12.18

实验5

MapReduce初级编程实践

 

1.实验目的

1通过实验掌握基本的MapReduce编程方法;

2掌握用MapReduce解决一些常见的数据处理问题,包括数据去重、数据排序和数据挖掘等。

2.实验平台

1操作系统:Linux(建议Ubuntu16.04Ubuntu18.04

2Hadoop版本:3.1.3

3.实验步骤

(一)编程实现文件合并和去重操作

对于两个输入文件,即文件A和文件B,请编写MapReduce程序,对两个文件进行合并,并剔除其中重复的内容,得到一个新的输出文件C。下面是输入文件和输出文件的一个样例供参考。

输入文件A的样例如下:

 

20170101     x

20170102     y

20170103     x

20170104     y

20170105     z

20170106     x

 

输入文件B的样例如下:

20170101      y

20170102      y

20170103      x

20170104      z

20170105      y

 

根据输入文件AB合并得到的输出文件C的样例如下:

20170101      x

20170101      y

20170102      y

20170103      x

20170104      y

20170104      z

20170105      y

20170105      z

20170106      x

 

代码:

package org.example.five;

import java.io.IOException;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.GenericOptionsParser;
public class Merge {
    /**
     * @paramargs
     * A,B 两个文件进行合并,并剔除其中重复的内容,得到一个新的输出文件 C
     */
//重载 map 函数,直接将输入中的 value 复制到输出数据的 key
    public static class Map extends Mapper<Object, Text, Text, Text>{
        private static Text text = new Text();
        public void map(Object key, Text value, Context context) throws
                IOException,InterruptedException{
            text = value;
            context.write(text, new Text(""));
        }
    }
    //重载 reduce 函数,直接将输入中的 key 复制到输出数据的 key
    public static class Reduce extends Reducer<Text, Text, Text, Text>{
        public void reduce(Text key, Iterable<Text> values, Context context )
                throws IOException,InterruptedException{
            context.write(key, new Text(""));
        }
    }
    public static void main(String[] args) throws Exception{
// TODO Auto-generated method stub
        Configuration conf = new Configuration();
        conf.set("fs.default.name","hdfs://192.168.91.128:9000");
        String[] otherArgs = new String[]{"input","output"}; /* 直接设置输入参
*/
        if (otherArgs.length != 2) {
            System.err.println("Usage: wordcount <in> <out>");
            System.exit(2);
        }
        Job job = Job.getInstance(conf,"Merge and duplicate removal");
        job.setJarByClass(Merge.class);
        job.setMapperClass(Map.class);
        job.setCombinerClass(Reduce.class);
        job.setReducerClass(Reduce.class);
        job.setOutputKeyClass(Text.class);
        job.setOutputValueClass(Text.class);
        FileInputFormat.addInputPath(job, new Path(otherArgs[0]));
        FileOutputFormat.setOutputPath(job, new Path(otherArgs[1]));
        System.exit(job.waitForCompletion(true) ? 0 : 1);
    }
}

 

 

(二)编写程序实现对输入文件的排序

现在有多个输入文件,每个文件中的每行内容均为一个整数。要求读取所有文件中的整数,进行升序排序后,输出到一个新的文件中,输出的数据格式为每行两个整数,第一个数字为第二个整数的排序位次,第二个整数为原待排列的整数。下面是输入文件和输出文件的一个样例供参考。

输入文件1的样例如下:

33

37

12

40

 

输入文件2的样例如下:

4

16

39

5

 

输入文件3的样例如下:

1

45

25

 

根据输入文件123得到的输出文件如下:

1 1

2 4

3 5

4 12

5 16

6 25

7 33

8 37

9 39

10 40

11 45

 

package org.example.five;

import java.io.IOException;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Partitioner;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.GenericOptionsParser;

public class MergeSort {
    /**
     * @paramargs 输入多个文件,每个文件中的每行内容均为一个整数
     * 输出到一个新的文件中,输出的数据格式为每行两个整数,第一个数字为第二个整
     * 数的排序位次,第二个整数为原待排列的整数
     */
    //map 函数读取输入中的 value,将其转化成 IntWritable 类型,最后作为输出 key
    public static class Map extends Mapper<Object, Text, IntWritable,
            IntWritable> {
        private static IntWritable data = new IntWritable();

        public void map(Object key, Text value, Context context) throws
                IOException, InterruptedException {
            String text = value.toString();
            data.set(Integer.parseInt(text));
            context.write(data, new IntWritable(1));
        }
    }

    //reduce 函数将 map 输入的 key 复制到输出的 value 上,然后根据输入的 value-list中元素的个数决定 key 的输出次数,定义一个全局变量 line_num 来代表 key 的位次
    public static class Reduce extends Reducer<IntWritable, IntWritable,
            IntWritable, IntWritable> {
        private static IntWritable line_num = new IntWritable(1);

        public void reduce(IntWritable key, Iterable<IntWritable> values,
                           Context context) throws IOException, InterruptedException {
            for (IntWritable val : values) {
                context.write(line_num, key);
                line_num = new IntWritable(line_num.get() + 1);
            }
        }
    }

    //自定义 Partition 函数,此函数根据输入数据的最大值和 MapReduce 框架中Partition 的数量获取将输入数据按照大小分块的边界,然后根据输入数值和边界的关系返回对应的 Partiton ID
    public static class Partition extends Partitioner<IntWritable,
            IntWritable> {
        public int getPartition(IntWritable key, IntWritable value, int
                num_Partition) {
            int Maxnumber = 65223;//int 型的最大数值
            int bound = Maxnumber / num_Partition + 1;
            int keynumber = key.get();
            for (int i = 0; i < num_Partition; i++) {
                if (keynumber < bound * (i + 1) && keynumber >= bound * i) {
                    return i;
                }
            }
            return -1;
        }
    }

    public static void main(String[] args) throws Exception {
        // TODO Auto-generated method stub
        Configuration conf = new Configuration();
        conf.set("fs.default.name", "hdfs://master:9000");
        String[] otherArgs = new String[]{"input", "output"}; /* 直接设置输入参  */
        if (otherArgs.length != 2) {
            System.err.println("Usage: wordcount <in> <out>");
            System.exit(2);
        }
        Job job = Job.getInstance(conf, "Merge and sort");
        job.setJarByClass(MergeSort.class);
        job.setMapperClass(Map.class);
        job.setReducerClass(Reduce.class);
        job.setPartitionerClass(Partition.class);
        job.setOutputKeyClass(IntWritable.class);
        job.setOutputValueClass(IntWritable.class);
        FileInputFormat.addInputPath(job, new Path(otherArgs[0]));
        FileOutputFormat.setOutputPath(job, new Path(otherArgs[1]));
        System.exit(job.waitForCompletion(true) ? 0 : 1);
    }
}

①上传文件到hdfs

 

 

②运行代码

 

 

 

 

(三)对给定的表格进行信息挖掘

下面给出一个child-parent的表格,要求挖掘其中的父子辈关系,给出祖孙辈关系的表格。

输入文件内容如下:

child          parent

Steven        Lucy

Steven        Jack

Jone         Lucy

Jone         Jack

Lucy         Mary

Lucy         Frank

Jack         Alice

Jack         Jesse

David       Alice

David       Jesse

Philip       David

Philip       Alma

Mark       David

Mark       Alma

 

输出文件内容如下:

grandchild       grandparent

Steven          Alice

Steven          Jesse

Jone            Alice

Jone            Jesse

Steven          Mary

Steven          Frank

Jone            Mary

Jone            Frank

Philip           Alice

Philip           Jesse

Mark           Alice

Mark           Jesse

package org.example.five;

import java.io.IOException;
import java.util.*;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.GenericOptionsParser;

public class simple_data_mining {
    public static int time = 0;

    /**
     * @paramargs 输入一个 child-parent 的表格
     * 输出一个体现 grandchild-grandparent 关系的表格
     */
//Map 将输入文件按照空格分割成 child parent,然后正序输出一次作为右表,反序输出一次作为左表,需要注意的是在输出的 value 中必须加上左右表区别标志
    public static class Map extends Mapper<Object, Text, Text, Text> {
        public void map(Object key, Text value, Context context) throws
                IOException, InterruptedException {
            String child_name = new String();
            String parent_name = new String();
            String relation_type = new String();
            String line = value.toString();
            int i = 0;
            while (line.charAt(i) != ' ') {
                i++;
            }
            String[] values = {line.substring(0, i), line.substring(i + 1)};
            if (values[0].compareTo("child") != 0) {
                child_name = values[0];
                parent_name = values[1];
                relation_type = "1";//左右表区分标志
                context.write(new Text(values[1]), new
                        Text(relation_type + "+" + child_name + "+" + parent_name));
//左表
                relation_type = "2";
                context.write(new Text(values[0]), new
                        Text(relation_type + "+" + child_name + "+" + parent_name));
//右表
            }
        }
    }

    public static class Reduce extends Reducer<Text, Text, Text, Text> {
        public void reduce(Text key, Iterable<Text> values, Context context)
                throws IOException, InterruptedException {
            if (time == 0) { //输出表头
                context.write(new Text("grand_child"), new
                        Text("grand_parent"));
                time++;
            }
            int grand_child_num = 0;
            String grand_child[] = new String[10];
            int grand_parent_num = 0;
            String grand_parent[] = new String[10];
            Iterator ite = values.iterator();
            while (ite.hasNext()) {
                String record = ite.next().toString();
                int len = record.length();
                int i = 2;
                if (len == 0) continue;
                char relation_type = record.charAt(0);
                String child_name = new String();
                String parent_name = new String();
//获取 value-list value child
                while (record.charAt(i) != '+') {
                    child_name = child_name + record.charAt(i);
                    i++;
                }
                i = i + 1;
//获取 value-list value parent
                while (i < len) {
                    parent_name = parent_name + record.charAt(i);
                    i++;
                }
//左表,取出 child 放入 grand_child
                if (relation_type == '1') {
                    grand_child[grand_child_num] = child_name;
                    grand_child_num++;
                } else {//右表,取出 parent 放入 grand_parent
                    grand_parent[grand_parent_num] = parent_name;
                    grand_parent_num++;
                }
            }
            if (grand_parent_num != 0 && grand_child_num != 0) {
                for (int m = 0; m < grand_child_num; m++) {
                    for (int n = 0; n < grand_parent_num; n++) {
                        context.write(new Text(grand_child[m]), new
                                Text(grand_parent[n]));
//输出结果
                    }
                }
            }
        }
    }

    public static void main(String[] args) throws Exception {
// TODO Auto-generated method stub
        Configuration conf = new Configuration();
        conf.set("fs.default.name", "hdfs://master:9000");
        String[] otherArgs = new String[]{"input", "output"}; /* 直接设置输入参
*/
        if (otherArgs.length != 2) {
            System.err.println("Usage: wordcount <in> <out>");
            System.exit(2);
        }
        Job job = Job.getInstance(conf, "Single table join ");
        job.setJarByClass(simple_data_mining.class);
        job.setMapperClass(Map.class);
        job.setReducerClass(Reduce.class);
        job.setOutputKeyClass(Text.class);
        job.setOutputValueClass(Text.class);
        FileInputFormat.addInputPath(job, new Path(otherArgs[0]));
        FileOutputFormat.setOutputPath(job, new Path(otherArgs[1]));
        System.exit(job.waitForCompletion(true) ? 0 : 1);
    }
}

①上传文件

 

 

②代码运行

4.实验报告

题目:

MapReduce初级编程实践

姓名

陈庆振

日期 12.2

  • 实验环境:操作系统:Linux(Ubuntu16.04)
  • Hadoop版本:2.7.2

实验内容与完成情况:

文件合并和去重操作:

编写MapReduce程序,对两个输入文件A和B进行合并,并剔除重复内容,生成新的输出文件C。实验成功,合并去重功能正常。

输入文件排序:

编写MapReduce程序,读取多个输入文件中的整数,进行升序排序,并输出排序位次和原整数。实验成功,排序功能正常。

信息挖掘:

编写MapReduce程序,挖掘给定的child-parent表格中的父子辈关系,输出祖孙辈关系的表格。实验成功,信息挖掘功能正常。

出现的问题:在配置Hadoop运行环境时,遇到了配置文件路径错误的问题,导致无法正确运行MapReduce任务。

在实现排序功能时,由于MapReduce框架的默认行为,遇到了数据分布不均匀的问题。

解决方案(列出遇到的问题和解决办法,列出没有解决的问题):对于配置文件路径错误,通过检查和修正Hadoop配置文件中的路径设置,确保所有路径正确无误,问题得以解决。

对于数据分布不均匀的问题,通过调整Partitioner类,使得数据能够更均匀地分布到各个Reducer上,从而解决了问题。

posted @ 2024-12-18 18:21  七安。  阅读(29)  评论(0编辑  收藏  举报