常见关系代数运算包括:选择、投影、并、交、差以及自然连接操作等,都可以十分容易利用MapReduce框架进行并行化计算
NAME | SEX | AGE |
小明 | 男 | 25 |
小红 | 女 | 18 |
小张 | 男 | 22 |
小米 | 女 | 23 |
小丽 | 女 | 21 |
小王 | 男 | 19 |
小美 | 女 | 25 |
小朱 | 女 | 26 |
选择操作
将关系R的数据存储在relationR文件,然后移入HDFS下的data文件夹,如代码1-1
代码1-1
root@lejian:/data# cat relationR 小明 男 25 小红 女 18 小张 男 22 小米 女 23 小丽 女 21 小王 男 19 小美 女 25 小朱 女 26 root@lejian:/data# hadoop fs -put selection /data root@lejian:/data# hadoop fs -ls -R /data -rw-r--r-- 1 root supergroup 112 2017-01-07 15:03 /data/relationR
对于关系R的应用条件C,选择性别为女的数据,只需在Map阶段对每个输入的记录进行判断,将满足条件的数据输出即可,输出键值为(key,null)。Reduce阶段无需做额外的工作
代码1-2
<?xml version="1.0"?> <configuration> <property> <name>sex</name> <value>女</value> </property> </configuration>
代码1-3
package com.hadoop.mapreduce; public class Person { private String name; private String sex; private int age; public Person(String line) { super(); String[] lines = line.split(" "); this.name = lines[0]; this.sex = lines[1]; this.age = Integer.parseInt(lines[2]); } public String getName() { return name; } public String getSex() { return sex; } public int getAge() { return age; } public String getVal(String col) { if ("name".equals(col)) { return name; } if ("sex".equals(col)) { return sex; } return age + ""; } @Override public String toString() { return name + " " + sex + " " + age; } }
代码1-4
package com.hadoop.mapreduce; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Mapper; public class SelectionMap extends Mapper<LongWritable, Text, Text, NullWritable> { private String sex = ""; private Text val = new Text(); protected void setup(Context context) throws java.io.IOException, InterruptedException { Configuration conf = context.getConfiguration(); sex = conf.get("sex"); }; protected void map(LongWritable key, Text value, Context context) throws java.io.IOException, InterruptedException { Person person = new Person(value.toString()); if (sex.equals(person.getVal("sex"))) { val.set(person.toString()); context.write(val, NullWritable.get()); } }; }
代码1-5
package com.hadoop.mapreduce; import java.io.IOException; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.fs.Path; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; public class Selection { public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException { if (args == null || args.length != 2) { throw new RuntimeException("请输入输入路径、输出路径"); } Configuration conf = new Configuration(); conf.addResource("conf.xml"); Job job = Job.getInstance(conf); job.setJobName("Selection"); job.setMapperClass(SelectionMap.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(NullWritable.class); FileInputFormat.addInputPaths(job, args[0]); FileOutputFormat.setOutputPath(job, new Path(args[1])); System.exit(job.waitForCompletion(true) ? 0 : 1); } }
运行代码1-5,运行结果如代码1-6
代码1-6
root@lejian:/data# hadoop jar selection.jar com.hadoop.mapreduce.Selection /data /output ………… root@lejian:/data# hadoop fs -ls -R /output -rw-r--r-- 1 root supergroup 0 2017-01-07 15:05 /output/_SUCCESS -rw-r--r-- 1 root supergroup 70 2017-01-07 15:05 /output/part-r-00000 root@lejian:/data# hadoop fs -cat /output/part-r-00000 小丽 女 21 小朱 女 26 小米 女 23 小红 女 18 小美 女 25
投影操作
例如在关系R上应用投影操作获得属性AGE的所有值,我们只需要在Map阶段将每条记录的AGE属性和NullWritable输出,而Reduce端仅获取key即可,注意,此时投影操作具有去重功能
代码1-7
package com.hadoop.mapreduce; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Mapper; public class ProjectionMap extends Mapper<LongWritable, Text, IntWritable, NullWritable> { private IntWritable age = new IntWritable(); protected void map(LongWritable key, Text value, Context context) throws java.io.IOException, InterruptedException { Person person = new Person(value.toString()); age.set(person.getAge()); context.write(age, NullWritable.get()); }; }
代码1-8
package com.hadoop.mapreduce; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.mapreduce.Reducer; public class ProjectionReduce extends Reducer<IntWritable, NullWritable, IntWritable, NullWritable> { protected void reduce(IntWritable key, Iterable<NullWritable> values, Context context) throws java.io.IOException, InterruptedException { context.write(key, NullWritable.get()); }; }
代码1-9
package com.hadoop.mapreduce; 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.NullWritable; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; public class Projection { public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException { if (args == null || args.length != 2) { throw new RuntimeException("请输入输入路径、输出路径"); } Configuration conf = new Configuration(); Job job = Job.getInstance(conf); job.setJobName("Projection"); job.setMapperClass(ProjectionMap.class); job.setReducerClass(ProjectionReduce.class); job.setOutputKeyClass(IntWritable.class); job.setOutputValueClass(NullWritable.class); FileInputFormat.addInputPaths(job, args[0]); FileOutputFormat.setOutputPath(job, new Path(args[1])); System.exit(job.waitForCompletion(true) ? 0 : 1); } }
运行代码1-9,运行结果如代码1-10
代码1-10
root@lejian:/data# hadoop jar projection.jar com.hadoop.mapreduce.Projection /data /output ………… root@lejian:/data# hadoop fs -ls -R /output -rw-r--r-- 1 root supergroup 0 2017-01-07 15:52 /output/_SUCCESS -rw-r--r-- 1 root supergroup 21 2017-01-07 15:52 /output/part-r-00000 root@lejian:/data# hadoop fs -cat /output/part-r-00000 18 19 21 22 23 25 26
交运算
如果有一个关系A和关系B为同一个模式,希望得到关系A和关系B的交集,那么在Map阶段对于A和B中的每一条记录r输出(r,1),在Reduce阶段汇总计数,如果计数为2,则将该条记录输出。依旧以Person类为例,这里把Person作为主键,为了使得关系A和关系B相同的Person发送到同一个Reduce节点进行计算,需要对原先代码1-3的Person类进行修改,如代码1-11,MapReduce默认会先调用对象的compareTo方法进行对象间的比较,如果对象相等,再比较其hashCode,如果hashCode相等,则认为这两个对象为同一个对象
修改代码1-3的Person类为代码1-11
代码1-11
package com.hadoop.mapreduce; import java.io.DataInput; import java.io.DataOutput; import java.io.IOException; import org.apache.hadoop.io.WritableComparable; public class Person implements WritableComparable<Person> { private String name; private String sex; private int age; public Person() { super(); // TODO Auto-generated constructor stub } public Person(String line) { super(); String[] lines = line.split(" "); this.name = lines[0]; this.sex = lines[1]; this.age = Integer.parseInt(lines[2]); } public String getName() { return name; } public String getSex() { return sex; } public int getAge() { return age; } public String getVal(String col) { if ("name".equals(col)) { return name; } if ("sex".equals(col)) { return sex; } return age + ""; } @Override public String toString() { return name + " " + sex + " " + age; } @Override public int hashCode() { int res = 20; res = name.hashCode() + 10 * res; res = sex.hashCode() + 10 * res; res = age + 10 * res; return res; } @Override public void write(DataOutput out) throws IOException { out.writeUTF(name); out.writeUTF(sex); out.writeInt(age); } @Override public void readFields(DataInput in) throws IOException { name = in.readUTF(); sex = in.readUTF(); age = in.readInt(); } @Override public int compareTo(Person o) { // TODO Auto-generated method stub if (hashCode() > o.hashCode()) { return 1; } if (hashCode() < o.hashCode()) { return -1; } return 0; } public static void main(String[] args) { System.out.println(new Person("Lily female 22").hashCode()); } }
将关系A和关系B移入HDFS下的data文件夹,如代码1-12
root@lejian:/data# cat relationA Tom male 21 Amy female 19 Daivd male 16 Lily female 22 Lucy female 20 John male 19 Rose female 19 Jojo female 26 root@lejian:/data# cat relationB Daivd male 16 Jack male 15 Lily female 22 Lucy female 20 Tom male 25 root@lejian:/data# hadoop fs -put relation* /data root@lejian:/data# hadoop fs -ls -R /data -rw-r--r-- 1 root supergroup 113 2017-01-07 20:48 /data/relationA -rw-r--r-- 1 root supergroup 69 2017-01-07 20:48 /data/relationB
代码1-13
package com.hadoop.mapreduce; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Mapper; public class IntersectionMap extends Mapper<LongWritable, Text, Person, IntWritable> { private static final IntWritable ONE = new IntWritable(1); protected void map(LongWritable key, Text value, Context context) throws java.io.IOException, InterruptedException { Person person = new Person(value.toString()); context.write(person, ONE); }; }
代码1-14
package com.hadoop.mapreduce; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.mapreduce.Reducer; public class IntersectionReduce extends Reducer<Person, IntWritable, Person, NullWritable> { protected void reduce(Person key, Iterable<IntWritable> values, Context context) throws java.io.IOException, InterruptedException { int count = 0; for (IntWritable val : values) { count += val.get(); } if (count == 2) { context.write(key, NullWritable.get()); } }; }
代码1-15
package com.hadoop.mapreduce; 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.NullWritable; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; public class Intersection { public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException { if (args == null || args.length != 2) { throw new RuntimeException("请输入输入路径、输出路径"); } Configuration conf = new Configuration(); Job job = Job.getInstance(conf); job.setJobName("Intersection"); job.setJarByClass(Intersection.class); job.setMapperClass(IntersectionMap.class); job.setMapOutputKeyClass(Person.class); job.setMapOutputValueClass(IntWritable.class); job.setReducerClass(IntersectionReduce.class); job.setOutputKeyClass(Person.class); job.setOutputValueClass(NullWritable.class); FileInputFormat.addInputPaths(job, args[0]); FileOutputFormat.setOutputPath(job, new Path(args[1])); System.exit(job.waitForCompletion(true) ? 0 : 1); } }
运行代码1-15,运行结果如代码1-16
代码1-16
root@lejian:/data# hadoop jar intersection.jar com.hadoop.mapreduce.Intersection /data /output ………… root@lejian:/data# hadoop fs -ls -R /output -rw-r--r-- 1 root supergroup 0 2017-01-07 20:30 /output/_SUCCESS -rw-r--r-- 1 root supergroup 44 2017-01-07 20:30 /output/part-r-00000 root@lejian:/data# hadoop fs -cat /output/part-r-00000 Daivd male 12 Lily female 22 Lucy female 20
差运算
计算关系A-关系B的差集,即找出在关系A中存在而在关系B中不存在的记录,在Map阶段,对于关系A和关系B中每一条记录r输出键值对(r,A),(r,B),在Reduce阶段检查每一条记录r和其对应的关系名称,只有关系名称只存在A,才输出记录
先显示HDFS中data文件夹下得relationA和relationB的文件内容,如代码1-17
代码1-17
root@lejian:/data# hadoop fs -ls -R /data -rw-r--r-- 1 root supergroup 113 2017-01-07 20:48 /data/relationA -rw-r--r-- 1 root supergroup 69 2017-01-07 20:48 /data/relationB root@lejian:/data# hadoop fs -cat /data/relationA Tom male 21 Amy female 19 Daivd male 16 Lily female 22 Lucy female 20 John male 19 Rose female 19 Jojo female 26 root@lejian:/data# hadoop fs -cat /data/relationB Daivd male 16 Jack male 15 Lily female 22 Lucy female 20 Tom male 25
代码1-18
package com.hadoop.mapreduce; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Mapper; import org.apache.hadoop.mapreduce.lib.input.FileSplit; public class DifferenceMap extends Mapper<LongWritable, Text, Person, Text> { private Text relationName = new Text(); protected void setup(Context context) throws java.io.IOException, InterruptedException { FileSplit fileSplit = (FileSplit) context.getInputSplit(); relationName.set(fileSplit.getPath().getName()); }; protected void map(LongWritable key, Text value, Context context) throws java.io.IOException, InterruptedException { Person person = new Person(value.toString()); context.write(person, relationName); }; }
代码1-19
package com.hadoop.mapreduce; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Reducer; public class DifferenceReduce extends Reducer<Person, Text, Person, NullWritable> { private String remove = ""; protected void setup(Context context) throws java.io.IOException, InterruptedException { Configuration conf = context.getConfiguration(); remove = conf.get("remove"); }; protected void reduce(Person key, Iterable<Text> values, Context context) throws java.io.IOException, InterruptedException { for (Text val : values) { if (remove.equals(val.toString())) { return; } } context.write(key, NullWritable.get()); }; }
代码1-20
package com.hadoop.mapreduce; import java.io.IOException; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.fs.Path; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; public class Difference { public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException { if (args == null || args.length != 3) { throw new RuntimeException("请输入输入路径、输出路径和被减集合"); } Configuration conf = new Configuration(); conf.set("remove", args[2]); Job job = Job.getInstance(conf); job.setJobName("Difference"); job.setJarByClass(Difference.class); job.setMapperClass(DifferenceMap.class); job.setMapOutputKeyClass(Person.class); job.setMapOutputValueClass(Text.class); job.setReducerClass(DifferenceReduce.class); job.setOutputKeyClass(Person.class); job.setOutputValueClass(NullWritable.class); FileInputFormat.addInputPaths(job, args[0]); FileOutputFormat.setOutputPath(job, new Path(args[1])); System.exit(job.waitForCompletion(true) ? 0 : 1); } }
运行代码1-20,运行结果如代码1-21
代码1-21
root@lejian:/data# hadoop jar difference.jar com.hadoop.mapreduce.Difference /data /output relationB ………… root@lejian:/data# hadoop fs -ls -R /output -rw-r--r-- 1 root supergroup 0 2017-01-08 08:59 /output/_SUCCESS -rw-r--r-- 1 root supergroup 69 2017-01-08 08:59 /output/part-r-00000 root@lejian:/data# hadoop fs -cat /output/part-r-00000 Tom male 21 Amy female 19 John male 19 Jojo female 26 Rose female 19
自然连接
如代码1-22,student集合的第一列是id,第二列是姓名,第三列是性别,第四列是年龄,grade集合第一列是id,第二列是科目,第三列是科目成绩,需要对student集合和grade集合做自然连接。在Map阶段将student和grade中每一条记录r作为value,而记录中的id作为key输出。在Reduce阶段则将同一键收集而来的数据根据它们的来源(student或grade)做笛卡尔积然后将结果输出
代码1-22中,将student集合和grade集合存储在HDFS下的data文件夹中
代码1-22
root@lejian:/data# cat student 1 Amy female 18 2 Tom male 19 3 Sam male 21 4 John male 19 5 Lily female 21 6 Rose female 20 root@lejian:/data# cat grade 1 Math 89 2 Math 75 4 English 85 3 English 95 5 Math 91 5 English 88 6 Math 78 6 English 99 2 English 80 root@lejian:/data# hadoop fs -put student /data root@lejian:/data# hadoop fs -put grade /data root@lejian:/data# hadoop fs -ls -R /data -rw-r--r-- 1 root supergroup 105 2017-01-08 09:59 /data/grade -rw-r--r-- 1 root supergroup 93 2017-01-08 09:59 /data/student
代码1-23
package com.hadoop.mapreduce; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Mapper; import org.apache.hadoop.mapreduce.lib.input.FileSplit; public class NaturalJoinMap extends Mapper<LongWritable, Text, IntWritable, Text> { private String fileName = ""; private Text val = new Text(); private IntWritable stuKey = new IntWritable(); protected void setup(Context context) throws java.io.IOException, InterruptedException { FileSplit fileSplit = (FileSplit) context.getInputSplit(); fileName = fileSplit.getPath().getName(); }; protected void map(LongWritable key, Text value, Context context) throws java.io.IOException, InterruptedException { String[] arr = value.toString().split(" "); stuKey.set(Integer.parseInt(arr[0])); val.set(fileName + " " + value.toString()); context.write(stuKey, val); }; }
代码1-24
package com.hadoop.mapreduce; import java.util.ArrayList; import java.util.List; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Reducer; public class NaturalJoinReduce extends Reducer<IntWritable, Text, Text, NullWritable> { private Text student = new Text(); private Text value = new Text(); protected void reduce(IntWritable key, Iterable<Text> values, Context context) throws java.io.IOException, InterruptedException { List<String> grades = new ArrayList<String>(); for (Text val : values) { if (val.toString().contains("student")) { student.set(studentStr(val.toString())); } else { grades.add(gradeStr(val.toString())); } } for (String grade : grades) { value.set(student.toString() + grade); context.write(value, NullWritable.get()); } }; private String studentStr(String line) { String[] arr = line.split(" "); StringBuilder str = new StringBuilder(); for (int i = 1; i < arr.length; i++) { str.append(arr[i] + " "); } return str.toString(); } private String gradeStr(String line) { String[] arr = line.split(" "); StringBuilder str = new StringBuilder(); for (int i = 2; i < arr.length; i++) { str.append(arr[i] + " "); } return str.toString(); } }
代码1-25
package com.hadoop.mapreduce; 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.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; public class NaturalJoin { public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException { if (args == null || args.length != 2) { throw new RuntimeException("请输入输入路径、输出路径"); } Configuration conf = new Configuration(); Job job = Job.getInstance(conf); job.setJobName("NaturalJoin"); job.setJarByClass(NaturalJoin.class); job.setMapperClass(NaturalJoinMap.class); job.setMapOutputKeyClass(IntWritable.class); job.setMapOutputValueClass(Text.class); job.setReducerClass(NaturalJoinReduce.class); job.setOutputKeyClass(IntWritable.class); job.setOutputValueClass(NullWritable.class); FileInputFormat.addInputPaths(job, args[0]); FileOutputFormat.setOutputPath(job, new Path(args[1])); System.exit(job.waitForCompletion(true) ? 0 : 1); } }
运行代码1-25,运行结果如代码1-26
代码1-26
root@lejian:/data# hadoop jar naturalJoin.jar com.hadoop.mapreduce.NaturalJoin /data /output ………… root@lejian:/data# hadoop fs -ls -R /output -rw-r--r-- 1 root supergroup 0 2017-01-08 11:19 /output/_SUCCESS -rw-r--r-- 1 root supergroup 237 2017-01-08 11:19 /output/part-r-00000 root@lejian:/data# hadoop fs -cat /output/part-r-00000 1 Amy female 18 Math 89 2 Tom male 19 English 80 2 Tom male 19 Math 75 3 Sam male 21 English 95 4 John male 19 English 85 5 Lily female 21 English 88 5 Lily female 21 Math 91 6 Rose female 20 English 99 6 Rose female 20 Math 78