温度排序代码,具体说明可以搜索其他博客
KeyPair.java
package temperaturesort; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.WritableComparable; import java.io.DataInput; import java.io.DataOutput; import java.io.IOException; public class KeyPair implements WritableComparable<KeyPair> { private int hot; private int year; public int getYear() { return year; } public void setYear(int year) { this.year = year; } public int getHot() { return hot; } public void setHot(int hot) { this.hot = hot; } public int compareTo(KeyPair o) { int result = this.year-o.getYear(); if(result!=0){ return result<0?-1:1; } return -( this.hot < o.getHot() ? -1 :(this.hot == o.getHot()?0:1)); } public void write(DataOutput dataOutput) throws IOException { dataOutput.writeInt(year); dataOutput.writeInt(hot); } public void readFields(DataInput dataInput) throws IOException { this.year=dataInput.readInt(); this.hot=dataInput.readInt(); } @Override public String toString() { return year+"/t"+hot; } @Override public int hashCode() { return new Integer(year+hot).hashCode(); } }
Sort.java:
package temperaturesort; import org.apache.hadoop.io.WritableComparable; import org.apache.hadoop.io.WritableComparator; public class Sort extends WritableComparator { public Sort(){ super(KeyPair.class,true); } @Override public int compare(WritableComparable a, WritableComparable b) { KeyPair key1 = (KeyPair)a; KeyPair key2 = (KeyPair)b; int result = key1.getYear()-key2.getYear(); if(result!=0){ return result<0?-1:1; } return key1.getHot()< key2.getHot() ? 1 :(key1.getHot() == key2.getHot()?0:-1); } }
Partition.java:
package temperaturesort; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Partitioner; public class Partition extends Partitioner<KeyPair,Text>{ @Override public int getPartition(KeyPair keyPair, Text text, int num) { return keyPair.getYear()*127 % num; } }
Group.java:
package temperaturesort; import org.apache.hadoop.io.WritableComparable; import org.apache.hadoop.io.WritableComparator; public class Group extends WritableComparator { public Group(){ super(KeyPair.class,true); } @Override public int compare(WritableComparable a, WritableComparable b) { KeyPair key1 = (KeyPair)a; KeyPair key2 = (KeyPair)b; return key1.getYear() < key2.getYear() ? -1 : (key1.getYear()==key2.getYear()?0:1); } }
RunJob.java:
package temperaturesort; 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 java.io.IOException; import java.text.ParseException; import java.text.SimpleDateFormat; import java.util.Calendar; import java.util.Date; public class RunJob { public static class TempSortMapper extends Mapper<Object,Text,KeyPair,Text>{ static SimpleDateFormat simpleDateFormat =new SimpleDateFormat("yyyy-MM-dd HH:mm:ss"); @Override protected void map(Object key, Text value, Context context) throws IOException, InterruptedException { String line = value.toString(); String[] strArr=line.split("/t"); if(strArr.length==2){ try { Date date = simpleDateFormat.parse(strArr[0]); Calendar calendar = Calendar.getInstance(); calendar.setTime(date); int year = calendar.get(1); int hot = Integer.parseInt(strArr[1].substring(0,strArr[1].indexOf("C"))); KeyPair keyPair =new KeyPair(); keyPair.setHot(hot); keyPair.setYear(year); /*System.out.println("-------------------------------------------------------------------"); System.out.println(keyPair);*/ context.write(keyPair,value); } catch (ParseException e) { e.printStackTrace(); } } } } public static class TempSortReducer extends Reducer<KeyPair,Text,KeyPair,Text>{ @Override protected void reduce(KeyPair key, Iterable<Text> values, Context context) throws IOException, InterruptedException { for(Text text:values) context.write(key,text); } } public static void main(String[] args) throws Exception { //System.setProperty("hadoop.home.dir","E://softs//majorSoft//hadoop-2.7.5"); Configuration conf = new Configuration(); conf.set("mapreduce.app-submission.cross-platform", "true"); Path fileInput = new Path("hdfs://mycluster/testFile/hot.txt"); Path fileOutput = new Path("hdfs://mycluster/output/hot"); Job job =Job.getInstance(conf ,"temperatureSort"); job.setJar("E://bigData//hadoopDemo//out//artifacts//wordCount_jar//hadoopDemo.jar"); job.setJarByClass(RunJob.class); job.setMapperClass(TempSortMapper.class); job.setReducerClass(TempSortReducer.class); job.setMapOutputKeyClass(KeyPair.class); job.setMapOutputValueClass(Text.class); job.setNumReduceTasks(3); job.setSortComparatorClass(Sort.class); job.setPartitionerClass(Partition.class); job.setGroupingComparatorClass(Group.class); FileInputFormat.addInputPath(job,fileInput); FileOutputFormat.setOutputPath(job,fileOutput); System.exit(job.waitForCompletion(true)?0:1); } }
其中自定义的sort和parititon是在mapTask任务之后使用的,而Group是在reduce任务使用的。
原创文章,作者:奋斗,如若转载,请注明出处:https://blog.ytso.com/9430.html