最近在研究Flink SQL流,这里写一个简单的实战。
每1分钟统计过去1个小时,各个商品的购买量。
数据格式为{"behavior":"cart","itemId":19,"nowTime":1562314387553}
behavior:为用户行为,有cart,pv,buy。我们这里要的就是buy
import com.alibaba.fastjson.JSON;
import com.closeli.writetohbase.fun.SQLKEYPro;
import com.closeli.writetohbase.fun.UDFToTime;
import com.closeli.writetohbase.fun.UserWaterMark;
import com.closeli.writetohbase.pojo.UserData;
import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.api.common.serialization.SimpleStringSchema;
import org.apache.flink.api.common.typeinfo.TypeHint;
import org.apache.flink.api.common.typeinfo.TypeInformation;
import org.apache.flink.api.java.tuple.Tuple3;
import org.apache.flink.streaming.api.TimeCharacteristic;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer010;
import org.apache.flink.table.api.Table;
import org.apache.flink.table.api.TableEnvironment;
import org.apache.flink.table.api.java.StreamTableEnvironment;
import java.util.Properties;
public class SQLTOKafka {
public static void main(String[] args) throws Exception {
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
StreamTableEnvironment tEnv = TableEnvironment.getTableEnvironment(env);
env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);
Properties prop = new Properties();
prop.put("group.id","aaaaaaa");
prop.put("bootstrap.servers","node223:6667,node224:6667,node225:6667");
FlinkKafkaConsumer010 consumer = new FlinkKafkaConsumer010<>("mydata", new SimpleStringSchema(), prop);
DataStreamSource dataSource = env.addSource(consumer);
// dataSource.print();
SingleOutputStreamOperator filterData = dataSource.map(new MapFunction() {
@Override
public UserData map(String value) throws Exception {
UserData userData = null;
try {
userData = JSON.parseObject(value, UserData.class);
} catch (Exception e) {
e.printStackTrace();
}
return userData;
}
}).filter(x -> x != null && x.getBehavior().equals("pv")).assignTimestampsAndWatermarks(new UserWaterMark());
//自定义UDF,将时间戳改成东八区的形式
tEnv.registerFunction("timetoformat",new UDFToTime());
tEnv.registerDataStream("userTable",filterData,"behavior,itemId,nowTime.rowTime");
// Table table = tEnv.sqlQuery("select url,count(1) from userTable group by TUMBLE(nowTime,INTERVAL '5' SECOND),url");
Table table = tEnv.sqlQuery("select itemId,count(1),timetoformat(HOP_END(nowTime, INTERVAL '10' SECOND, INTERVAL '30' MINUTE)) from userTable group by HOP(nowTime, INTERVAL '10' SECOND, INTERVAL '30' MINUTE),itemId");
// DataStream resultData = tEnv.toAppendStream(table, Types.TUPLE(Types.INT, Types.LONG ,Types.SQL_TIMESTAMP));
// DataStream resultData = tEnv.toAppendStream(table, Types.TUPLE(Types.INT, Types.LONG ,Types.LONG));
DataStream> resultData = tEnv.toAppendStream(table, TypeInformation.of(new TypeHint>() {}));
resultData.print();
env.execute("SQLTOKafka");
}
}
是一个pojo类,属性是behavior、itemId、nowTime
这个是自定义的UDF,负责将timestamp转成东八区的时间,Flink SQL中的时间距离北京时间差了8个小时,所以,如果后续有需要,可以进行转换。
import org.apache.flink.table.functions.ScalarFunction;
import java.sql.Timestamp;
public class UDFToTime extends ScalarFunction {
public Long eval(Timestamp s){
return s.getTime() + 28800000;
}
}
1、目前Flinik SQL在流式处理中不支持limit 。对于order by,说是只可以对time进行排序,但我试了下,发现并没有成功。感觉支持的不是很好,可能是我没有用好。
2、对于时间,如何要使用event time,那么在前面的datastream先注册好,然后在注册table的时候,在对应的属性后 .rowTime,这样就表示这个是event time了。如果用的是processingTime,那么就需要在注册表的时候,属性那里,增加一个proctime.proctime,这个时候proctime就是系统当前时间,可以直接拿来用。
对于其他的知识点的话,大家可以多看看官网
https://ci.apache.org/projects/flink/flink-docs-release-1.9/dev/table/sql.html