Trident学习笔记（一）

1. Trident入门

Trident

-------------------

　三叉戟

　storm高级抽象，支持有状态流处理；

　好处是确保消费被处理一次；

　以小批次方式处理输入流，得到精准一次性处理 ；

　不再使用bolt，使用functions、aggreates、filters以及states。

　Trident Tuple: trident top的数据模型，trident处理数据的单元；

　　　　　　　　每个tuple有预定义的字段列表构成，字段类型可以是byte；

　　　　　　　　character，integer，long，float，double，Boolean or byte array。

　Trident functions: 包含修改tuple的业务逻辑，输入的是tuple的字段，输出多个tuple。

import org.apache.storm.trident.operation.BaseFunction;
import org.apache.storm.trident.operation.TridentCollector;
import org.apache.storm.trident.tuple.TridentTuple;
import org.apache.storm.tuple.Values;

/**
 * 求和函数
 */
public class SumFunction extends BaseFunction {

    @Override
    public void execute(TridentTuple input, TridentCollector collector) {
        Integer num1 = input.getInteger(0);
        Integer num2 = input.getInteger(1);
        int sum = num1 + num2;
        collector.emit(new Values(sum));
    }

}

如果tuple有a, b, c, d四个field，只有a和b作为输入传给function，functions会生成新的sum字段，

sum字段和输入的元祖进行合并，生成一个完成tuple，因此，新的tuple的总和字段个数是a, b, c, d, sum。

Trident Filter

--------------------

　　1. 描述

　　获取字段集合作为输入，输出boolean，如果反悔true，tuple在流中保留，否则删除，

　　a, b, c, d, sum是元祖的字段，sum作为输入传递给filter，判断sum是否为偶数，

　　如果是偶数，tuple(a, b, c, d, sum)保留，否则tuple删除。

　　2. 代码

import org.apache.storm.trident.operation.BaseFilter;
import org.apache.storm.trident.tuple.TridentTuple;

/**
 * 校验是否是偶数的过滤器
 */
public class CheckEvenFilter extends BaseFilter {

    @Override
    public boolean isKeep(TridentTuple input) {
        Integer sum = input.getInteger(0);
        if (sum % 2 == 0) {
            return true;
        }
        return false;
    }

}

Trident projections

--------------------

　　1. 描述

　　　投影操作中，trident值保留在投影中制定的字段，

　　　x, y, z --> projection(x) --> x

　　2. 调用投影的方式

　　　mystream.project(new fields("x"));

写一个topology

import org.apache.storm.trident.operation.BaseFunction;
import org.apache.storm.trident.operation.TridentCollector;
import org.apache.storm.trident.tuple.TridentTuple;

public class PrintFunction extends BaseFunction {

    @Override
    public void execute(TridentTuple input, TridentCollector collector) {
        Integer sum = input.getInteger(0);
        System.out.println(this.getCLass.getSimpleName + ": " + sum);
    }

}

import com.google.common.collect.ImmutableList;
import org.apache.storm.Config;
import org.apache.storm.LocalCluster;
import org.apache.storm.trident.Stream;
import org.apache.storm.trident.TridentTopology;
import org.apache.storm.trident.testing.FeederBatchSpout;
import org.apache.storm.tuple.Fields;
import org.apache.storm.tuple.Values;

public class TridentTopologyApp {

    public static void main(String[] args) {
        // 创建topology
        TridentTopology topology = new TridentTopology();

        // 创建spout
        FeederBatchSpout testSpout = new FeederBatchSpout(ImmutableList.of("a", "b", "c", "d"));

        // 创建流
        Stream stream = topology.newStream("spout", testSpout);
        stream.shuffle().each(new Fields("a", "b"), new SumFunction(), new Fields("sum")).parallelismHint(1)
                .shuffle().each(new Fields("sum"), new CheckEvenFilter()).parallelismHint(1)
                .shuffle().each(new Fields("sum"), new PrintFunction(), new Fields("xxx")).parallelismHint(1);

        // 本地提交
        LocalCluster cluster = new LocalCluster();
        cluster.submitTopology("TridentDemo", new Config(), topology.build());

        // 测试数据
        testSpout.feed(ImmutableList.of(new Values(1, 2, 3, 4)));
        testSpout.feed(ImmutableList.of(new Values(2, 3, 4, 5)));
        testSpout.feed(ImmutableList.of(new Values(3, 4, 5, 6)));
        testSpout.feed(ImmutableList.of(new Values(4, 5, 6, 7)));
    }

}

输出结果

SumFunction:,
CheckEvenFilter:
PrintFunction:
SumFunction:,
CheckEvenFilter:
PrintFunction:
SumFunction:,
CheckEvenFilter:
PrintFunction:
SumFunction:,
CheckEvenFilter:
PrintFunction: 

加入一个求平均数的函数

import org.apache.storm.trident.operation.BaseFunction;
import org.apache.storm.trident.operation.TridentCollector;
import org.apache.storm.trident.tuple.TridentTuple;

/**
 * 求平均值方法
 */
public class AverageFunction extends BaseFunction {

    @Override
    public void execute(TridentTuple input, TridentCollector collector) {
        int a = input.getIntegerByField("a");
        int b = input.getIntegerByField("b");
        int c = input.getIntegerByField("c");
        int d = input.getIntegerByField("d");
        int sum = input.getIntegerByField("sum");
        float avg = (float) ((a+b+c+d+sum) / 5.0);
        System.out.println(this.getClass().getSimpleName() + ": avg = " + avg);
    }

}

import com.google.common.collect.ImmutableList;
import org.apache.storm.Config;
import org.apache.storm.LocalCluster;
import org.apache.storm.trident.Stream;
import org.apache.storm.trident.TridentTopology;
import org.apache.storm.trident.testing.FeederBatchSpout;
import org.apache.storm.tuple.Fields;
import org.apache.storm.tuple.Values;

public class TridentTopologyApp {

    public static void main(String[] args) {
        // 创建topology
        TridentTopology topology = new TridentTopology();

        // 创建spout
        FeederBatchSpout testSpout = new FeederBatchSpout(ImmutableList.of("a", "b", "c", "d"));

        // 创建流
        Stream stream = topology.newStream("spout", testSpout);
        stream.shuffle().each(new Fields("a", "b"), new SumFunction(), new Fields("sum")).parallelismHint(1)
                .shuffle().each(new Fields("sum"), new CheckEvenFilter()).parallelismHint(1)
                .shuffle().each(new Fields("sum"), new PrintFunction(), new Fields("res")).parallelismHint(1)
                .shuffle().each(new Fields("a", "b", "c", "d", "sum"), new AverageFunction(), new Fields("avg")).parallelismHint(1);

        // 本地提交
        LocalCluster cluster = new LocalCluster();
        cluster.submitTopology("TridentDemo", new Config(), topology.build());

        // 测试数据
        testSpout.feed(ImmutableList.of(new Values(1, 2, 3, 4)));
        testSpout.feed(ImmutableList.of(new Values(2, 3, 4, 5)));
        testSpout.feed(ImmutableList.of(new Values(3, 4, 5, 6)));
        testSpout.feed(ImmutableList.of(new Values(4, 5, 6, 7)));
    }

}

2. Trident聚合函数

分区聚合

import com.google.common.collect.ImmutableList;
import org.apache.storm.Config;
import org.apache.storm.LocalCluster;
import org.apache.storm.trident.Stream;
import org.apache.storm.trident.TridentTopology;
import org.apache.storm.trident.testing.FeederBatchSpout;
import org.apache.storm.tuple.Fields;
import org.apache.storm.tuple.Values;

public class TridentTopologyApp2 {

    public static void main(String[] args) {
        // 创建topology
        TridentTopology topology = new TridentTopology();

        // 创建spout
        FeederBatchSpout testSpout = new FeederBatchSpout(ImmutableList.of("a", "b"));

        // 创建流
        Stream stream = topology.newStream("testSpout", testSpout);
        stream.shuffle().each(new Fields("a", "b"), new MyFilter1()).parallelismHint(1)
                .global().each(new Fields("a", "b"), new MyFilter2()).parallelismHint(1)
                .partitionBy(new Fields("a"))
                //.each(new Fields("a", "b"), new MyFunction1(), new Fields("none")).parallelismHint(1)
                .partitionAggregate(new Fields("a"), new MyCount(), new Fields("count"))
                .each(new Fields("count"), new MyPrintFunction1(), new Fields("xxx")).parallelismHint(1);

        // 本地提交
        LocalCluster cluster = new LocalCluster();
        cluster.submitTopology("TridentDemo2", new Config(), topology.build());

        // 测试数据
        testSpout.feed(ImmutableList.of(new Values(1, 2)));
        testSpout.feed(ImmutableList.of(new Values(2, 3)));
        testSpout.feed(ImmutableList.of(new Values(2, 4)));
        testSpout.feed(ImmutableList.of(new Values(3, 5)));
    }

}

批次聚合

3. 自定义聚合函数-Sum-SumAsAggregator