kafkaStream
实时计算
一般流式计算会与批量计算相比较。在流式计算模型中,输入是持续的,可以认为在时间上是无界的,也就意味着,永远拿不到全量数据去做计算。同时,计算结果是持续输出的,也即计算结果在时间上也是无界的。流式计算一般对实时性要求较高,同时一般是先定义目标计算,然后数据到来之后将计算逻辑应用于数据。同时为了提高计算效率,往往尽可能采用增量计算代替全量计算。
流式计算就相当于上图的右侧扶梯,是可以源源不断的产生数据,源源不断的接收数据,没有边界。
网站的用户访问日志进行实时的分析,计算访问量,用户画像,留存率等等,实时的进行数据分析,帮助企业进行决策
可以实时的查看网站注册数量,订单数量,购买数量,金额等。
可以随时更新公交车方位,计算多久到达站牌等
头条类文章的分值计算,通过用户的行为实时文章的分值,分值越高就越被推荐。
Storm 是一个分布式实时大数据处理系统,可以帮助我们方便地处理海量数据,具有高可靠、高容错、高扩展的特点。是流式框架,有很高的数据吞吐能力。
可以轻松地将其嵌入任何Java应用程序中,并与用户为其流应用程序所拥有的任何现有打包,部署和操作工具集成。
Kafka Stream是Apache Kafka从0.10版本引入的一个新Feature。它是提供了对存储于Kafka内的数据进行流式处理和分析的功能。
Kafka Stream的特点如下:
(1)数据结构类似于map,如下图,key-value键值对
(2)KStream
KStream数据流(data stream),即是一段顺序的,可以无限长,不断更新的数据集。
数据流中比较常记录的是事件,这些事件可以是一次鼠标点击(click),一次交易,或是传感器记录的位置数据。
KStream负责抽象的,就是数据流。与Kafka自身topic中的数据一样,类似日志,每一次操作都是向其中插入(insert)新数据。
为了说明这一点,让我们想象一下以下两个数据记录正在发送到流中:
(“ alice”,1)->(“” alice“,3)
如果您的流处理应用是要总结每个用户的价值,它将返回4
了alice
。为什么?因为第二条数据记录将不被视为先前记录的更新。(insert)新数据
(1)需求分析,求单词个数(word count)
(2)引入依赖
在之前的kafka-demo工程的pom文件中引入
org.apache.kafka
kafka-streams
connect-json
org.apache.kafka
org.apache.kafka
kafka-clients
(3)创建原生的kafka staream入门案例
package com.heima.kafka.sample;
import java.time.Duration;
import java.util.Arrays;
import java.util.Properties;
/**
* 流式处理
*/
public class KafkaStreamQuickStart {
public static void main(String[] args) {
//kafka的配置信息
Properties prop = new Properties();
prop.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG,"192.168.200.130:9092");
prop.put(StreamsConfig.DEFAULT_KEY_SERDE_CLASS_CONFIG, Serdes.String().getClass());
prop.put(StreamsConfig.DEFAULT_VALUE_SERDE_CLASS_CONFIG, Serdes.String().getClass());
prop.put(StreamsConfig.APPLICATION_ID_CONFIG,"streams-quickstart");
//stream 构建器
StreamsBuilder streamsBuilder = new StreamsBuilder();
//流式计算
streamProcessor(streamsBuilder);
//创建kafkaStream对象
KafkaStreams kafkaStreams = new KafkaStreams(streamsBuilder.build(),prop);
//开启流式计算
kafkaStreams.start();
}
/**
* 流式计算
* 消息的内容:hello kafka hello itcast
* @param streamsBuilder
*/
private static void streamProcessor(StreamsBuilder streamsBuilder) {
//创建kstream对象,同时指定从那个topic中接收消息
KStream stream = streamsBuilder.stream("itcast-topic-input");
/**
* 处理消息的value
*/
stream.flatMapValues(new ValueMapper>() {
@Override
public Iterable apply(String value) {
return Arrays.asList(value.split(" "));
}
})
//按照value进行聚合处理
.groupBy((key,value)->value)
//时间窗口
.windowedBy(TimeWindows.of(Duration.ofSeconds(10)))
//统计单词的个数
.count()
//转换为kStream
.toStream()
.map((key,value)->{
System.out.println("key:"+key+",vlaue:"+value);
return new KeyValue<>(key.key().toString(),value.toString());
})
//发送消息
.to("itcast-topic-out");
}
}
(4)测试准备
结果:
(1)自定配置参数
package com.heima.kafka.config;
import java.util.HashMap;
import java.util.Map;
/**
* 通过重新注册KafkaStreamsConfiguration对象,设置自定配置参数
*/
@Setter
@Getter
@Configuration
@EnableKafkaStreams
@ConfigurationProperties(prefix="kafka")
public class KafkaStreamConfig {
private static final int MAX_MESSAGE_SIZE = 16* 1024 * 1024;
private String hosts;
private String group;
@Bean(name = KafkaStreamsDefaultConfiguration.DEFAULT_STREAMS_CONFIG_BEAN_NAME)
public KafkaStreamsConfiguration defaultKafkaStreamsConfig() {
Map props = new HashMap<>();
props.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, hosts);//连接信息
props.put(StreamsConfig.APPLICATION_ID_CONFIG, this.getGroup()+"_stream_aid");//组
props.put(StreamsConfig.CLIENT_ID_CONFIG, this.getGroup()+"_stream_cid");//应用名称
props.put(StreamsConfig.RETRIES_CONFIG, 10);//重试次数
props.put(StreamsConfig.DEFAULT_KEY_SERDE_CLASS_CONFIG, Serdes.String().getClass());//key序列化器
props.put(StreamsConfig.DEFAULT_VALUE_SERDE_CLASS_CONFIG, Serdes.String().getClass());
return new KafkaStreamsConfiguration(props);
}
}
修改application.yml文件,在最下方添加自定义配置
kafka:
hosts: 192.168.200.130:9092
group: ${spring.application.name}
(2)新增配置类,创建KStream对象,进行聚合
package com.heima.kafka.stream;
import java.time.Duration;
import java.util.Arrays;
@Configuration
@Slf4j
public class KafkaStreamHelloListener {
@Bean
public KStream kStream(StreamsBuilder streamsBuilder){
//创建kstream对象,同时指定从那个topic中接收消息
KStream stream = streamsBuilder.stream("itcast-topic-input");
stream.flatMapValues(new ValueMapper>() {
@Override
public Iterable apply(String value) {
return Arrays.asList(value.split(" "));
}
})
//根据value进行聚合分组
.groupBy((key,value)->value)
//聚合计算时间间隔
.windowedBy(TimeWindows.of(Duration.ofSeconds(10)))
//求单词的个数
.count()
.toStream()
//处理后的结果转换为string字符串
.map((key,value)->{
System.out.println("key:"+key+",value:"+value);
return new KeyValue<>(key.key().toString(),value.toString());
})
//发送消息
.to("itcast-topic-out");
return stream;
}
}
测试:
启动微服务,正常发送消息,可以正常接收到消息
①在heima-leadnews-behavior微服务中集成kafka生产者配置
修改nacos,新增内容
spring:
application:
name: leadnews-behavior
kafka:
bootstrap-servers: 192.168.200.130:9092
producer:
retries: 10
key-serializer: org.apache.kafka.common.serialization.StringSerializer
value-serializer: org.apache.kafka.common.serialization.StringSerializer
②修改ApLikesBehaviorServiceImpl新增发送消息
定义消息发送封装类:UpdateArticleMess
package com.heima.model.mess;
import lombok.Data;
@Data
public class UpdateArticleMess {
/**
* 修改文章的字段类型
*/
private UpdateArticleType type;
/**
* 文章ID
*/
private Long articleId;
/**
* 修改数据的增量,可为正负
*/
private Integer add;
public enum UpdateArticleType{
COLLECTION,COMMENT,LIKES,VIEWS;
}
}
topic常量类:
package com.heima.common.constants;
public class HotArticleConstants {
public static final String HOT_ARTICLE_SCORE_TOPIC="hot.article.score.topic";
}
完整代码如下:
package com.heima.behavior.service.impl;
import org.springframework.transaction.annotation.Transactional;
@Service
@Transactional
@Slf4j
public class ApLikesBehaviorServiceImpl implements ApLikesBehaviorService {
@Autowired
private CacheService cacheService;
@Autowired
private KafkaTemplate kafkaTemplate;
@Override
public ResponseResult like(LikesBehaviorDto dto) {
//1.检查参数
if (dto == null || dto.getArticleId() == null || checkParam(dto)) {
return ResponseResult.errorResult(AppHttpCodeEnum.PARAM_INVALID);
}
//2.是否登录
ApUser user = AppThreadLocalUtil.getUser();
if (user == null) {
return ResponseResult.errorResult(AppHttpCodeEnum.NEED_LOGIN);
}
UpdateArticleMess mess = new UpdateArticleMess();
mess.setArticleId(dto.getArticleId());
mess.setType(UpdateArticleMess.UpdateArticleType.LIKES);
//3.点赞 保存数据
if (dto.getOperation() == 0) {
Object obj = cacheService.hGet(BehaviorConstants.LIKE_BEHAVIOR + dto.getArticleId().toString(), user.getId().toString());
if (obj != null) {
return ResponseResult.errorResult(AppHttpCodeEnum.PARAM_INVALID, "已点赞");
}
// 保存当前key
log.info("保存当前key:{} ,{}, {}", dto.getArticleId(), user.getId(), dto);
cacheService.hPut(BehaviorConstants.LIKE_BEHAVIOR + dto.getArticleId().toString(), user.getId().toString(), JSON.toJSONString(dto));
mess.setAdd(1);
} else {
// 删除当前key
log.info("删除当前key:{}, {}", dto.getArticleId(), user.getId());
cacheService.hDelete(BehaviorConstants.LIKE_BEHAVIOR + dto.getArticleId().toString(), user.getId().toString());
mess.setAdd(-1);
}
//发送消息,数据聚合
kafkaTemplate.send(HotArticleConstants.HOT_ARTICLE_SCORE_TOPIC,JSON.toJSONString(mess));
return ResponseResult.okResult(AppHttpCodeEnum.SUCCESS);
}
/**
* 检查参数
*
* @return
*/
private boolean checkParam(LikesBehaviorDto dto) {
if (dto.getType() > 2 || dto.getType() < 0 || dto.getOperation() > 1 || dto.getOperation() < 0) {
return true;
}
return false;
}
}
③修改阅读行为的类ApReadBehaviorServiceImpl发送消息
完整代码:
package com.heima.behavior.service.impl;
import org.springframework.transaction.annotation.Transactional;
@Service
@Transactional
@Slf4j
public class ApReadBehaviorServiceImpl implements ApReadBehaviorService {
@Autowired
private CacheService cacheService;
@Autowired
private KafkaTemplate kafkaTemplate;
@Override
public ResponseResult readBehavior(ReadBehaviorDto dto) {
//1.检查参数
if (dto == null || dto.getArticleId() == null) {
return ResponseResult.errorResult(AppHttpCodeEnum.PARAM_INVALID);
}
//2.是否登录
ApUser user = AppThreadLocalUtil.getUser();
if (user == null) {
return ResponseResult.errorResult(AppHttpCodeEnum.NEED_LOGIN);
}
//更新阅读次数
String readBehaviorJson = (String) cacheService.hGet(BehaviorConstants.READ_BEHAVIOR + dto.getArticleId().toString(), user.getId().toString());
if (StringUtils.isNotBlank(readBehaviorJson)) {
ReadBehaviorDto readBehaviorDto = JSON.parseObject(readBehaviorJson, ReadBehaviorDto.class);
dto.setCount((short) (readBehaviorDto.getCount() + dto.getCount()));
}
// 保存当前key
log.info("保存当前key:{} {} {}", dto.getArticleId(), user.getId(), dto);
cacheService.hPut(BehaviorConstants.READ_BEHAVIOR + dto.getArticleId().toString(), user.getId().toString(), JSON.toJSONString(dto));
//发送消息,数据聚合
UpdateArticleMess mess = new UpdateArticleMess();
mess.setArticleId(dto.getArticleId());
mess.setType(UpdateArticleMess.UpdateArticleType.VIEWS);
mess.setAdd(1);
kafkaTemplate.send(HotArticleConstants.HOT_ARTICLE_SCORE_TOPIC,JSON.toJSONString(mess));
return ResponseResult.okResult(AppHttpCodeEnum.SUCCESS);
}
}
①在leadnews-article微服务中集成kafkaStream (参考kafka-demo)
②定义实体类,用于聚合之后的分值封装
package com.heima.model.article.mess;
import lombok.Data;
@Data
public class ArticleVisitStreamMess {
/**
* 文章id
*/
private Long articleId;
/**
* 阅读
*/
private int view;
/**
* 收藏
*/
private int collect;
/**
* 评论
*/
private int comment;
/**
* 点赞
*/
private int like;
}
修改常量类:增加常量
package com.heima.common.constans;
public class HotArticleConstants {
public static final String HOT_ARTICLE_SCORE_TOPIC="hot.article.score.topic";
public static final String HOT_ARTICLE_INCR_HANDLE_TOPIC="hot.article.incr.handle.topic";
}
③ 定义stream,接收消息并聚合
package com.heima.article.stream;
import java.time.Duration;
@Configuration
@Slf4j
public class HotArticleStreamHandler {
@Bean
public KStream kStream(StreamsBuilder streamsBuilder){
//接收消息
KStream stream = streamsBuilder.stream(HotArticleConstants.HOT_ARTICLE_SCORE_TOPIC);
//聚合流式处理
stream.map((key,value)->{
UpdateArticleMess mess = JSON.parseObject(value, UpdateArticleMess.class);
//重置消息的key:1234343434 和 value: likes:1
return new KeyValue<>(mess.getArticleId().toString(),mess.getType().name()+":"+mess.getAdd());
})
//按照文章id进行聚合
.groupBy((key,value)->key)
//时间窗口
.windowedBy(TimeWindows.of(Duration.ofSeconds(10)))
/**
* 自行的完成聚合的计算
*/
.aggregate(new Initializer() {
/**
* 初始方法,返回值是消息的value
* @return
*/
@Override
public String apply() {
return "COLLECTION:0,COMMENT:0,LIKES:0,VIEWS:0";
}
/**
* 真正的聚合操作,返回值是消息的value
*/
}, new Aggregator() {
@Override
public String apply(String key, String value, String aggValue) {
if(StringUtils.isBlank(value)){
return aggValue;
}
String[] aggAry = aggValue.split(",");
int col = 0,com=0,lik=0,vie=0;
for (String agg : aggAry) {
String[] split = agg.split(":");
/**
* 获得初始值,也是时间窗口内计算之后的值
*/
switch (UpdateArticleMess.UpdateArticleType.valueOf(split[0])){
case COLLECTION:
col = Integer.parseInt(split[1]);
break;
case COMMENT:
com = Integer.parseInt(split[1]);
break;
case LIKES:
lik = Integer.parseInt(split[1]);
break;
case VIEWS:
vie = Integer.parseInt(split[1]);
break;
}
}
/**
* 累加操作
*/
String[] valAry = value.split(":");
switch (UpdateArticleMess.UpdateArticleType.valueOf(valAry[0])){
case COLLECTION:
col += Integer.parseInt(valAry[1]);
break;
case COMMENT:
com += Integer.parseInt(valAry[1]);
break;
case LIKES:
lik += Integer.parseInt(valAry[1]);
break;
case VIEWS:
vie += Integer.parseInt(valAry[1]);
break;
}
String formatStr = String.format("COLLECTION:%d,COMMENT:%d,LIKES:%d,VIEWS:%d", col, com, lik, vie);
System.out.println("文章的id:"+key);
System.out.println("当前时间窗口内的消息处理结果:"+formatStr);
return formatStr;
}
}, Materialized.as("hot-atricle-stream-count-001"))
.toStream()
.map((key,value)->{
return new KeyValue<>(key.key().toString(),formatObj(key.key().toString(),value));
})
//发送消息
.to(HotArticleConstants.HOT_ARTICLE_INCR_HANDLE_TOPIC);
return stream;
}
/**
* 格式化消息的value数据
* @param articleId
* @param value
* @return
*/
public String formatObj(String articleId,String value){
ArticleVisitStreamMess mess = new ArticleVisitStreamMess();
mess.setArticleId(Long.valueOf(articleId));
//COLLECTION:0,COMMENT:0,LIKES:0,VIEWS:0
String[] valAry = value.split(",");
for (String val : valAry) {
String[] split = val.split(":");
switch (UpdateArticleMess.UpdateArticleType.valueOf(split[0])){
case COLLECTION:
mess.setCollect(Integer.parseInt(split[1]));
break;
case COMMENT:
mess.setComment(Integer.parseInt(split[1]));
break;
case LIKES:
mess.setLike(Integer.parseInt(split[1]));
break;
case VIEWS:
mess.setView(Integer.parseInt(split[1]));
break;
}
}
log.info("聚合消息处理之后的结果为:{}",JSON.toJSONString(mess));
return JSON.toJSONString(mess);
}
}
①在ApArticleService添加方法,用于更新数据库中的文章分值
/**
* 更新文章的分值 同时更新缓存中的热点文章数据
* @param mess
*/
public void updateScore(ArticleVisitStreamMess mess);
实现类方法
/**
* 更新文章的分值 同时更新缓存中的热点文章数据
* @param mess
*/
@Override
public void updateScore(ArticleVisitStreamMess mess) {
//1.更新文章的阅读、点赞、收藏、评论的数量
ApArticle apArticle = updateArticle(mess);
//2.计算文章的分值
Integer score = computeScore(apArticle);
score = score * 3;
//3.替换当前文章对应频道的热点数据
replaceDataToRedis(apArticle, score, ArticleConstants.HOT_ARTICLE_FIRST_PAGE + apArticle.getChannelId());
//4.替换推荐对应的热点数据
replaceDataToRedis(apArticle, score, ArticleConstants.HOT_ARTICLE_FIRST_PAGE + ArticleConstants.DEFAULT_TAG);
}
/**
* 替换数据并且存入到redis
* @param apArticle
* @param score
* @param s
*/
private void replaceDataToRedis(ApArticle apArticle, Integer score, String s) {
String articleListStr = cacheService.get(s);
if (StringUtils.isNotBlank(articleListStr)) {
List hotArticleVoList = JSON.parseArray(articleListStr, HotArticleVo.class);
boolean flag = true;
//如果缓存中存在该文章,只更新分值
for (HotArticleVo hotArticleVo : hotArticleVoList) {
if (hotArticleVo.getId().equals(apArticle.getId())) {
hotArticleVo.setScore(score);
flag = false;
break;
}
}
//如果缓存中不存在,查询缓存中分值最小的一条数据,进行分值的比较,如果当前文章的分值大于缓存中的数据,就替换
if (flag) {
if (hotArticleVoList.size() >= 30) {
hotArticleVoList = hotArticleVoList.stream().sorted(Comparator.comparing(HotArticleVo::getScore).reversed()).collect(Collectors.toList());
HotArticleVo lastHot = hotArticleVoList.get(hotArticleVoList.size() - 1);
if (lastHot.getScore() < score) {
hotArticleVoList.remove(lastHot);
HotArticleVo hot = new HotArticleVo();
BeanUtils.copyProperties(apArticle, hot);
hot.setScore(score);
hotArticleVoList.add(hot);
}
} else {
HotArticleVo hot = new HotArticleVo();
BeanUtils.copyProperties(apArticle, hot);
hot.setScore(score);
hotArticleVoList.add(hot);
}
}
//缓存到redis
hotArticleVoList = hotArticleVoList.stream().sorted(Comparator.comparing(HotArticleVo::getScore).reversed()).collect(Collectors.toList());
cacheService.set(s, JSON.toJSONString(hotArticleVoList));
}
}
/**
* 更新文章行为数量
* @param mess
*/
private ApArticle updateArticle(ArticleVisitStreamMess mess) {
ApArticle apArticle = getById(mess.getArticleId());
apArticle.setCollection(apArticle.getCollection()==null?0:apArticle.getCollection()+mess.getCollect());
apArticle.setComment(apArticle.getComment()==null?0:apArticle.getComment()+mess.getComment());
apArticle.setLikes(apArticle.getLikes()==null?0:apArticle.getLikes()+mess.getLike());
apArticle.setViews(apArticle.getViews()==null?0:apArticle.getViews()+mess.getView());
updateById(apArticle);
return apArticle;
}
/**
* 计算文章的具体分值
* @param apArticle
* @return
*/
private Integer computeScore(ApArticle apArticle) {
Integer score = 0;
if(apArticle.getLikes() != null){
score += apArticle.getLikes() * ArticleConstants.HOT_ARTICLE_LIKE_WEIGHT;
}
if(apArticle.getViews() != null){
score += apArticle.getViews();
}
if(apArticle.getComment() != null){
score += apArticle.getComment() * ArticleConstants.HOT_ARTICLE_COMMENT_WEIGHT;
}
if(apArticle.getCollection() != null){
score += apArticle.getCollection() * ArticleConstants.HOT_ARTICLE_COLLECTION_WEIGHT;
}
return score;
}