Spark 读取Hbase表数据并实现类似groupByKe

package com.union.bigdata.spark.hbase;import org.apache.hadoop.hbase.HBaseConfiguration;import org.apache.hadoop.hbase.mapreduce.TableSplit;import org.apache.hadoop.hbase.util.Base64;import org.apache.hadoop.hbase.util.Bytes;import org.apache.spark.api.java.JavaSparkContext;import org.apache.spark.api.java.JavaRDD;import org.apache.spark.SparkConf;import org.apache.spark.api.java.function.Function;import org.apache.spark.api.java.function.Function2;import org.apache.hadoop.conf.Configuration;import org.apache.hadoop.hbase.client.Scan;import org.apache.hadoop.hbase.client.Result;import org.apache.hadoop.hbase.io.ImmutableBytesWritable;import org.apache.hadoop.hbase.mapreduce.TableInputFormat;import org.apache.spark.api.java.JavaPairRDD;import org.apache.hadoop.hbase.protobuf.ProtobufUtil;import org.apache.hadoop.hbase.protobuf.generated.ClientProtos;import org.apache.spark.api.java.function.PairFunction;import scala.Tuple10;import scala.Tuple2;import java.io.IOException;import java.util.ArrayList;import java.util.List;public class ReadHbase {

    private static String appName = "ReadTable";

    public static void main(String[] args) {
        SparkConf sparkConf = new SparkConf();

    //we can also run it at local:"local[3]"  the number 3 means 3 threads
        sparkConf.setMaster("spark://master:7077").setAppName(appName);
        JavaSparkContext jsc = new JavaSparkContext(sparkConf);

        Configuration conf = HBaseConfiguration.create();
        conf.set("hbase.zookeeper.quorum", "master"); 
        conf.set("hbase.zookeeper.property.clientPort", "2181"); 
        Scan scan = new Scan();
        scan.addFamily(Bytes.toBytes("baseINFO"));
        scan.addColumn(Bytes.toBytes("baseINFO"), Bytes.toBytes("telphone_number"));
        scan.addColumn(Bytes.toBytes("baseINFO"), Bytes.toBytes("age"));
        scan.addColumn(Bytes.toBytes("baseINFO"), Bytes.toBytes("gender"));

        String scanToString = "";
        try {
            ClientProtos.Scan proto = ProtobufUtil.toScan(scan);
            scanToString = Base64.encodeBytes(proto.toByteArray());
        } catch (IOException io) {
            System.out.println(io);
        }

        for (int i = 0; i < 2; i++) {
            try {
                String tableName = "VIPUSER";
                conf.set(TableInputFormat.INPUT_TABLE, tableName);
                conf.set(TableInputFormat.SCAN, scanToString);

                //get the Result of query from the Table of Hbase
                JavaPairRDD<ImmutableBytesWritable, Result> hBaseRDD = jsc.newAPIHadoopRDD(conf,
                        TableInputFormat.class, ImmutableBytesWritable.class,
                        Result.class);

                //group by row key like : [(Andy,110,21,0),(Tom,120,18,1)]
                JavaPairRDD<String, List<Integer>> art_scores = hBaseRDD.mapToPair(
                        new PairFunction<Tuple2<ImmutableBytesWritable, Result>, String, List<Integer>>() {
                            @Override
                            public Tuple2<String, List<Integer>> call(Tuple2<ImmutableBytesWritable, Result> results) {

                                List<Integer> list = new ArrayList<Integer>();

                                byte[] telphone_number = results._2().getValue(Bytes.toBytes("baseINFO"), Bytes.toBytes("telphone_number"));
                                byte[] age = results._2().getValue(Bytes.toBytes("baseINFO"), Bytes.toBytes("age"));
                                byte[] gender = results._2().getValue(Bytes.toBytes("baseINFO"), Bytes.toBytes("gender"));

                //the type of storage at Hbase is Byte Array,so we must let it be normal like Int,String and so on 
                                list.add(Integer.parseInt(Bytes.toString(telphone_number)));
                                list.add(Integer.parseInt(Bytes.toString(age)));
                                list.add(Integer.parseInt(Bytes.toString(gender)));

                                return new Tuple2<String, List<Integer>>(Bytes.toString(results._1().get()), list);
                            }
                        }
                );

                //switch to Cartesian product 
                JavaPairRDD<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>> cart = art_scores.cartesian(art_scores);

                //use Row Key to delete the repetition from the last step "Cartesian product"  
                JavaPairRDD<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>> cart2 = cart.filter(
                        new Function<Tuple2<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>>, Boolean>() {
                            public Boolean call(Tuple2<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>> tuple2Tuple2Tuple2) throws Exception {

                                return tuple2Tuple2Tuple2._1()._1().compareTo(tuple2Tuple2Tuple2._2()._1()) < 0;
                            }
                        }
                );

                System.out.println("Create the List 'collect'...");

        //get the result we need
                 List<Tuple2<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>>> collect = cart2.collect();
                 System.out.println("Done..");
                 System.out.println(collect.size() > i ? collect.get(i):"STOP");

                 if (collect.size() > i ) break;
            } catch (Exception e) {
                System.out.println(e);
            }
        }
    }
}