Flink中BloomFilter（布隆过滤器）和ValueState的结合使用

一、需求

有以下数据：

用户ID,活动ID,时间,事件类型,省份
u001,A1,2019-09-02 10:10:11,1,北京市
u001,A1,2019-09-02 14:10:11,1,北京市
u001,A1,2019-09-02 14:10:11,2,北京市
u002,A1,2019-09-02 14:10:11,1,北京市
u002,A2,2019-09-02 14:10:11,1,北京市
u002,A2,2019-09-02 15:10:11,1,北京市
u002,A2,2019-09-02 15:10:11,2,北京市

事件类型：
0:曝光
1:点击
2:参与

• 统计上面点击、参与某个活动的人数和次数？

• 要求使用：ValueState结合BloomFilter完成

---

二、分析

1、人数

• 按活动id 和 事件类型id 分组
• 把uid放入布隆过滤器中，对uid去重并计数

2、点击次数

每个活动和每个事件类型下的：

• 全局count

3、技术点

• 使用BloomFilter，对中间结果的判断储存
• 自定义ValueState
• 底层Api 的Process的使用

---

三、代码实现

• 布隆过滤器包名：
  import org.apache.flink.shaded.guava18.com.google.common.hash.BloomFilter;

1、数据的读取，做切割处理

import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.api.common.state.ValueState;
import org.apache.flink.api.common.state.ValueStateDescriptor;
import org.apache.flink.api.common.typeinfo.TypeHint;
import org.apache.flink.api.common.typeinfo.TypeInformation;
import org.apache.flink.api.java.tuple.Tuple;
import org.apache.flink.api.java.tuple.Tuple4;
import org.apache.flink.api.java.tuple.Tuple5;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.shaded.guava18.com.google.common.hash.BloomFilter;
import org.apache.flink.shaded.guava18.com.google.common.hash.Funnels;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.datastream.KeyedStream;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.KeyedProcessFunction;
import org.apache.flink.util.Collector;

/**
 * @date: 2020/3/13 07:03
 * @site: www.ianlou.cn
 * @author: lekko 六水
 * @qq: 496208110
 * @description: 使用布隆过滤器，对中间结果的判断储存 < ValusState<BloomFilter> >
 */
public class ActivityCountAndBloomFilter {
    public static void main(String[] args) throws Exception {

        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        //u001,A1,2019-09-02 10:10:11,1,北京市 数据
        DataStreamSource<String> lines = env.socketTextStream("linux01", 8888);

        SingleOutputStreamOperator<Tuple5<String, String, String, String, String>> tpDataStream =
                lines.map(new MapFunction<String, Tuple5<String, String, String, String, String>>() {

                    @Override
                    public Tuple5<String, String, String, String, String> map(String lines) throws Exception {

                        String[] spArr = lines.split(",");
                        String uid = spArr[0];
                        String act = spArr[1];
                        String dt = spArr[2];
                        String type = spArr[3];
                        String province = spArr[4];

                        return Tuple5.of(uid, act, dt, type, province);
                    }
                });

2、按照活动ID和事件类型分组

 KeyedStream<Tuple5<String, String, String, String, String>, Tuple> keyedStream =
                tpDataStream.keyBy(1, 3);

3、调用process，自定义ValueState,对数据进行统计

• ① 在open方法里，自定义state状态描述器，然后通过context拿到ValueState
• ② 在processElement中，初始化三个ValueState中的值，然后写逻辑
• ③ 将最后计算的结果out.collect

SingleOutputStreamOperator<Tuple4<String, String, Integer, Integer>> result =
                keyedStream.process(new KeyedProcessFunction<Tuple, Tuple5<String, String, String, String, String>,
                        Tuple4<String, String, Integer, Integer>>() {

                    //保存分组数据去重后用户ID的布隆过滤器
                    private transient ValueState<BloomFilter> bloomState = null;
                    //保存去重后总人数的state，加transient禁止参与反序列化
                    private transient ValueState<Integer> uidCountState = null;
                    //保存活动的点击数的state
                    private transient ValueState<Integer> clickState = null;


                    @Override
                    public void open(Configuration parameters) throws Exception {

                        ValueStateDescriptor<BloomFilter> bloomDescriptor = new ValueStateDescriptor<>(
                                "actId-type",
                                // 数据类型的class对象，因为是布隆过滤器，所有要使用这种方式来拿
                                TypeInformation.of(new TypeHint<BloomFilter>() {
                                })
                        );
                        ValueStateDescriptor<Integer> uidCountDescriptor = new ValueStateDescriptor<>(
                                "uid-count",
                                Integer.class
                        );
                        ValueStateDescriptor<Integer> clickDescripor = new ValueStateDescriptor<>(
                                "click-count",
                                Integer.class
                        );

                        bloomState = getRuntimeContext().getState(bloomDescriptor);
                        uidCountState = getRuntimeContext().getState(uidCountDescriptor);
                        clickState = getRuntimeContext().getState(clickDescripor);
                    }


                    @Override
                    public void processElement(Tuple5<String, String, String, String, String> input, Context ctx,
                                               Collector<Tuple4<String, String, Integer, Integer>> out) throws Exception {

                        String uid = input.f0;
                        String actId = input.f1;
                        String type = input.f3;

                        //将值取出来，因为ValueState中实质上是以特殊的map集合存储的，一个key,一个value
                        BloomFilter bloomFilter = bloomState.value();
                        Integer uidCound = uidCountState.value();
                        Integer clickCount = clickState.value();

                        //初始化上面三个变量
                        if (clickCount == null) {
                            clickCount = 0;
                        }
                        if (bloomFilter == null) {
                            bloomFilter = BloomFilter.create(Funnels.unencodedCharsFunnel(), 10000000);
                            uidCound = 0;
                        }
                        //布隆过滤器中是否不包含uid,是的话就返回false
                        if (!bloomFilter.mightContain(uid)) {
                            bloomFilter.put(uid); //不包含就添加进去
                            uidCound += 1;
                        }

                        clickCount += 1;
                        bloomState.update(bloomFilter);
                        uidCountState.update(uidCound);
                        clickState.update(clickCount);

                        out.collect(Tuple4.of(actId, type, uidCound, clickCount));
                    }
                });

        result.print();
        env.execute("ActivityCountAndBloomFilter");
    }
}