null)
    {
        if (current.Key.Equals(key))
        {
            target = current;
            break;
        }

        /*这个地方来修改获取下一个元素位置*/
        curIndex++;

        /*到尾了,但是没有走完一圈*/
        if (curIndex == hashArray.Length)
            curIndex = 0;

        current = hashArray[curIndex];

        /*走完一圈了,没找到空位*/
        if (current != null && current.Key.Equals(hitKey)) throw new Exception("No such item for given key");
    }

    if (target == null)
    {
        throw new Exception("未找到元素key");
    }
    #endregion  


    //删除,将当前位置置空
    hashArray[curIndex] = null;

    #region 之前讲过删除,造成元素丢失,所以在此处处理

    curIndex++;

    /*到尾了,但是没有走完一圈*/
    if (curIndex == hashArray.Length)
        curIndex = 0;

    current = hashArray[curIndex];

    //直到下一个为空的点,到空说明后边的还没有被线性探测插入污染
    while (current != null)
    {
        //先删除
        hashArray[curIndex] = null;

        //重新插入
        Add(current.Key, current.Value);
        Count--;

        curIndex++;

        /*到尾了,但是没有走完一圈*/
        if (curIndex == hashArray.Length)
            curIndex = 0;

        current = hashArray[curIndex];
    }
    #endregion


    Count--;

    Shrink();
}

/// <summary>
/// 减容
/// </summary>
private void Shrink()
{
    /*这个地方判断元素在什么程度算少*/
    if (Count <= hashArray.Length * 0.3 && hashArray.Length / 2 > 0)
    {
        var orghashArray = hashArray.Length;
        var currentArray = hashArray;

        /*这个地方改变扩容大小的规则*/
        hashArray = new DictionaryKeyValuePair<TK, TV>[hashArray.Length / 2];

        for (var i = 0; i < orghashArray; i++)
        {
            var current = currentArray[i];

            /*旧数组中存在元素,添加到新数组中,Add方法会对Count++,所以加入后要Count--*/
            if (current != null)
            {
                Add(current.Key, current.Value);
                Count--;
            }
        }

        currentArray = null;
    }
}

最终代码

线性探测:最终代码

/// <summary>
/// 使用线性探测法实现哈希表
/// </summary>
/// <typeparam name="TK"></typeparam>
/// <typeparam name="TV"></typeparam>
public class OpenAddressDictionary<TK, TV> : IDictionary<TK, TV>
{
    //创建一个数组,用来存储元素
    private DictionaryKeyValuePair<TK, TV>[] hashArray;
    //记录已插入元素的数量
    public int Count { get; private set; }

    public TV this[TK key]
    {
        get => GetValue(key);
        set => SetValue(key, value);
    }

    public OpenAddressDictionary(int capacity)
    {
        if (capacity < 0)
            throw new ArgumentOutOfRangeException("初始值容量不能小于0");
        hashArray = new DictionaryKeyValuePair<TK, TV>[capacity];
    }
    /// <summary>
    /// 清除最简单
    /// </summary>
    public void Clear()
    {
        if (Count > 0)
            Array.Clear(hashArray, 0, hashArray.Length);
    }

    /// <summary>
    /// 查找,按照上文线性探测查找步骤
    /// </summary>
    /// <param name="key"></param>
    /// <returns></returns>
    public bool ContainsKey(TK key)
    {
        //1.给定待查找的关键字key,获取原始应该插入的下标index
        var hashCode = GetHash(key);
        var index = hashCode % hashArray.Length;

        //2.如果原始下标index处,元素为空,则所查找的元素不存在
        if (hashArray[index] == null) return false;

        var current = hashArray[index];//当前元素

        /*这个点用来判断是否走了一整圈*/
        var hitKey = current.Key;

        //4.否则重复下述解决冲突的过程           
        while (current != null)
        {
            //3.如果index处的元素等于key,则查找成功
            if (current.Key.Equals(key)) return true;

            /*这个地方来修改获取下一个元素位置*/
            index++;

            /*到尾了,但是没有走完一圈*/
            if (index == hashArray.Length)
                index = 0;

            current = hashArray[index];

            /*走完一圈了,没找到*/
            if (current != null && current.Key.Equals(hitKey)) break;
        }

        return false;
    }

    /// <summary>
    /// 添加元素
    /// </summary>
    /// <param name="key"></param>
    /// <param name="value"></param>
    /// <exception cref="Exception"></exception>
    public void Add(TK key, TV value)
    {
        Grow();
        //1.获取原始插入位置
        var hashCode = GetHash(key);
        var index = hashCode % hashArray.Length;

        //2.此位置为空,直接插入
        if (hashArray[index] == null)
        {
            hashArray[index] = new DictionaryKeyValuePair<TK, TV>(key, value);
        }
        //3.坑被占了,去看看下一个
        else
        {
            var current = hashArray[index];
            /*这个点用来判断是否走了一整圈*/
            var hitKey = current.Key;
            while (current != null)
            {
                if (current.Key.Equals(key)) throw new Exception("重复key");

                /*这个地方来修改获取下一个元素位置*/
                index++;

                /*到尾了,但是没有走完一圈*/
                if (index == hashArray.Length)
                    index = 0;

                current = hashArray[index];

                /*走完一圈了,没找到空位*/
                if (current != null && current.Key.Equals(hitKey)) throw new Exception("容器满了");
            }
            hashArray[index] = new DictionaryKeyValuePair<TK, TV>(key, value);
        }
        Count++;
    }
    /// <summary>
    /// 删除元素key
    /// </summary>
    /// <param name="key"></param>
    /// <exception cref="Exception"></exception>
    public void Remove(TK key)
    {
        //1.获取原始插入位置
        var hashCode = GetHash(key);
        var curIndex = hashCode % hashArray.Length;

        //2.此位置为空,无法删除
        if (hashArray[curIndex] == null) throw new Exception("未找到元素key");

        var current = hashArray[curIndex];

        /*这个点用来判断是否走了一整圈*/
        var hitKey = current.Key;

        #region 找到待删除元素
        DictionaryKeyValuePair<TK, TV> target = null;

        while (current != null)
        {
            if (current.Key.Equals(key))
            {
                target = current;
                break;
            }

            /*这个地方来修改获取下一个元素位置*/
            curIndex++;

            /*到尾了,但是没有走完一圈*/
            if (curIndex == hashArray.Length)
                curIndex = 0;

            current = hashArray[curIndex];

            /*走完一圈了,没找到空位*/
            if (current != null && current.Key.Equals(hitKey)) throw new Exception("No such item for given key");
        }

        if (target == null)
        {
            throw new Exception("未找到元素key");
        }
        #endregion  


        //删除,将当前位置置空
        hashArray[curIndex] = null;

        #region 之前讲过删除,造成元素丢失,所以在此处处理

        curIndex++;

        /*到尾了,但是没有走完一圈*/
        if (curIndex == hashArray.Length)
            curIndex = 0;

        current = hashArray[curIndex];

        //直到下一个为空的点,到空说明后边的还没有被线性探测插入污染
        while (current != null)
        {
            //先删除
            hashArray[curIndex] = null;

            //重新插入
            Add(current.Key, current.Value);
            Count--;

            curIndex++;

            /*到尾了,但是没有走完一圈*/
            if (curIndex == hashArray.Length)
                curIndex = 0;

            current = hashArray[curIndex];
        }
        #endregion


        Count--;

        Shrink();
    }

    /// <summary>
    /// 扩容
    /// </summary>
    private void Grow()
    {
        /*这个地方判断使用多少扩容*/
        if (hashArray.Length * 0.7 <= Count)
        {
            var orghashArray = hashArray.Length;
            var currentArray = hashArray;

            /*这个地方改变扩容大小的规则*/
            hashArray = new DictionaryKeyValuePair<TK, TV>[hashArray.Length * 2];
            for (var i = 0; i < orghashArray; i++)
            {
                var current = currentArray[i];

                /*旧数组中存在元素,添加到新数组中,Add方法会对Count++,所以加入后要Count--*/
                if (current != null)
                {
                    Add(current.Key, current.Value);
                    Count--;
                }
            }
            currentArray = null;
        }
    }
    /// <summary>
    /// 减容
    /// </summary>
    private void Shrink()
    {
        /*这个地方判断元素在什么程度算少*/
        if (Count <= hashArray.Length * 0.3 && hashArray.Length / 2 > 0)
        {
            var orghashArray = hashArray.Length;
            var currentArray = hashArray;

            /*这个地方改变扩容大小的规则*/
            hashArray = new DictionaryKeyValuePair<TK, TV>[hashArray.Length / 2];

            for (var i = 0; i < orghashArray; i++)
            {
                var current = currentArray[i];

                /*旧数组中存在元素,添加到新数组中,Add方法会对Count++,所以加入后要Count--*/
                if (current != null)
                {
                    Add(current.Key, current.Value);
                    Count--;
                }
            }

            currentArray = null;
        }
    }

    private void SetValue(TK key, TV value)
    {
        var index = GetHash(key) % hashArray.Length;

        if (hashArray[index] == null)
        {
            Add(key, value);
        }
        else
        {
            var current = hashArray[index];
            var hitKey = current.Key;

            while (current != null)
            {
                if (current.Key.Equals(key))
                {
                    Remove(key);
                    Add(key, value);
                    return;
                }

                index++;

                //wrap around
                if (index == hashArray.Length)
                    index = 0;

                current = hashArray[index];

                //reached original hit again
                if (current != null && current.Key.Equals(hitKey)) throw new Exception("Item not found");
            }
        }

        throw new Exception("Item not found");
    }

    private TV GetValue(TK key)
    {
        var index = GetHash(key) % hashArray.Length;

        if (hashArray[index] == null) throw new Exception("Item not found");

        var current = hashArray[index];
        var hitKey = current.Key;

        while (current != null)
        {
            if (current.Key.Equals(key)) return current.Value;

            index++;

            //wrap around
            if (index == hashArray.Length)
                index = 0;

            current = hashArray[index];

            //reached original hit again
            if (current != null && current.Key.Equals(hitKey)) throw new Exception("Item not found");
        }

        throw new Exception("Item not found");
    }

    private int GetHash(TK key)
    {
        return Math.Abs(key.GetHashCode());
    }
    IEnumerator IEnumerable.GetEnumerator()
    {
        return GetEnumerator();
    }
    //迭代器就不写了,想了解看我博客容器栏目
    public IEnumerator<KeyValuePair<TK, TV>> GetEnumerator()
    {
        throw new System.NotImplementedException();
    }
}
internal class DictionaryKeyValuePair<TK, TV>
{
    internal TK Key;
    internal TV Value;

    internal DictionaryKeyValuePair(TK key, TV value)
    {
        Key = key;
        Value = value;
    }
}

.NET实现拉链法

实现过程

回想一下,上边的拉链法,每个下标位置放置的是一个链条,所以我们先实现一个双向链表

1. 实现一个双向链表

拉链法:构建双向链表

internal class DLinkedNode<T>
{
    public T Data;
    public DLinkedNode<T> Next;
    public DLinkedNode<T> Previous;

    public DLinkedNode(T data)
    {
        Data = data;
    }
}

2. 创建一个拉链法实体类

拉链法:实现类

/// <summary>
/// 拉链法:实现类
/// </summary>
/// <typeparam name="TK"></typeparam>
/// <typeparam name="TV"></typeparam>
internal class SeparateChainingDictionary<TK, TV>:IDictionary<TK, TV>
{
    //构建一个数组,数组每个节点都是链表
    private DLinkedNode<KeyValuePair<TK, TV>>[] hashArray;
    //已使用数组下标个数
    private int filledBuckets;

    public SeparateChainingDictionary(int capacity) {
        if (capacity < 0)
            throw new ArgumentOutOfRangeException("初始值容量不能小于0");
        hashArray = new DLinkedNode<KeyValuePair<TK, TV>>[capacity];
    }
    public TV this[TK key] { 
        get => throw new NotImplementedException(); 
        set => throw new NotImplementedException(); 
    }

    public int Count => throw new NotImplementedException();

    public void Add(TK key, TV value)
    {
        throw new NotImplementedException();
    }

    public void Clear()
    {
        throw new NotImplementedException();
    }

    public bool ContainsKey(TK key)
    {
        throw new NotImplementedException();
    }

    public void Remove(TK key)
    {
        throw new NotImplementedException();
    }

    public IEnumerator<KeyValuePair<TK, TV>> GetEnumerator()
    {
        throw new NotImplementedException();
    }

    IEnumerator IEnumerable.GetEnumerator()
    {
        throw new NotImplementedException();
    }
}

3. 拉链法:查找

拉链法:查找

/// <summary>
/// 查找
/// </summary>
/// <param name="key"></param>
/// <returns></returns>
public bool ContainsKey(TK key)
{
    /*1.获取原始下标*/
    var index = Math.Abs(key.GetHashCode()) % hashArray.Length;

    /*2.为空即无*/
    if (hashArray[index] == null) return false;

    var current = hashArray[index];

    /*3.遍历链表*/
    while (current != null)
    {               
        if (current.Data.Key.Equals(key)) return true;

        current = current.Next;
    }

    return false;
}

4. 拉链法:添加

拉链法:添加

/// <summary>
/// 添加
/// </summary>
/// <param name="key"></param>
/// <param name="value"></param>
/// <exception cref="Exception"></exception>
public void Add(TK key, TV value)
{
    Grow();

    var index = Math.Abs(key.GetHashCode()) % hashArray.Length;

    if (hashArray[index] == null)
    {
        hashArray[index] = new DLinkedNode<KeyValuePair<TK, TV>>(new KeyValuePair<TK, TV>(key, value));
        filledBuckets++;
    }
    else
    {
        var current = hashArray[index];

        while (current != null && current.Next != null)
        {
            /*此处可以判断是重复修改,还是抛异常*/
            if (current.Data.Key.Equals(key)) throw new Exception("重复key");

            current = current.Next;
        }
        if (current.Data.Key.Equals(key)) throw new Exception("重复key");
        current.Next = new DLinkedNode<KeyValuePair<TK, TV>>(new KeyValuePair<TK, TV>(key, value));
    }

    Count++;
}

/// <summary>
/// 扩容
/// </summary>
private void Grow()
{
    if (filledBuckets >= hashArray.Length * 0.7)
    {
        filledBuckets = 0;

        var newBucketSize = hashArray.Length * 2;

        var biggerArray = new DLinkedNode<KeyValuePair<TK, TV>>[newBucketSize];

        for (var i = 0; i < hashArray.Length; i++)
        {
            var item = hashArray[i];

            if (item != null)
            {
                var current = item;

                while (current != null)
                {
                    var next = current.Next;

                    var newIndex = Math.Abs(current.Data.Key.GetHashCode()) % newBucketSize;

                    if (biggerArray[newIndex] == null)
                    {
                        filledBuckets++;
                        biggerArray[newIndex] = current;
                    }

                    var bItem = biggerArray[newIndex];
                    while(bItem.Next != null)
                        bItem = bItem.Next;
                    bItem.Next = current;

                    current = next;
                }
            }
        }

        hashArray = biggerArray;
    }
}

5. 拉链法:删除

拉链法:删除

public void Remove(TK key)
{
    var index = Math.Abs(key.GetHashCode()) % hashArray.Length;

    if (hashArray[index] == null) throw new Exception("未找到key");

    var current = hashArray[index];

    /*查找待删除元素*/
    DLinkedNode<KeyValuePair<TK, TV>> item = null;
    while (current != null)
    {
        if (current.Data.Key.Equals(key))
        {
            item = current;
            break;
        }
        current = current.Next;
    }

    if (item == null)
    {
        throw new Exception("未找到key");
    }

    /*删除*/
    if (item.Next == null)
        item = null;
    else
    {
        item.Previous = item.Next; 
        item.Next.Previous =item.Previous ;
        item = null;
    }


    if (hashArray[index] == null)
    {
        filledBuckets--;
    }

    Count--;

    Shrink();
}
private void Shrink()
{
    /*是否减容*/
    if (Math.Abs(filledBuckets - hashArray.Length * 0.3) < 0.1 && hashArray.Length / 2 > 0)
    {
        filledBuckets = 0;
        var newBucketSize = hashArray.Length / 2;

        var smallerArray = new DLinkedNode<KeyValuePair<TK, TV>>[newBucketSize];

        for (var i = 0; i < hashArray.Length; i++)
        {
            var item = hashArray[i];

            if (item != null)
            {
                var current = item;

                /*找到新的存储点*/
                while (current != null)
                {
                    var next = current.Next;

                    var newIndex = Math.Abs(current.Data.Key.GetHashCode()) % newBucketSize;

                    if (smallerArray[newIndex] == null)
                    {
                        filledBuckets++;
                        smallerArray[newIndex] = current;
                    }

                    var newItem = smallerArray[newIndex];
                    while(newItem.Next != null)
                        newItem = newItem.Next;
                    newItem.Next = current;

                    current = next;
                }
            }
        }

        hashArray = smallerArray;
    }
}

最终代码

拉链法:最终代码


internal class DLinkedNode<T>
{
  public T Data;
  public DLinkedNode<T> Next;
  public DLinkedNode<T> Previous;

  public DLinkedNode(T data)
  {
      Data = data;
  }
}
internal class SeparateChainingDictionary<TK, TV> : IDictionary<TK, TV>
{
//构建一个数组,数组每个节点都是链表
private DLinkedNode<KeyValuePair<TK, TV>>[] hashArray;
//已使用数组下标个数
private int filledBuckets;

public SeparateChainingDictionary(int capacity)
{
    if (capacity < 0)
        throw new ArgumentOutOfRangeException("初始值容量不能小于0");
    hashArray = new DLinkedNode<KeyValuePair<TK, TV>>[capacity];
}
public TV this[TK key]
{
    get => throw new NotImplementedException();
    set => throw new NotImplementedException();
}

public int Count { get; private set; }

/// <summary>
/// 添加
/// </summary>
/// <param name="key"></param>
/// <param name="value"></param>
/// <exception cref="Exception"></exception>
public void Add(TK key, TV value)
{
    Grow();

    var index = Math.Abs(key.GetHashCode()) % hashArray.Length;

    if (hashArray[index] == null)
    {
        hashArray[index] = new DLinkedNode<KeyValuePair<TK, TV>>(new KeyValuePair<TK, TV>(key, value));
        filledBuckets++;
    }
    else
    {
        var current = hashArray[index];

        while (current != null && current.Next != null)
        {
            /*此处可以判断是重复修改,还是抛异常*/
            if (current.Data.Key.Equals(key)) throw new Exception("重复key");

            current = current.Next;
        }
        if (current.Data.Key.Equals(key)) throw new Exception("重复key");
        current.Next = new DLinkedNode<KeyValuePair<TK, TV>>(new KeyValuePair<TK, TV>(key, value));
    }

    Count++;
}

/// <summary>
/// 扩容
/// </summary>
private void Grow()
{
    if (filledBuckets >= hashArray.Length * 0.7)
    {
        filledBuckets = 0;

        var newBucketSize = hashArray.Length * 2;

        var biggerArray = new DLinkedNode<KeyValuePair<TK, TV>>[newBucketSize];

        for (var i = 0; i < hashArray.Length; i++)
        {
            var item = hashArray[i];

            if (item != null)
            {
                var current = item;

                while (current != null)
                {
                    var next = current.Next;

                    var newIndex = Math.Abs(current.Data.Key.GetHashCode()) % newBucketSize;

                    if (biggerArray[newIndex] == null)
                    {
                        filledBuckets++;
                        biggerArray[newIndex] = current;
                    }

                    var bItem = biggerArray[newIndex];
                    while(bItem.Next != null)
                        bItem = bItem.Next;
                    bItem.Next = current;

                    current = next;
                }
            }
        }

        hashArray = biggerArray;
    }
}


public void Clear()
{
    throw new NotImplementedException();
}

/// <summary>
/// 查找
/// </summary>
/// <param name="key"></param>
/// <returns></returns>
public bool ContainsKey(TK key)
{
    /*1.获取原始下标*/
    var index = Math.Abs(key.GetHashCode()) % hashArray.Length;

    /*2.为空即无*/
    if (hashArray[index] == null) return false;

    var current = hashArray[index];

    /*3.遍历链表*/
    while (current != null)
    {
        if (current.Data.Key.Equals(key)) return true;

        current = current.Next;
    }

    return false;
}

public void Remove(TK key)
{
    var index = Math.Abs(key.GetHashCode()) % hashArray.Length;

    if (hashArray[index] == null) throw new Exception("未找到key");

    var current = hashArray[index];

    /*查找待删除元素*/
    DLinkedNode<KeyValuePair<TK, TV>> item = null;
    while (current != null)
    {
        if (current.Data.Key.Equals(key))
        {
            item = current;
            break;
        }
        current = current.Next;
    }

    if (item == null)
    {
        throw new Exception("未找到key");
    }

    /*删除*/
    if (item.Next == null)
        item = null;
    else
    {
        item.Previous = item.Next; 
        item.Next.Previous =item.Previous ;
        item = null;
    }


    if (hashArray[index] == null)
    {
        filledBuckets--;
    }

    Count--;

    Shrink();
}
private void Shrink()
{
    /*是否减容*/
    if (Math.Abs(filledBuckets - hashArray.Length * 0.3) < 0.1 && hashArray.Length / 2 > 0)
    {
        filledBuckets = 0;
        var newBucketSize = hashArray.Length / 2;

        var smallerArray = new DLinkedNode<KeyValuePair<TK, TV>>[newBucketSize];

        for (var i = 0; i < hashArray.Length; i++)
        {
            var item = hashArray[i];

            if (item != null)
            {
                var current = item;

                /*找到新的存储点*/
                while (current != null)
                {
                    var next = current.Next;

                    var newIndex = Math.Abs(current.Data.Key.GetHashCode()) % newBucketSize;

                    if (smallerArray[newIndex] == null)
                    {
                        filledBuckets++;
                        smallerArray[newIndex] = current;
                    }

                    var newItem = smallerArray[newIndex];
                    while(newItem.Next != null)
                        newItem = newItem.Next;
                    newItem.Next = current;

                    current = next;
                }
            }
        }

        hashArray = smallerArray;
    }
}
public IEnumerator<KeyValuePair<TK, TV>> GetEnumerator()
{
    throw new NotImplementedException();
}

IEnumerator IEnumerable.GetEnumerator()
{
    throw new NotImplementedException();
}
}

Dictionary源码分析

模拟实现:一个Dictionary,存储数据{1,’a’},{‘4′,’b’},{5,’c’}

1. 创建一个单链表,用来存储K-V

private struct Entry
{
    public uint hashCode;
    //值为-1,表示是该链条最后一个节点
    //值小于-1,表示已经被删除的自由节点
    public int next;
    public TKey key;     // Key of entry
    public TValue value; // Value of entry
}

2. 创建一个数组当桶,还有一个链表数组(核心就这两个数组)

private int[]? _buckets;
private Entry[]? _entries;

3. 模拟实现插入{1,’a’},{‘4′,’b’},{5,’c’}

初始化
《.NET深入了解哈希表和Dictionary》

第一次插入{1,’a’}
《.NET深入了解哈希表和Dictionary》

第二次插入{‘4′,’b’}

《.NET深入了解哈希表和Dictionary》

第三次插入{5,’c’}
《.NET深入了解哈希表和Dictionary》

仔细看一下这三个数据的插入,及数据的变化,应该可以理解_buckets和_entries的关系

4.删除

上边再讲哈希表,包括我们自己实现的代码中,删除一个节点后,都要重新计算后边的位置。如何解决这个问题呢?我们可以使用Entry的next,来表示是否已经删除,小于0就表示是自由节点。

关于删除就这样几个变量:

private int _freeList;//最后一个删除的Entry下标
private int _freeCount;//当前已删除,但是还未重新使用的节点数量
private const int StartOfFreeList = -3;//帮助寻找自由节点的一个常量

看一下StartOfFreeList和_freeList和Entry.next如何寻找自由节点

  • 删除时:Entry[i].next=上一层中的StartOfFreeList-_freeList
  • 添加&&_freeCount>0:_freeList=StartOfFreeList – entries[_freeList].next

请看图理解:

《.NET深入了解哈希表和Dictionary》

源码:简化版(debug理解)

源码:简化版可直接运行

public static void Main(string[] args)
{
    Dictionary<int, char> dic = new Dictionary<int, char>();
    dic.TryInsert(1, 'a');
    dic.TryInsert(4, 'b');
    dic.TryInsert(5, 'c');
    dic.Remove(4);
    dic.Remove(5);
    dic.TryInsert(0, 'd');
    dic.TryInsert(1, 'e');
}
public class Dictionary<TKey, TValue>
{
private int[]? _buckets;
private Entry[]? _entries;
private int _count;
private int _freeList;
private int _freeCount;
private int _version;
private const int StartOfFreeList = -3;
public Dictionary()
{
    /*初始值为素数,这里就不动态了,获取素数可以使用埃及筛选法*/
    Initialize(7);
}
private int Initialize(int capacity)
{
    int size = capacity;
    int[] buckets = new int[size];
    Entry[] entries = new Entry[size];
    _freeList = -1;
    _buckets = buckets;
    _entries = entries;
    return size;
}

public bool TryInsert(TKey key, TValue value)
{
    Entry[]? entries = _entries;
    uint hashCode = (uint)key.GetHashCode();

    uint collisionCount = 0;
    ref int bucket = ref GetBucket(hashCode);
    int i = bucket - 1; // Value in _buckets is 1-based
    if (typeof(TKey).IsValueType)
    {
        while (true)
        {
            if ((uint)i >= (uint)entries.Length)
            {
                break;
            }

            if (entries[i].hashCode == hashCode && EqualityComparer<TKey>.Default.Equals(entries[i].key, key))
            {
                entries[i].value = value;
                return true;
            }

            i = entries[i].next;

            collisionCount++;
            if (collisionCount > (uint)entries.Length)
            {
                throw new Exception("");
            }
        }
    }
    int index;
    if (_freeCount > 0)
    {
        index = _freeList;
        // Debug.Assert((StartOfFreeList - entries[_freeList].next) >= -1, "shouldn't overflow because `next` cannot underflow");
        _freeList = StartOfFreeList - entries[_freeList].next;
        _freeCount--;
    }
    else
    {
        int count = _count;
        if (count == entries.Length)
        {
            //Resize();
            bucket = ref GetBucket(hashCode);
        }
        index = count;
        _count = count + 1;
        entries = _entries;
    }

    ref Entry entry = ref entries![index];
    entry.hashCode = hashCode;
    entry.next = bucket - 1; // Value in _buckets is 1-based
    entry.key = key;
    entry.value = value; // Value in _buckets is 1-based
    bucket = index + 1;
    _version++;
    return true;
}
public bool Remove(TKey key)
{
    if (key == null) return false;

    if (_buckets != null)
    {
        uint collisionCount = 0;
        uint hashCode = (uint)key.GetHashCode();
        ref int bucket = ref GetBucket(hashCode);
        Entry[]? entries = _entries;
        int last = -1;
        int i = bucket - 1; // Value in buckets is 1-based
        while (i >= 0)
        {
            ref Entry entry = ref entries[i];

            if (entry.hashCode == hashCode && EqualityComparer<TKey>.Default.Equals(entry.key, key))
            {
                if (last < 0)
                {
                    bucket = entry.next + 1; 
                }
                else
                {
                    entries[last].next = entry.next;
                }

                entry.next = StartOfFreeList - _freeList;

                entry.key = default!;

                entry.value = default!;

                _freeList = i;
                _freeCount++;
                return true;
            }
            last = i;
            i = entry.next;
            collisionCount++;
            if (collisionCount > (uint)entries.Length)
            {

            }
        }
    }
    return false;
}
private ref int GetBucket(uint hashCode)
{
    int[] buckets = _buckets!;
    return ref buckets[hashCode % (uint)buckets.Length];
}
private struct Entry
{
    public uint hashCode;
    //值为-1,表示是该链条最后一个节点
    public int next;
    public TKey key;     // Key of entry
    public TValue value; // Value of entry
}
    原文作者:Broder
    原文地址: https://www.cnblogs.com/Z7TS/p/16894154.html
    本文转自网络文章,转载此文章仅为分享知识,如有侵权,请联系博主进行删除。
点赞