flink 1.13 streamload写入doris

前言

官方教程详细介绍了基于flink 1.16的各种写入方式，本文主要介绍的是基于flink 1.13的RowData 数据流(RowDataSerializer)写入

    <properties>
        <maven.compiler.source>8</maven.compiler.source>
        <maven.compiler.target>8</maven.compiler.target>
        <flink.version>1.13.1</flink.version>
    </properties>


        <dependency>
            <groupId>org.apache.doris</groupId>
            <artifactId>flink-doris-connector-1.13_2.12</artifactId>
            <version>1.0.3</version>
        </dependency>

        <dependency>
            <groupId>org.apache.flink</groupId>
            <artifactId>flink-table-planner-blink_2.12</artifactId>
            <version>${flink.version}</version>
            <scope>provided</scope>
        </dependency>

一、目标表

-- doris sink table 
CREATE TABLE IF NOT EXISTS events (
    `user_id` INT NOT NULL COMMENT '用户id',
    `date` DATE COMMENT '接收日期',
    `event_type` varchar(256)  COMMENT '事件类型',
    `province` varchar(128)  COMMENT '省份',
    `city` varchar(128)  COMMENT '城市'
    `receive_time` DATETIME
)
DUPLICATE KEY(`user_id`, `receive_date`)
COMMENT '事件表'
PARTITION BY RANGE (`receive_date`)
(
FROM
    ("2023-01-01") TO ("2023-09-14") INTERVAL 1 DAY
)
DISTRIBUTED BY HASH (`user_id`) BUCKETS 1
PROPERTIES (
   "replication_allocation" = "tag.location.default: 1",
   "compression"="LZ4",
   "dynamic_partition.enable" = "true",
   "dynamic_partition.time_unit" = "DAY",
   "dynamic_partition.end" = "3",
   "dynamic_partition.prefix" = "p",
   "dynamic_partition.buckets" = "1"
);

二、map

public class EventMapFunction extends RichMapFunction<Event, RowData> {
    private static final Logger log = LoggerFactory.getLogger(EventMapFunction.class);

    @Override
    public RowData map(Event event) {
        GenericRowData genericRowData = new GenericRowData(6);
        try {
        //map 字段映射
            genericRowData.setField(0, event.getUserId());
            // 字符型需要转化，否则会报错
            genericRowData.setField(1, StringData.fromString(event.getReceiveDate()));
            genericRowData.setField(2, StringData.fromString(event.getEventType()));
            genericRowData.setField(3, StringData.fromString(event.getProvince()));
            genericRowData.setField(4, StringData.fromString(event.getCity()));
            genericRowData.setField(5, StringData.fromString(event.getReceiveTime()));

        } catch (Exception e) {
            log.error("Event data map error : " + e);
        }
        return genericRowData;
    }
}

三、DorisSink

public class DorisSinkUtil {
    public static SinkFunction<RowData> getDorisSink(String table, String labelPrefix) {
 		//写入格式   
        Properties properties = new Properties();
        properties.setProperty("read_json_by_line", "true");
        properties.setProperty("format", "json");
        properties.setProperty("strip_outer_array", "true");

        SinkFunction<RowData> dorisSink = DorisSink.sink(getEventFields(),
                getEventDataType(),
                DorisReadOptions.builder().build(),
                DorisExecutionOptions.builder()
                        .setBatchSize(3)
                        .setBatchIntervalMs(0L)
                        .setMaxRetries(3)
                        .setStreamLoadProp(properties)
                        .build(),
                DorisOptions.builder()
                        .setFenodes(readValue("doris.fenodes"))
                        .setTableIdentifier(table)
                        .setUsername(readValue("doris.username"))
                        .setPassword(readValue("doris.password"))
                        .build());
          
        return dorisSink;
    }
	//字段及类型一一对应
    public static String[] getEventFields() {
        return new String[]{
                "user_id",
                "receive_date",
                "event_type",
                "province",
                "city",
                "receive_time"
        };
    }

    public static LogicalType[] getEventDataType() {
        return new LogicalType[]{
                new IntType(),
                new VarCharType(256),
                new VarCharType(256),
                new VarCharType(128),
                new VarCharType(128),
                new VarCharType(256)
        };
    }

}

四、job主类

    public static void main(String[] args) {

        try {
            /** 一、 创建flink流式执行环境 */
            final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
            env.getConfig().setAutoWatermarkInterval(0);

            /** 二、 获取Source */
            KafkaSource<String> eventSource = getKafkaSource(
                    readValue("kafka.broker.id"),
                    readValue("kafka.event.topic"),
                    readValue("kafka.event.group.id"));
            /** 三、消费 Source */
            SingleOutputStreamOperator<String> eventSourceStream = env.fromSource(eventSource, WatermarkStrategy.noWatermarks(), "kafkaSource_event").setParallelism(12);

            /** 四、解析数据 */
            SingleOutputStreamOperator<Event> eventSingleOutputStreamOperator = eventSourceStream.flatMap(new EventParseFlatMap()).setParallelism(12);


			/** 五、将java bean类型转化为rowdata类型 */
            SingleOutputStreamOperator<RowData> eventStream =   eventSingleOutputStreamOperator   .map(new EventMapFunction()).setParallelism(12);
            

            /** 六、构建doris sink */
            SinkFunction<RowData> eventSink = DorisSinkUtil.getDorisSink(readValue("doris.table.event"), Constants.EVENT_LABEL_PREFIX);


            /** 七、输出至doris */
            eventStream.addSink(eventSink).setParallelism(12);


            env.execute("EventJob");
        } catch (Exception e) {
            LOG.error("EventJob failed to activate", e);
        }
    }