Flink-输出算子(Sink)使用

5.5 输出算子

5.5.1 概述

调用print是返回输出类，作为最后一环sink存在

该方法创建了一个PrintSinkFunction 操作，然后作为addSink方法的参数

PrintSinkFunction这个类继承自RichSinkFunction富函数类

RichSinkFunction类

继承了AbstractRichFunction富函数类

因此就可以调用富函数类(是一个实现类)的声明周期方法，例如open，close，以及获取运行时上下文，运行环境，定义状态等等

RichSinkFunction类同时也实现了SinkFunction这个接口，所以本质上也是SinkFunction

SinkFunction接口的抽象方法有invoke，传入是value，以及当前的上下文

关系图

如果需要自定义输出算子

可以调用DataStream的addSink方法

然后传入自己实现的SinkFunction

flink提供的第三方系统连接器

5.5.2 输出到文件

StreamingFileSink流文件输出类

来源

继承RichSinkFunction类，并实现CheckpointedFunction，CheckpointListener(检查点)

创建实例

在StreamingFileSink类中调用forRowFormat()方法传入Path以及Encoder返回StreamingFileSink.DefaultBulkFormatBuilder，DefaultBulkFormatBuilder是一个静态类并继承RowFormatBuilder类，RowFormatBuilder类又继承BucketsBuilder类，底层将数据写入bucket(桶)，桶里面分大小存储分区文件，实现了分布式存储

使用Builder构建器构建

RowFormatBuilder是行编码

BulkFormatBuilder是列存储编码格式

关系图
代码

public class SinkToFileTest {public static void main(String[] args) throws Exception{StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();env.setParallelism(4);DataStreamSource<Event> stream = env.fromElements(new Event("Mary", "./home", 1000L),new Event("Bob", "./cart", 2000L),new Event("Alice", "./prod?id=100", 3000L),new Event("Bob", "./prod?id=1", 3300L),new Event("Alice", "./prod?id=200", 3000L),new Event("Bob", "./home", 3500L),new Event("Bob", "./prod?id=2", 3800L),new Event("Bob", "./prod?id=3", 4200L));//2.为了得到并传入SinkFunction，需要构建StreamingFileSink的一个对象//调用forRowFormat方法或者forBulkformat方法得到一个DefaultRowFormatBuilder//  其中forBulkformat方法前面还有类型参数，以及传参要求一个目录名称，一个编码器//写入文件需要序列化，需要定义序列化方法并进行编码转换，当成Stream写入文件//然后再使用builder创建实例StreamingFileSink<String> streamingFileSink = StreamingFileSink.<String>forRowFormat(new Path("./output"),new SimpleStringEncoder<>("UTF-8")).withRollingPolicy(//指定滚动策略，根据事件或者文件大小新产生文件归档保存DefaultRollingPolicy.builder()//使用builder构建实例.withMaxPartSize(1024 * 1024 * 1024).withRolloverInterval(TimeUnit.MINUTES.toMinutes(15))//事件间隔毫秒数.withInactivityInterval(TimeUnit.MINUTES.toMinutes(15))//当前不活跃的间隔事件，隔多长事件没有数据到来.build()).build();//1.写入文件调用addSink()方法，并传入SinkFunctionstream.map(data -> data.toString())//把Event类型转换成String.addSink(streamingFileSink);env.execute();}
}

结果

5.5.3 输出到kafka

构造FlinkKafkaProducer类传入三个参数：brokerList(主机+端口号)和topicId(topic)以及serializationSchema(编码序列化)完成构造

代码

public class SinkToKafka {public static void main(String[] args) throws Exception{StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();env.setParallelism(1);//1.从kafka中读取数据Properties properties = new Properties();properties.setProperty("bootstrap.servers","hadoop2:9092");properties.setProperty("group.id", "consumer-group");DataStreamSource<String> kafkaStream = env.addSource(new FlinkKafkaConsumer<String>("clicks", new SimpleStringSchema(), properties));//2.用flink进行简单的etl处理转换SingleOutputStreamOperator<String> result = kafkaStream.map(new MapFunction<String, String>() {@Overridepublic String map(String value) throws Exception {String[] fields = value.split(",");return new Event(fields[0].trim(), fields[1].trim(), Long.valueOf(fields[2].trim())).toString();}});//3.结果数据写入kafka//FlinkKafkaProducer传参borckList，topicid,序列化result.addSink(new FlinkKafkaProducer<String>("hadoop2:9092","events",new SimpleStringSchema()));env.execute();}
}

kafka输出结果

5.5.4 输出到redis

引入依赖

<dependency><groupId>org.apache.bahir</groupId><artifactId>flink-connector-redis_2.11</artifactId><version>1.0</version>
</dependency>

分析

RedisSink类分析

RedisSink类继承自RichSinkFunction

参数分析

去调构造方法，传入redis集群的配置FlinkJedisConfigBase以及RedisMapper写入命令

new FlinkJedisConfigBase的时候，可以使用FlinkJedisPoolConfig没毛病，直接继承的FlinkJedisConfigBase

FlinkJedisConfigBase是一个接口

实例FlinkJedisPoolConfig的时候也是使用的构造器Builder()的设计模式即，同样再使用.build实例它

第二个参数分析

RedisMapper是一个接口

自定义一个实现类并重写方法getCommandDescription(),getKeyFromData(Event data),getValueFromData(Event data)

关系图

代码

public class SinkToRedis {public static void main(String[] args) throws Exception{StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();env.setParallelism(1);//1.输入ClickSource是自定义输入DataStreamSource<Event> stream = env.addSource(new ClickSource());//2.创建一个jedis连接配置//FlinkJedisPoolConfig直接继承的FlinkJedisConfigBaseFlinkJedisPoolConfig config = new FlinkJedisPoolConfig.Builder().setHost("hadoop2").build();//3.写入redisstream.addSink(new RedisSink<>(config,new MyRedisMapper()));env.execute();}//3.自定义类实现 redisMapper接口public static class MyRedisMapper implements RedisMapper<Event>{@Override//返回一个redis命令的描述public RedisCommandDescription getCommandDescription() {return new RedisCommandDescription(RedisCommand.HSET,"clicks");//写入哈希表}@Override//把key定义成userpublic String getKeyFromData(Event data) {return data.user;}@Override//把value定义成urlpublic String getValueFromData(Event data) {return data.url;}}
}

结果

运行redis

[hadoop1@hadoop2 redis]$ ./src/redis-server [hadoop1@hadoop2 bin]$ pwd
/usr/local/bin

5.5.5 输出到ElasticSearch

引入依赖

<dependency><groupId>org.apache.flink</groupId>
<artifactId>flink-connector-elasticsearch6_${scala.binary.version}</artifact
Id>
<version>${flink.version}</version>
</dependency>

分析

ElasticsearchSink类分析

ElasticsearchSink类继承ElasticsearchSinkBase抽象类，ElasticsearchSinkBase抽象类继承RichSinkFunction接口

实例

ElasticsearchSink类调用Builder()传入参数是List和ElasticsearchSinkFunction

HttpHost需要参数主机名和端口号

是一个接口，写一个实现类重写他的方法，写入逻辑

关系图

代码

public class SinToES {public static void main(String[] args) throws Exception{StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();env.setParallelism(1);//1.输入DataStreamSource<Event> stream = env.fromElements(new Event("Mary", "./home", 1000L),new Event("Bob", "./cart", 2000L),new Event("Alice", "./prod?id=100", 3000L),new Event("Bob", "./prod?id=1", 3300L),new Event("Alice", "./prod?id=200", 3000L),new Event("Bob", "./home", 3500L),new Event("Bob", "./prod?id=2", 3800L),new Event("Bob", "./prod?id=3", 4200L));//2.定义hosts的列表ArrayList<HttpHost> httpHosts = new ArrayList<>();httpHosts.add(new HttpHost("hadoop",9200));//3.定义ElasticsearchSinkFunction<T>,是个接口，重写process方法//向es发送请求，并插入数据ElasticsearchSinkFunction<Event> elasticsearchSinkFunction = new ElasticsearchSinkFunction<Event>() {@Override//输入，运行上下文，发送任务请求public void process(Event element, RuntimeContext ctx, RequestIndexer indexer) {HashMap<String, String> map = new HashMap<>();map.put(element.user, element.url);//构建一个indexrequestIndexRequest request = Requests.indexRequest().index("clicks").type("types").source(map);indexer.add(request);}};//4.写入es//传入参数是List<HttpHost>和ElasticsearchSinkFunction<T>stream.addSink(new ElasticsearchSink.Builder<>(httpHosts,elasticsearchSinkFunction).build());env.execute();}
}

结果

5.5.6 输入到Mysql

引入依赖

<dependency><groupId>org.apache.flink</groupId><artifactId>flink-connector-jdbc_${scala.binary.version}</artifactId><version>${flink.version}</version>
</dependency>
<dependency><groupId>mysql</groupId><artifactId>mysql-connector-java</artifactId><version>5.1.47</version>
</dependency>

分析

JdbcSink来源

无继承，无实现

定义了sink方法，三个参数，sql，JdbcStatementBuilder构造，JdbcConnectionOptions等sql的连接配置，然后返回SinkFunction

参数分析

JdbcStatementBuilder是个接口，实现了BiConsumerWithException接口

单一抽象方法accept()，lambda使用

构造器私有，因此调用JdbcConnectionOptionsBuilder.build()进行实例化

关系图

代码

public class SinkToMysql {public static void main(String[] args) throws Exception{StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();env.setParallelism(1);//1.输入DataStreamSource<Event> stream = env.fromElements(new Event("Mary", "./home", 1000L),new Event("Bob", "./cart", 2000L),new Event("Alice", "./prod?id=100", 3000L),new Event("Bob", "./prod?id=1", 3300L),new Event("Alice", "./prod?id=200", 3000L),new Event("Bob", "./home", 3500L),new Event("Bob", "./prod?id=2", 3800L),new Event("Bob", "./prod?id=3", 4200L));//三个参数，sql，JdbcStatementBuilder构造，JdbcConnectionOptions等sql的连接配置stream.addSink(JdbcSink.sink("INSERT INTO clicks (user,url) VALUES(?,?)",((statement,event)->{statement.setString(1,event.user);statement.setString(2,event.url);}),new JdbcConnectionOptions.JdbcConnectionOptionsBuilder().withUrl("jdbc:mysql://localhost:3306/test2").withDriverName("com.mysql.jdbc.Driver").withUsername("root").withPassword("123456").build()));env.execute();}
}

mysql前期准备

创建mysql的test2
创建clicks表

mysql> create table clicks(-> user varchar(20) not null,-> url varchar(100) not null);
Query OK, 0 rows affected (0.02 sec)

结果

5.5.7 自定义Sink输出

分析

调用DataStream的addSink()方法，并传入自定义好的SinkFunction（采用富函数类），重写关键方法invoke()，并且重写富函数类的生命周期相关方法open和close

导入依赖

<dependency><groupId>org.apache.hbase</groupId><artifactId>hbase-client</artifactId><version>${hbase.version}</version>
</dependency>

代码

略