Flink去重计数统计用户数-Toy模板网

这篇具有很好参考价值的文章主要介绍了Flink去重计数统计用户数。希望对大家有所帮助。如果存在错误或未考虑完全的地方，请大家不吝赐教，您也可以点击"举报违法"按钮提交疑问。

1.数据

订单表，分别是店铺id、用户id和支付金额

"店铺id,用户id,支付金额",
"shop-1,user-1,1",
"shop-1,user-2,1",
"shop-1,user-2,1",
"shop-1,user-3,1",
"shop-1,user-3,1",
"shop-1,user-1,1",
"shop-1,user-2,1",
"shop-1,user-4,1",
"shop-2,user-4,1",
"shop-2,user-4,1",
"shop-2,user-2,1"

2.可运行案例

import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.table.api.Table;
import org.apache.flink.table.api.bridge.java.StreamTableEnvironment;
import org.apache.flink.types.Row;


public class Test03 {
    public static void main(String[] args) throws Exception {
        // 1. 创建流式执行环境
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        // 2.创建表执行环境
        StreamTableEnvironment tableEnv = StreamTableEnvironment.create(env);
        // 3.读取数据源
        SingleOutputStreamOperator<String> jsonStream = env
                .fromElements("shop-1,user-1,1",
                        "shop-1,user-2,1",
                        "shop-1,user-2,1",
                        "shop-1,user-3,1",
                        "shop-1,user-3,1",
                        "shop-1,user-1,1",
                        "shop-1,user-2,1",
                        "shop-1,user-4,1",
                        "shop-2,user-4,1",
                        "shop-2,user-4,1",
                        "shop-2,user-2,1"
                );
        // 4.流转换为表
        Table table = tableEnv.fromDataStream(jsonStream);

        // 5. 把注册为一个临时视图
        tableEnv.createTemporaryView("tableTmp", table);

        // 6.求每个商店的用户数
        Table table1 = tableEnv.sqlQuery("select shop_id,sum(num) as num,sum(gmv) as gmv from (select shop_id,user_id, 1 as num,sum(gmv) as gmv from (select SPLIT_INDEX(f0,',',0) as shop_id,SPLIT_INDEX(f0,',',1) as user_id,cast(SPLIT_INDEX(f0,',',2) as bigint) as gmv from tableTmp) t1 group by shop_id,user_id) t2 group by shop_id");

        // 7.打印
        tableEnv.toRetractStream(table1, Row.class).print(">>>>>>");

        // 8.执行
        env.execute("test");
    }
}

sql：

select
  shop_id,
  sum(num) as num,
  sum(gmv) as gmv
from
  (
    select
      shop_id,
      user_id,
      1 as num,
      sum(gmv) as gmv
    from
      (
        select
          SPLIT_INDEX(f0, ',', 0) as shop_id,
          SPLIT_INDEX(f0, ',', 1) as user_id,
          cast(SPLIT_INDEX(f0, ',', 2) as bigint) as gmv
        from
          tableTmp
      ) t1
    group by
      shop_id,
      user_id
  ) t2
group by
  shop_id

3.运行结果

>>>>>>:7> (true,+U[shop-2, 2, 3])

>>>>>>:1> (true,+U[shop-1, 4, 8])

>>>>>>:7> (true,+I[shop-2, 1, 1])
>>>>>>:1> (true,+I[shop-1, 1, 1])
>>>>>>:1> (false,-U[shop-1, 1, 1])
>>>>>>:7> (false,-U[shop-2, 1, 1])
>>>>>>:1> (true,+U[shop-1, 2, 2])
>>>>>>:7> (true,+U[shop-2, 2, 2])
>>>>>>:1> (false,-U[shop-1, 2, 2])
>>>>>>:7> (false,-U[shop-2, 2, 2])
>>>>>>:1> (true,+U[shop-1, 1, 1])
>>>>>>:7> (true,+U[shop-2, 1, 1])
>>>>>>:1> (false,-U[shop-1, 1, 1])
>>>>>>:7> (false,-U[shop-2, 1, 1])
>>>>>>:7> (true,+U[shop-2, 2, 3])
>>>>>>:1> (true,+U[shop-1, 2, 3])
>>>>>>:1> (false,-U[shop-1, 2, 3])
>>>>>>:1> (true,+U[shop-1, 3, 4])
>>>>>>:1> (false,-U[shop-1, 3, 4])
>>>>>>:1> (true,+U[shop-1, 2, 3])
>>>>>>:1> (false,-U[shop-1, 2, 3])
>>>>>>:1> (true,+U[shop-1, 3, 5])
>>>>>>:1> (false,-U[shop-1, 3, 5])
>>>>>>:1> (true,+U[shop-1, 2, 3])
>>>>>>:1> (false,-U[shop-1, 2, 3])
>>>>>>:1> (true,+U[shop-1, 3, 6])
>>>>>>:1> (false,-U[shop-1, 3, 6])
>>>>>>:1> (true,+U[shop-1, 4, 7])
>>>>>>:1> (false,-U[shop-1, 4, 7])
>>>>>>:1> (true,+U[shop-1, 3, 6])
>>>>>>:1> (false,-U[shop-1, 3, 6])
>>>>>>:1> (true,+U[shop-1, 4, 8])

4.原理

Flink回撤流原理文章来源地址https://www.toymoban.com/news/detail-769562.html

到了这里，关于Flink去重计数统计用户数的文章就介绍完了。如果您还想了解更多内容，请在右上角搜索TOY模板网以前的文章或继续浏览下面的相关文章，希望大家以后多多支持TOY模板网！