好吧 老师作业 晚交好久

 

分析一个文本（英文文章）（300k—500k）中的词出现的频率，并且把频率最高的10个词打印出来

 

 

 

 

import java.io.*;

public class  Account_num

{

    public static void main(String args[])

    {

        int member[][],max=0,mid=0; //member记录前10个单词的位置

        int i,j,k,m,n;

        int all=0,num[],num1=0,num2[],b=0,a=0,tmp11,tmp22;//num记录每层上单词的个数

        member=new int[10][3];

        char tmp1,tmp2,temp[]; //tmp用于从文件读取的中介，temp用于暂时记录每个单词

        temp=new char[20];

        for(i=0;i<20;i++)

        temp[i]='\0';

        String words[][][];//words用于记录所有的单词；

        try{

        FileInputStream f1=new FileInputStream("E:/Software_eng/a.txt");

        tmp11=f1.read();

        tmp22=f1.read();

        while(tmp11!=-1)      //第一遍读取，用于计算出总单词的个数

        {tmp1=(char)tmp11;

        tmp2=(char)tmp22;

        ///////////////

            if(((tmp1>='a')&&(tmp1<='z'))||((tmp1>='A')&&(tmp1<='Z')))

            {

                mid++;

                if((((tmp1>='a')&&(tmp1<='z'))||((tmp1>='A')&&(tmp1<='Z')))&&!(((tmp2>='a')&&(tmp2<='z'))||((tmp2>='A')&&(tmp2<='Z'))))

                {

                    all++;

                    if(mid>max)  

                    {

                        max=mid;

                    }   

                mid=0;

                }

                  

            }

            tmp11=tmp22;

            tmp22=f1.read();

        }

        f1.close();}

        catch(IOException ie){

              System.out.println(ie);

              }

        num=new int[max];

        num2=new int[max];

        for(i=0;i<max;i++)

        {

            num2[i]=0;

            num[i]=0;

        }

        words=new String[max][][];

        try{

            FileInputStream f2=new FileInputStream("E:/Software_eng/a.txt");

            tmp11=f2.read();

            tmp22=f2.read();

            mid=0;

        while(tmp11!=-1)                      //第二次读取用于计算每一层上的单词的个数

        {

            tmp1=(char)tmp11;

            tmp2=(char)tmp22;

            if(((tmp1>='a')&&(tmp1<='z'))||((tmp1>='A')&&(tmp1<='Z')))

            {

                    mid++;                //用于记录中间每个单词的长度

            }

            if((((tmp1>='a')&&(tmp1<='z'))||((tmp1>='A')&&(tmp1<='Z')))&&!(((tmp2>='a')&&(tmp2<='z'))||((tmp2>='A')&&(tmp2<='Z'))))

            {

                 num[mid-1]++;  //相应的记录每层的长度的一维数组加一

                 mid=0;     //一个单词结束，记录长度的数置0

            }

            tmp11=tmp22;

            tmp22=f2.read();    

        }

        f2.close();

            }

        catch(IOException ie){

              System.out.println(ie);

              }

        for(i=0;i<max;i++)

        {   num1=num[i];

            num2[i]=num[i];               //num2数组用于记录每层上的数组的个数，以便于后面的使用

            words[i]=new String[num1][2];//对每一层上的二维数组进行空间的申请

        }

        for(i=0;i<max;i++)

        {

            for(j=0;j<num[i];j++)

            {

                words[i][j][0]="1";   //用于记录该单词的个数， 

                words[i][j][1]=null;

            }

        }

        try{

        FileInputStream f3=new FileInputStream("E:/Software_eng/a.txt");

        tmp11=f3.read();

        tmp22=f3.read();

        mid=0; 

            //第三次读取用于吧所有的单词存储到申请好的空间里

        while(tmp11!=-1)         

        {tmp1=(char)tmp11;

        tmp2=(char)tmp22;

            if((((tmp1>='a')&&(tmp1<='z'))||((tmp1>='A')&&(tmp1<='Z'))))

            {

                    temp[mid]=tmp1;    //用于暂时保存读到的单词

                    mid++;

            }

            if((((tmp1>='a')&&(tmp1<='z'))||((tmp1>='A')&&(tmp1<='Z')))&&!(((tmp2>='a')&&(tmp2<='z'))||((tmp2>='A')&&(tmp2<='Z'))))

            {

                num1=num[mid-1]-1;     //

                num[mid-1]=num1;

                words[mid-1][num1][1]=new String(temp);//将读到的单词放入相应的层地相应位置

                words[mid-1][num1][1]=words[mid-1][num1][1].toLowerCase();//将单词转化为小写，便于后面的匹配

                for(;mid>0;mid--)

                    temp[mid-1]='\0';//将数组重新置为空

                 mid=0;     

            }

            tmp11=tmp22;

            tmp22=f3.read();    

        }

        f3.close();

        }

        catch(IOException ie){

      System.out.println(ie);

      }

        /////////

      for(i=0;i<10;i++)

      {

        for(j=0;j<3;j++)

        {member[i][j]=0;//将记录出现次数最多单词位置的数组全置为0，初始化。

        }

      }

      int mid1=0;

        for(i=0;i<max;i++)

        {

            for(k=0;k<num2[i];k++)

            {mid=0;

            if(!(words[i][k][0]=="0"))

            {       

                for(j=k;j<num2[i];j++)

                {

                    if(words[i][k][1].compareTo(words[i][j][1])==0)

                    {   

                        mid++;  

                        if(mid!=1)                //如果遇到相等的单词就个数加一

                        words[i][j][0]=""+0+"";

                    }

                }

                words[i][k][0]=""+mid+"";

                if(mid>member[9][2])          //如果当前单词的出现次数大于已储存的

                {                             //最小的单词次数，就提换掉最小的，

                    member[9][0]=i;           //在此声明member数组是用于存储当前出现次数

                    member[9][1]=k;            //最多的单词在三维数组中的位置。

                    member[9][2]=mid;          //顺续是按从大到小的顺序排的

                    for(n=0;n<9;n++)           //然后是循环找到当前出现次数在已记录的

                    {                          //次数中的位置

                        if(member[9-n][2]>member[9-n-1][2])

                        {

                            m=member[9-n][0];

                            member[9-n][0]=member[9-n-1][0];

                            member[9-n-1][0]=m;

                            m=member[9-n][1];

                            member[9-n][1]=member[9-n-1][1];

                            member[9-n-1][1]=m;

                            m=member[9-n][2];

                            member[9-n][2]=member[9-n-1][2];

                            member[9-n-1][2]=m;

                        }

                    }

                }

                ///////////////////////////

            }

        }

        }

        System.out.println(member[0][2]);

        System.out.println("出现频率最高的10个单词为：");

        for(i=0;i<10;i++)

        {   

            a=member[i][0];                       //根据记录的位置输出单词

            b=member[i][1];

            System.out.println(a);

            System.out.println(b);

            System.out.println(words[a][b][1]);

        }

    }

}