Flink处理--异步IO

async

import com.alibaba.fastjson.JSONObject;
import org.apache.commons.io.IOUtils;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.streaming.api.functions.async.ResultFuture;
import org.apache.flink.streaming.api.functions.async.RichAsyncFunction;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.http.HttpEntity;
import org.apache.http.HttpResponse;
import org.apache.http.client.config.RequestConfig;
import org.apache.http.client.methods.HttpGet;
import org.apache.http.impl.nio.client.CloseableHttpAsyncClient;
import org.apache.http.impl.nio.client.HttpAsyncClients;
import org.apache.http.util.EntityUtils;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import pers.aishuang.flink.streaming.entity.ItcastDataPartObj;
import pers.aishuang.flink.streaming.entity.VehicleLocationModel;
import pers.aishuang.flink.streaming.utils.GaoDeMapUtils;
import pers.aishuang.flink.streaming.utils.GeoHashUtil;
import pers.aishuang.flink.streaming.utils.RedisUtil;

import java.io.BufferedReader;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.util.Collections;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.Future;
import java.util.function.Consumer;
import java.util.function.Supplier;

/**
 * 通过异步请求获取指定经纬度的位置信息,从高德API获取位置数据
 * 将指定vin某个时间的位置数据保存到redis中
 */
public class AsyncHttpQueryFunction extends RichAsyncFunction<ItcastDataPartObj, ItcastDataPartObj> {
    //创建日志打印器
    private static final Logger logger = LoggerFactory.getLogger(AsyncHttpQueryFunction.class);

    //实现读取异步请求的客户端 (可关闭的http异步请求客户端)
    private static CloseableHttpAsyncClient httpAsyncClient = null;

    @Override
    public void open(Configuration parameters) throws Exception {
        super.open(parameters);
        //设置HttpAsyncClient配置
        RequestConfig config = RequestConfig.custom()
                //-- 设置连接超时时间
                .setConnectTimeout(5000)
                //-- 设置socket超时时间
                .setSocketTimeout(3000)
                .build();
        //初始化异步Http的client
        httpAsyncClient = HttpAsyncClients
                .custom()
                //设置最大连接数量
                .setMaxConnTotal(5)
                .setDefaultRequestConfig(config)
                .build();
        //开启异步http的客户端
        httpAsyncClient.start();
    }

    //实现读取高德API获取位置数据并将位置数据保存到redis中并返回ItcastDataPartObj
    @Override
    public void asyncInvoke(ItcastDataPartObj input, ResultFuture<ItcastDataPartObj> resultFuture) throws Exception {
        //1. 获取当前车辆的经纬度
        Double lng = input.getLng();
        Double lat = input.getLat();
        //2. 通过GaoDeMapUtils工具类根据参数获取请求的url
        String urlByLonLat = GaoDeMapUtils.getUrlByLonLat(lng,lat);
        //3. 创建http get请求对象
        HttpGet httpGet = new HttpGet(urlByLonLat);
        //4. 使用刚创建的http异步客户端执行 http请求对象
        Future<HttpResponse> future = httpAsyncClient.execute(httpGet, null);
        //5. 从执行完成的future中获取数据,返回ItcastDataPartObj对象
        CompletableFuture<ItcastDataPartObj> completableFuture = CompletableFuture.supplyAsync(new Supplier<ItcastDataPartObj>(){
                        //重写get方法
                        //成功时,Redis写入了数据,ItcastDataPartObj的相关字段数据也补齐了。
                        //失败时,什么也不做,原样返回
                        @Override
                        public ItcastDataPartObj get() {
                            try {
                            String country = null;
                            String province = null;
                            String city = null;
                            String district = null;
                            String address = null;
                            //再开个线程自己去拿
                            HttpResponse httpResponse = future.get();

                            //使用future获取到返回的值
                            if(httpResponse.getStatusLine().getStatusCode() == 200 ){
                                HttpEntity entity = httpResponse.getEntity();
                                InputStream contentStream = entity.getContent();
                                //①通过IO流工具类直接生成字符串
                                String content1 = IOUtils.toString(contentStream);

                                //②通过将InputStream转换成输入Reader (转换流:字节流->字符流)
                                InputStreamReader inputStreamReader = new InputStreamReader(contentStream);
                                //--再读取数据流到buffer缓冲区(字符流->高效字符流)
                                BufferedReader bufferedReader = new BufferedReader(inputStreamReader);
                                final int bufferSize = 1024;
                                final char[] buffer = new char[bufferSize];
                                final StringBuilder out = new StringBuilder();
                                int len;
                                while ((len = bufferedReader.read(buffer)) != -1){
                                    out.append(new String(buffer,0,len));
                                }
                                inputStreamReader.close();
                                bufferedReader.close();
                                String content2 = out.toString();

                                //③Entity工具类
                                String content3 = EntityUtils.toString(entity);

                                //----------------------------
                                //将json字符串转换成对象然后读取出来国家,省、市、区、address
                                JSONObject jsonObject = JSONObject.parseObject(content3);
                                JSONObject regeocode = jsonObject.getJSONObject("regeocode");
                                if(regeocode !=null && regeocode.size() > 0 ){
                                    address = regeocode.getString("formatted_address");
                                    JSONObject addressComponent = regeocode.getJSONObject("addressComponent");
                                    if(addressComponent != null && addressComponent.size() > 0) {
                                        country = addressComponent.getString("country");
                                        province = addressComponent.getString("province");
                                        city = addressComponent.getString("city");
                                        district = addressComponent.getString("district");
                                        //将其封装为VehicleLocationModel 并写入到redis
                                        VehicleLocationModel vehicleLocationModel = new VehicleLocationModel(
                                                country,
                                                province,
                                                city,
                                                district,
                                                address,
                                                lat,
                                                lng
                                        );
                                        //获取geohash值作为存储到redis的key
                                        String geoHash = GeoHashUtil.encode(lat,lng);
                                        RedisUtil.set(
                                                Bytes.toBytes(geoHash), //字节数组 (二进制数据)
                                                vehicleLocationModel.toJsonStringArr()//字节数组(二进制数据)
                                        );
                                        //将当前车辆的位置信息赋值
                                        input.setCountry(country);
                                        input.setProvince(province);
                                        input.setCity(city);
                                        input.setDistrict(district);
                                        input.setAddress(address);
                                    }else{
                                        logger.error("当前解析出来的地理信息为空,请检查");
                                    }
                                }else  {
                                    logger.error("当前解析出来的对象为空,请检查!");
                                }


                            }else {
                                logger.error("当前url请求返回reponse错误!");
                            }
                        } catch (Exception e) {
                            e.printStackTrace();
                            }
                            return input;
                        }
                    });

        //6. 从future的thenAccept
        completableFuture.thenAccept(new Consumer<ItcastDataPartObj>() {
            //重写accept方法,使用集合中只放一个对象
            @Override
            public void accept(ItcastDataPartObj itcastDataPartObj) {
                resultFuture.complete(Collections.singleton(itcastDataPartObj));
            }
        });
    }

    //超时了怎么处理(如果当前请求超时,打印输出超时日志或告警信息)
    @Override
    public void timeout(ItcastDataPartObj input, ResultFuture<ItcastDataPartObj> resultFuture) throws Exception {
        //超时时间,打印输出异步请求的超时警告
        System.out.println("当前异步请求超时!");
    }

    //关闭当前的http异步请求客户端

    @Override
    public void close() throws Exception {
        if(httpAsyncClient.isRunning()) httpAsyncClient.close();
    }
}

实例

import org.apache.commons.lang.StringUtils;
import org.apache.flink.api.common.serialization.SimpleStringSchema;
import org.apache.flink.streaming.api.datastream.*;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.timestamps.BoundedOutOfOrdernessTimestampExtractor;
import org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import pers.aishuang.flink.streaming.async.AsyncHttpQueryFunction;
import pers.aishuang.flink.streaming.entity.ItcastDataPartObj;
import pers.aishuang.flink.streaming.entity.OnlineDataObj;
import pers.aishuang.flink.streaming.entity.VehicleInfoModel;
import pers.aishuang.flink.streaming.function.flatmap.VehicleInfoMapMysqlFunction;
import pers.aishuang.flink.streaming.function.map.LocactionInfoReidsFunction;
import pers.aishuang.flink.streaming.function.window.OnlineStatisticsWindowFunction;
import pers.aishuang.flink.streaming.sink.mysql.OnlineStatisticsMysqlSink;
import pers.aishuang.flink.streaming.source.mysql.VehicleInfoMysqlSource;
import pers.aishuang.flink.streaming.utils.JsonParsePartUtil;

import java.util.HashMap;
import java.util.concurrent.TimeUnit;

/**
 * 实现车辆的实时上报故障诊断业务分析
 * 1、读取车辆的数据,将jsob字符串转换成对象
 * 2、读取出来正确的数据
 * 3、将车辆的数据通过地理位置(经纬度)去redis中拉取(geoHash算法)
 * -- 如果拉取数据成功,直接封装成对象
 * -- 如果拉取省市区地理位置失败,异步数据流读取高德API请求地理位置并将数据保存到redis中
 * 4、将从redis和高德API拉宽的数据进行合并处理
 * 5、使用窗口操作,比如30s统计一些窗口内的故障告警对象返回
 * 6、读取mysql数据库中的车辆静态数据,车辆车型车系,销售时间等
 * 7、窗口数据和静态数据进行connect并flatMap,拉宽数据
 * 8、将数据写入到mysql中
 * 9、执行任务流环境
 *
 */
public class OnlineStatisticsTask extends BaseTask{
    private static final Logger logger = LoggerFactory.getLogger(OnlineStatisticsTask.class);

    public static void main(String[] args) throws Exception{
        //1. 初始化Flink流处理的执行环境(事件时间、checkpoint、hadoop name)
        StreamExecutionEnvironment env = getEnv(OnlineStatisticsTask.class.getSimpleName());
        //2. 接入kafka数据源,消费kafka数据
        DataStreamSource<String> kafkaStream = getKafkaStream(
                    env,
                    "__consumer_online_alarm_analysis_",
                    SimpleStringSchema.class);
        //3. 将消费到的json字符串转换成ItcastDataPartObj对象
        DataStream<ItcastDataPartObj> source = kafkaStream
                .map(JsonParsePartUtil::parseJsonToObject)
                //4. 过滤掉异常数据,根据errorDara属性判断(没有VIN号和终端时间 和json解析失败的数据都视为异常数据)
                .filter(obj -> StringUtils.isEmpty(obj.getErrorData()));
        //5. 读取redis中的位置数据<geohash,VehicleLocationModel> ,生成新的数据流
        SingleOutputStreamOperator<ItcastDataPartObj> itcastDataMapStream = source.map(new LocactionInfoReidsFunction());
        //6. 过滤出 redis拉宽成功的地理位置数据
        SingleOutputStreamOperator<ItcastDataPartObj> okWithLocationStream = itcastDataMapStream
                .filter(obj -> StringUtils.isNotEmpty(obj.getProvince()));
        //7. 过滤出 redis拉框失败的地理位置数据
        SingleOutputStreamOperator<ItcastDataPartObj> ngWithLocationStream = itcastDataMapStream
                .filter(obj -> StringUtils.isEmpty(obj.getProvince()));
        //8. 对redis拉框失败的地理位置数据使用异步IO访问高德地图地理位置查询地理位置信息,并将返回结果写入到reids中
        //-- 异步数据流 :处理之后的数据(成功补齐数据和失败的ItcastDataPartObj)
        //-- 存在问题,http请求失败的数据还在里面,仍然缺少坐标详细信息
        SingleOutputStreamOperator<ItcastDataPartObj> withLocationAsyncStream = AsyncDataStream
                //无序返回(可设置返回是否有序,先访问先返回,后访问后返回,设置有序会造成效率低,所以设置为无序)
                .unorderedWait(
                        ngWithLocationStream,
                        new AsyncHttpQueryFunction(),
                        3000, //设置超时时间,超过设定时间,认为任务请求失败,3000ms=》 3s
                        TimeUnit.MICROSECONDS //超时单位
                );

        //9. 将redis拉宽的地理位置数据与高德API拉宽的地理位置数据进行上下合并(合流)
        //flatmap(FlatMap) / map(Map) 用于单流
        // broadcast + connect + flatmap(CoFlatMap)/map(CoMap) 数据拉宽,主要用于两流的数据左右合并(不要求两流的数据类型一致)
        //              union 数据数据上下合并,要求数据类型一致。
        //FlatMap 和 Map是用于单流的,CoFlatMap和CoMap是用于两条流连接(co:connect)
        WindowedStream<ItcastDataPartObj, String, TimeWindow> windowStream = okWithLocationStream
                .union(withLocationAsyncStream)
                //10. 创建原始数据的30s的滚动窗口,根据vin进行分流操作
                .assignTimestampsAndWatermarks(
                        //水印乱序时间设为3s
                        new BoundedOutOfOrdernessTimestampExtractor<ItcastDataPartObj>(Time.seconds(3)) {
                            @Override
                            public long extractTimestamp(ItcastDataPartObj element) {
                                //指定JavaBean中某个字段数据作为事件时间,必须是long类型
                                return element.getTerminalTimeStamp();
                            }
                        }
                )
                //设置分组,指定JavaBean的vin字段作为分组字段
                .keyBy(obj -> obj.getVin())
                //设置窗口类型:为滚动事件时间窗口,并设置窗口大小
                .window(TumblingEventTimeWindows.of(Time.seconds(30)));
        //11. 对原始数据的窗口流数据进行实时故障分析(区分出来告警数据和非告警数据19个告警字段)
        SingleOutputStreamOperator<OnlineDataObj> onlineStatisticsStream = windowStream
                .apply(new OnlineStatisticsWindowFunction());
        //12. 加载业务中间表(7张表:车辆表、车辆类型表、车辆销售记录表、车辆用途表4张),并进行广播
        DataStream<HashMap<String, VehicleInfoModel>> vehicleInfoBroadcastStream = env
                .addSource(new VehicleInfoMysqlSource()).broadcast();
        //13. 将第11步和第12步的广播流结果进行关联,并应用拉宽操作。
        //上报车辆不在库记载的直接丢了
        SingleOutputStreamOperator<OnlineDataObj> result = onlineStatisticsStream
                .connect(vehicleInfoBroadcastStream)
                .flatMap(new VehicleInfoMapMysqlFunction());
        //14. 将拉框后的结果数据写入到mysql数据库中
        result.addSink(new OnlineStatisticsMysqlSink());

        //15. 启动作业(触发执行)
        env.execute(OnlineStatisticsTask.class.getSimpleName());
    }
}

上一篇:mysql进阶(二十六)MySQL 索引类型(初学者必看)


下一篇:【Flink】Flink checkpoint 本地磁盘 没有清除