/* 
为了便于管理，先引入个基础类： 
*/ 
package algorithms; 

/** 
* @author yovn 
* 
*/ 
public abstract class Sorter<E extends Comparable<E>> { 
         
        public abstract void sort(E[] array,int from ,int len); 
         
        public final void sort(E[] array) 
        { 
                sort(array,0,array.length); 
        } 
        protected final void swap(E[] array,int from ,int to) 
        { 
                E tmp=array[from]; 
                array[from]=array[to]; 
                array[to]=tmp; 
        } 

} 
/*一 插入排序 
该算法在数据规模小的时候十分高效，该算法每次插入第K+1到前K个有序数组中一个合适位置，K从0开始到N-1,从而完成排序： 
*/ 
// 
/** 
* @author yovn 
*/ 
public class InsertSorter<E extends Comparable<E>> extends Sorter<E> { 

        /* (non-Javadoc) 
         * @see algorithms.Sorter#sort(E[], int, int) 
         */ 
        public void sort(E[] array, int from, int len) { 
                 E tmp=null; 
                    for(int i=from+1;i<from+len;i++) 
                    { 
                            tmp=array[i]; 
                            int j=i; 
                            for(;j>from;j--) 
                            { 
                                    if(tmp.compareTo(array[j-1])<0) 
                                    { 
                                            array[j]=array[j-1]; 
                                    } 
                                    else break; 
                            } 
                            array[j]=tmp; 
                    } 
        } 
                 
         

} 
/* 
二 冒泡排序 
这可能是最简单的排序算法了，算法思想是每次从数组末端开始比较相邻两元素，把第i小的冒泡到数组的第i个位置。i从0一直到N-1从而完成排序。（当然也可以从数组开始端开始比较相邻两元素，把第i大的冒泡到数组的第N-i个位置。i从0一直到N-1从而完成排序。) 

*/ 

/** 
* @author yovn 
* 
*/ 
public class BubbleSorter<E extends Comparable<E>> extends Sorter<E> { 

        private static    boolean DWON=true; 
         
        public final void bubble_down(E[] array, int from, int len) 
        { 
                for(int i=from;i<from+len;i++) 
                { 
                        for(int j=from+len-1;j>i;j--) 
                        { 
                                if(array[j].compareTo(array[j-1])<0) 
                                { 
                                        swap(array,j-1,j); 
                                } 
                        } 
                } 
        } 
         
        public final void bubble_up(E[] array, int from, int len) 
        { 
                for(int i=from+len-1;i>=from;i--) 
                { 
                        for(int j=from;j<i;j++) 
                        { 
                                if(array[j].compareTo(array[j+1])>0) 
                                { 
                                        swap(array,j,j+1); 
                                } 
                        } 
                } 
        } 
         
        public void sort(E[] array, int from, int len) { 
                 
                if(DWON) 
                { 
                        bubble_down(array,from,len); 
                } 
                else 
                { 
                        bubble_up(array,from,len); 
                } 
        } 
         
} 
/* 
三，选择排序 
选择排序相对于冒泡来说，它不是每次发现逆序都交换，而是在找到全局第i小的时候记下该元素位置，最后跟第i个元素交换，从而保证数组最终的有序。 
相对与插入排序来说，选择排序每次选出的都是全局第i小的，不会调整前i个元素了。 
*/ 
/** 
* @author yovn 
* 
*/ 
public class SelectSorter<E extends Comparable<E>> extends Sorter<E> { 

        /* (non-Javadoc) 
         * @see algorithms.Sorter#sort(E[], int, int) 
         */ 
        @Override 
        public void sort(E[] array, int from, int len) { 
                for(int i=0;i<len;i++) 
                { 
                        int smallest=i; 
                        int j=i+from; 
                        for(;j<from+len;j++) 
                        { 
                                if(array[j].compareTo(array[smallest])<0) 
                                { 
                                        smallest=j; 
                                } 
                        } 
                        swap(array,i,smallest); 
                                        
                } 

        } 
     
} 
/* 
四 Shell排序 
Shell排序可以理解为插入排序的变种，它充分利用了插入排序的两个特点： 
1）当数据规模小的时候非常高效 
2）当给定数据已经有序时的时间代价为O(N) 
所以，Shell排序每次把数据分成若个小块，来使用插入排序，而且之后在这若个小块排好序的情况下把它们合成大一点的小块，继续使用插入排序，不停的合并小块，知道最后成一个块，并使用插入排序。 

这里每次分成若干小块是通过“增量” 来控制的，开始时增量交大，接近N/2,从而使得分割出来接近N/2个小块，逐渐的减小“增量“最终到减小到1。 

一直较好的增量序列是2^k-1,2^(k-1)-1,.....7,3,1,这样可使Shell排序时间复杂度达到O(N^1.5) 
所以我在实现Shell排序的时候采用该增量序列 
*/ 

/** 
* @author yovn 
*/ 
public class ShellSorter<E extends Comparable<E>> extends Sorter<E>    { 

        /* (non-Javadoc) 
         * Our delta value choose 2^k-1,2^(k-1)-1,.7,3,1. 
         * complexity is O(n^1.5) 
         * @see algorithms.Sorter#sort(E[], int, int) 
         */ 
        @Override 
        public void sort(E[] array, int from, int len) { 
                 
                //1.calculate    the first delta value; 
                int value=1; 
                while((value+1)*2<len) 
                { 
                        value=(value+1)*2-1; 
                 
                } 
         
                for(int delta=value;delta>=1;delta=(delta+1)/2-1) 
                { 
                        for(int i=0;i<delta;i++) 
                        { 
                                modify_insert_sort(array,from+i,len-i,delta); 
                        } 
                } 

        } 
         
        private final    void modify_insert_sort(E[] array, int from, int len,int delta) { 
                    if(len<=1)return; 
                    E tmp=null; 
                    for(int i=from+delta;i<from+len;i+=delta) 
                    { 
                            tmp=array[i]; 
                            int j=i; 
                            for(;j>from;j-=delta) 
                            { 
                                    if(tmp.compareTo(array[j-delta])<0) 
                                    { 
                                            array[j]=array[j-delta]; 
                                    } 
                                    else break; 
                            } 
                            array[j]=tmp; 
                    } 

        } 
} 
/* 
五 快速排序 
快速排序是目前使用可能最广泛的排序算法了。 
一般分如下步骤： 
1）选择一个枢纽元素（有很对选法，我的实现里采用去中间元素的简单方法） 
2）使用该枢纽元素分割数组，使得比该元素小的元素在它的左边，比它大的在右边。并把枢纽元素放在合适的位置。 
3）根据枢纽元素最后确定的位置，把数组分成三部分，左边的，右边的，枢纽元素自己，对左边的，右边的分别递归调用快速排序算法即可。 
快速排序的核心在于分割算法，也可以说是最有技巧的部分。 
*/ 
/** 
* @author yovn 
* 
*/ 
public class QuickSorter<E extends Comparable<E>> extends Sorter<E> { 

        /* (non-Javadoc) 
         * @see algorithms.Sorter#sort(E[], int, int) 
         */ 
        @Override 
        public void sort(E[] array, int from, int len) { 
                q_sort(array,from,from+len-1); 
        } 

         
        private final void q_sort(E[] array, int from, int to) { 
                if(to-from<1)return; 
                int pivot=selectPivot(array,from,to); 

                 
                 
                pivot=partion(array,from,to,pivot); 
                 
                q_sort(array,from,pivot-1); 
                q_sort(array,pivot+1,to); 
                 
        } 


        private int partion(E[] array, int from, int to, int pivot) { 
                E tmp=array[pivot]; 
                array[pivot]=array[to];//now to's position is available 
                 
                while(from!=to) 
                { 
                        while(from<to&&array[from].compareTo(tmp)<=0)from++; 
                        if(from<to) 
                        { 
                                array[to]=array[from];//now from's position is available 
                                to--; 
                        } 
                        while(from<to&&array[to].compareTo(tmp)>=0)to--; 
                        if(from<to) 
                        { 
                                array[from]=array[to];//now to's position is available now    
                                from++; 
                        } 
                } 
                array[from]=tmp; 
                return from; 
        } 


        private int selectPivot(E[] array, int from, int to) { 
         
                return (from+to)/2; 
        } 

} 
/* 
六 归并排序 
算法思想是每次把待排序列分成两部分，分别对这两部分递归地用归并排序，完成后把这两个子部分合并成一个 
序列。 
归并排序借助一个全局性临时数组来方便对子序列的归并，该算法核心在于归并。 
*/ 

import java.lang.reflect.Array; 

/** 
* @author yovn 
* 
*/ 
public class MergeSorter<E extends Comparable<E>> extends Sorter<E>    { 

        /* (non-Javadoc) 
         * @see algorithms.Sorter#sort(E[], int, int) 
         */ 
        @SuppressWarnings("unchecked") 
        @Override 
        public void sort(E[] array, int from, int len) { 
                if(len<=1)return; 
                E[] temporary=(E[])Array.newInstance(array[0].getClass(),len); 
                merge_sort(array,from,from+len-1,temporary); 

        } 

        private final void merge_sort(E[] array, int from, int to, E[] temporary) { 
                if(to<=from) 
                { 
                        return; 
                } 
                int middle=(from+to)/2; 
                merge_sort(array,from,middle,temporary); 
                merge_sort(array,middle+1,to,temporary); 
                merge(array,from,to,middle,temporary); 
        } 

        private final void merge(E[] array, int from, int to, int middle, E[] temporary) { 
                int k=0,leftIndex=0,rightIndex=to-from; 
                System.arraycopy(array, from, temporary, 0, middle-from+1); 
                for(int i=0;i<to-middle;i++) 
                { 
                        temporary[to-from-i]=array[middle+i+1]; 
                } 
                while(k<to-from+1) 
                { 
                        if(temporary[leftIndex].compareTo(temporary[rightIndex])<0) 
                        { 
                                array[k+from]=temporary[leftIndex++]; 
                                 
                        } 
                        else 
                        { 
                                array[k+from]=temporary[rightIndex--]; 
                        } 
                        k++; 
                } 
                 
        } 

} 
/*七 堆排序 
堆是一种完全二叉树，一般使用数组来实现。 
堆主要有两种核心操作， 
1）从指定节点向上调整(shiftUp) 
2）从指定节点向下调整(shiftDown) 
建堆，以及删除堆定节点使用shiftDwon,而在插入节点时一般结合两种操作一起使用。 
堆排序借助最大值堆来实现，第i次从堆顶移除最大值放到数组的倒数第i个位置，然后shiftDown到倒数第i+1个位置,一共执行N此调整，即完成排序。 
显然，堆排序也是一种选择性的排序，每次选择第i大的元素。 
*/ 

/** 
* @author yovn 
* 
*/ 
public class HeapSorter<E extends Comparable<E>> extends Sorter<E>    { 

        /* (non-Javadoc) 
         * @see algorithms.Sorter#sort(E[], int, int) 
         */ 
        @Override 
        public void sort(E[] array, int from, int len) { 
                build_heap(array,from,len); 

                for(int i=0;i<len;i++) 
                { 
                        //swap max value to the (len-i)-th position 
                        swap(array,from,from+len-1-i); 
                        shift_down(array,from,len-1-i,0);//always shiftDown from 0 
                } 
        } 

        private final void build_heap(E[] array, int from, int len) { 
                int pos=(len-1)/2;//we start from (len-1)/2, because branch's node +1=leaf's node, and all leaf node is already a heap 
                for(int i=pos;i>=0;i--) 
                { 
                        shift_down(array,from,len,i); 
                } 
                 
        } 
         
        private final void shift_down(E[] array,int from, int len, int pos) 
        { 
                 
                E tmp=array[from+pos]; 
                int index=pos*2+1;//use left child 
                while(index<len)//until no child 
                { 
                        if(index+1<len&&array[from+index].compareTo(array[from+index+1])<0)//right child is bigger 
                        { 
                                index+=1;//switch to right child 
                        } 
                        if(tmp.compareTo(array[from+index])<0) 
                        { 
                                array[from+pos]=array[from+index]; 
                                pos=index; 
                                index=pos*2+1; 
                                 
                        } 
                        else 
                        { 
                                break; 
                        } 
                         
                } 
                array[from+pos]=tmp; 
                         
        } 

         
} 
/* 
八 桶式排序 
桶式排序不再是基于比较的了，它和基数排序同属于分配类的排序，这类排序的特点是事先要知道待排序列的一些特征。 
桶式排序事先要知道待排序列在一个范围内，而且这个范围应该不是很大的。 
比如知道待排序列在[0,M）内，那么可以分配M个桶，第I个桶记录I的出现情况，最后根据每个桶收到的位置信息把数据输出成有序的形式。 
这里我们用两个临时性数组，一个用于记录位置信息，一个用于方便输出数据成有序方式，另外我们假设数据落在0到MAX,如果所给数据不是从0开始，你可以把每个数减去最小的数。 
    
*/ 
/** 
* @author yovn 
* 
*/ 
public class BucketSorter { 

         
         
        public void sort(int[] keys,int from,int len,int max) 
        { 
                int[] temp=new int[len]; 
                int[] count=new int[max]; 
                 
                 
                for(int i=0;i<len;i++) 
                { 
                        count[keys[from+i]]++; 
                } 
                //calculate position info 
                for(int i=1;i<max;i++) 
                { 
                        count[i]=count[i]+count[i-1];//this means how many number which is less or equals than i,thus it is also position + 1    
                } 
                 
                System.arraycopy(keys, from, temp, 0, len); 
                for(int k=len-1;k>=0;k--)//from the ending to beginning can keep the stability 
                { 
                        keys[--count[temp[k]]]=temp[k];// position +1 =count 
                } 
        } 
        /** 
         * @param args 
         */ 
        public static void main(String[] args) { 

                int[] a={1,4,8,3,2,9,5,0,7,6,9,10,9,13,14,15,11,12,17,16}; 
                BucketSorter sorter=new BucketSorter(); 
                sorter.sort(a,0,a.length,20);//actually is 18, but 20 will also work 
                 
                 
                for(int i=0;i<a.length;i++) 
                { 
                        System.out.print(a[i]+","); 
                } 

        } 

} 
/* 
九 基数排序 
基数排序可以说是扩展了的桶式排序，比如当待排序列在一个很大的范围内，比如0到999999内，那么用桶式排序是很浪费空间的。而基数排序把每个排序码拆成由d个排序码，比如任何一个6位数（不满六位前面补0）拆成6个排序码，分别是个位的，十位的，百位的。。。。 
排序时，分6次完成，每次按第i个排序码来排。 
一般有两种方式: 
1) 高位优先(MSD)： 从高位到低位依次对序列排序 
2）低位优先(LSD)： 从低位到高位依次对序列排序 
计算机一般采用低位优先法（人类一般使用高位优先），但是采用低位优先时要确保排序算法的稳定性。 
基数排序借助桶式排序，每次按第N位排序时，采用桶式排序。对于如何安排每次落入同一个桶中的数据有两种安排方法： 
1）顺序存储：每次使用桶式排序，放入r个桶中，，相同时增加计数。 
2）链式存储：每个桶通过一个静态队列来跟踪。 

*/ 

import java.util.Arrays; 


/** 
* @author yovn 
* 
*/ 
public class RadixSorter { 
         
        public static boolean USE_LINK=true; 
         
        /** 
         *    
         * @param keys 
         * @param from 
         * @param len 
         * @param radix    key's radix 
         * @param d            how many sub keys should one key divide to 
         */ 
        public void sort(int[] keys,int from ,int len,int radix, int d) 
        { 
                if(USE_LINK) 
                { 
                        link_radix_sort(keys,from,len,radix,d); 
                } 
                else 
                { 
                        array_radix_sort(keys,from,len,radix,d); 
                } 
                 
        } 
         
         
        private final void array_radix_sort(int[] keys, int from, int len, int radix, 
                        int d)    
        { 
                int[] temporary=new int[len]; 
                int[] count=new int[radix]; 
                int R=1; 
                 
                for(int i=0;i<d;i++) 
                { 
                        System.arraycopy(keys, from, temporary, 0, len); 
                        Arrays.fill(count, 0); 
                        for(int k=0;k<len;k++) 
                        { 
                                int subkey=(temporary[k]/R)%radix; 
                                count[subkey]++; 
                        } 
                        for(int j=1;j<radix;j++) 
                        { 
                                count[j]=count[j]+count[j-1]; 
                        } 
                        for(int m=len-1;m>=0;m--) 
                        { 
                                int subkey=(temporary[m]/R)%radix; 
                                --count[subkey]; 
                                keys[from+count[subkey]]=temporary[m]; 
                        } 
                        R*=radix; 
                } 
                        
        } 


        private static class LinkQueue 
        { 
                int head=-1; 
                int tail=-1; 
        } 
        private final void link_radix_sort(int[] keys, int from, int len, int radix, int d) { 
                 
                int[] nexts=new int[len]; 
                 
                LinkQueue[] queues=new LinkQueue[radix]; 
                for(int i=0;i<radix;i++) 
                { 
                        queues[i]=new LinkQueue(); 
                } 
                for(int i=0;i<len-1;i++) 
                { 
                        nexts[i]=i+1; 
                } 
                nexts[len-1]=-1; 
                 
                int first=0; 
                for(int i=0;i<d;i++) 
                { 
                        link_radix_sort_distribute(keys,from,len,radix,i,nexts,queues,first); 
                        first=link_radix_sort_collect(keys,from,len,radix,i,nexts,queues); 
                } 
                int[] tmps=new int[len]; 
                int k=0; 
                while(first!=-1) 
                { 
                 
                        tmps[k++]=keys[from+first]; 
                        first=nexts[first]; 
                } 
                System.arraycopy(tmps, 0, keys, from, len); 
                 
                 
        } 
        private final void link_radix_sort_distribute(int[] keys, int from, int len, 
                        int radix, int d, int[] nexts, LinkQueue[] queues,int first) { 
                 
                for(int i=0;i<radix;i++)queues[i].head=queues[i].tail=-1; 
                while(first!=-1) 
                { 
                        int val=keys[from+first]; 
                        for(int j=0;j<d;j++)val/=radix; 
                        val=val%radix; 
                        if(queues[val].head==-1) 
                        { 
                                queues[val].head=first; 
                        } 
                        else    
                        { 
                                nexts[queues[val].tail]=first; 
                                 
                        } 
                        queues[val].tail=first; 
                        first=nexts[first]; 
                } 
                 
        } 
        private int link_radix_sort_collect(int[] keys, int from, int len, 
                        int radix, int d, int[] nexts, LinkQueue[] queues) { 
                int first=0; 
                int last=0; 
                int fromQueue=0; 
                for(;(fromQueue<radix-1)&&(queues[fromQueue].head==-1);fromQueue++); 
                first=queues[fromQueue].head; 
                last=queues[fromQueue].tail; 
                 
                while(fromQueue<radix-1&&queues[fromQueue].head!=-1) 
                { 
                        fromQueue+=1; 
                        for(;(fromQueue<radix-1)&&(queues[fromQueue].head==-1);fromQueue++); 
                         
                        nexts[last]=queues[fromQueue].head; 
                        last=queues[fromQueue].tail; 
                         
                } 
                if(last!=-1)nexts[last]=-1; 
                return first; 
        } 
         
        /** 
         * @param args 
         */ 
        public static void main(String[] args) { 
                int[] a={1,4,8,3,2,9,5,0,7,6,9,10,9,135,14,15,11,222222222,1111111111,12,17,45,16}; 
                USE_LINK=true; 
                RadixSorter sorter=new RadixSorter(); 
                sorter.sort(a,0,a.length,10,10); 
                for(int i=0;i<a.length;i++) 
                { 
                        System.out.print(a[i]+","); 
                } 


        } 

}