Spark实时流计算Java案例

现在，网上基于spark的代码基本上都是Scala，很多书上也都是基于Scala，没办法，谁叫spark是Scala写出来的了，但是我现在还没系统的学习Scala，所以只能用java写spark程序了，spark支持java，而且Scala也基于JVM,不说了，直接上代码

这是官网上给出的例子，大数据学习中经典案例单词计数

在linux下一个终端 输入 $ nc -lk 9999

然后运行下面的代码

package com.tg.spark.stream;

import java.util.Arrays;

import org.apache.spark.*;
import org.apache.spark.api.java.function.*;
import org.apache.spark.streaming.*;
import org.apache.spark.streaming.api.java.*;
import scala.Tuple2;
/**
 *
 * @author 汤高
 *
 */
public class SparkStream {
    public static void main(String[] args) {

        // Create a local StreamingContext with two working thread and batch
        // interval of 1 second
        SparkConf conf = new SparkConf().setMaster("local[4]").setAppName("NetworkWordCount").set("spark.testing.memory",
                "2147480000");
        JavaStreamingContext jssc = new JavaStreamingContext(conf, Durations.seconds(1));
        System.out.println(jssc);

        // Create a DStream that will connect to hostname:port, like
        // localhost:9999
        JavaReceiverInputDStream<String> lines = jssc.socketTextStream("master", 9999);
        //JavaDStream<String> lines = jssc.textFileStream("hdfs://master:9000/stream");

        // Split each line into words
        JavaDStream<String> words = lines.flatMap(new FlatMapFunction<String, String>() {
            @Override
            public Iterable<String> call(String x) {
                System.out.println(Arrays.asList(x.split(" ")).get(0));
                return Arrays.asList(x.split(" "));
            }
        });

        // Count each word in each batch
        JavaPairDStream<String, Integer> pairs = words.mapToPair(new PairFunction<String, String, Integer>() {
            @Override
            public Tuple2<String, Integer> call(String s) {
                return new Tuple2<String, Integer>(s, 1);
            }
        });
        System.out.println(pairs);
        JavaPairDStream<String, Integer> wordCounts = pairs.reduceByKey(new Function2<Integer, Integer, Integer>() {
            @Override
            public Integer call(Integer i1, Integer i2) {
                return i1 + i2;
            }
        });

        // Print the first ten elements of each RDD generated in this DStream to
        // the console

        wordCounts.print();
        //wordCounts.saveAsHadoopFiles("hdfs://master:9000/testFile/", "spark", new Text(), new IntWritable(), JavaPairDStream<Text,IntWritable>());
        wordCounts.dstream().saveAsTextFiles("hdfs://master:9000/testFile/", "spark");
        //wordCounts.saveAsHadoopFiles("hdfs://master:9000/testFile/", "spark",Text,IntWritable);
        //System.out.println(wordCounts.count());
        jssc.start();
        //System.out.println(wordCounts.count());// Start the computation
        jssc.awaitTermination();   // Wait for the computation to terminate
    }

}

然后再刚刚的终端输入 hello world

# TERMINAL 1:
# Running Netcat

$ nc -lk 9999

hello world

就可以通过控制台看到

-------------------------------------------
Time: 1357008430000 ms
-------------------------------------------
(hello,1)
(world,1)
...

并且hdfs上也可以看到通过计算生成的实时文件

第二个案例是，不是通过socketTextStream套接字，而是直接通过hdfs上的某个文件目录来作为输入数据源

package com.tg.spark.stream;

import java.util.Arrays;

import org.apache.spark.*;
import org.apache.spark.api.java.function.*;
import org.apache.spark.streaming.*;
import org.apache.spark.streaming.api.java.*;
import scala.Tuple2;
/**
 *
 * @author 汤高
 *
 */
public class SparkStream2 {
    public static void main(String[] args) {

        // Create a local StreamingContext with two working thread and batch
        // interval of 1 second
        SparkConf conf = new SparkConf().setMaster("local[4]").setAppName("NetworkWordCount").set("spark.testing.memory",
                "2147480000");
        JavaStreamingContext jssc = new JavaStreamingContext(conf, Durations.seconds(1));
        System.out.println(jssc);

        // Create a DStream that will connect to hostname:port, like
        // localhost:9999
        //JavaReceiverInputDStream<String> lines = jssc.socketTextStream("master", 9999);
        JavaDStream<String> lines = jssc.textFileStream("hdfs://master:9000/stream");

        // Split each line into words
        JavaDStream<String> words = lines.flatMap(new FlatMapFunction<String, String>() {
            @Override
            public Iterable<String> call(String x) {
                System.out.println(Arrays.asList(x.split(" ")).get(0));
                return Arrays.asList(x.split(" "));
            }
        });

        // Count each word in each batch
        JavaPairDStream<String, Integer> pairs = words.mapToPair(new PairFunction<String, String, Integer>() {
            @Override
            public Tuple2<String, Integer> call(String s) {
                return new Tuple2<String, Integer>(s, 1);
            }
        });
        System.out.println(pairs);
        JavaPairDStream<String, Integer> wordCounts = pairs.reduceByKey(new Function2<Integer, Integer, Integer>() {
            @Override
            public Integer call(Integer i1, Integer i2) {
                return i1 + i2;
            }
        });

        // Print the first ten elements of each RDD generated in this DStream to
        // the console

        wordCounts.print();
        //wordCounts.saveAsHadoopFiles("hdfs://master:9000/testFile/", "spark", new Text(), new IntWritable(), JavaPairDStream<Text,IntWritable>());
        wordCounts.dstream().saveAsTextFiles("hdfs://master:9000/testFile/", "spark");
        //wordCounts.saveAsHadoopFiles("hdfs://master:9000/testFile/", "spark",Text,IntWritable);
        //System.out.println(wordCounts.count());
        jssc.start();
        //System.out.println(wordCounts.count());// Start the computation
        jssc.awaitTermination();   // Wait for the computation to terminate
    }

}

这样就存在端口一直在监控你的那个目录，只要它有文件生成，就会马上读取到它里面的内容，你可以先运行程序，然后手动添加一个文件到刚刚的目录，就可以看到输出结果了

码字不易，转载请指明出处http://blog.csdn.net/tanggao1314/article/details/51606721

参考

spark编程指南