python flat_map字典写入es

[root@master pyflink]# cat test.txt 
aaaaa 111111
bbbbb 222222
ccccc 333333
ddddd 444444
eeeee 555555


[root@master pyflink]# cat test.py 
# -*- coding: utf-8 -*-
from pyflink.datastream import StreamExecutionEnvironment
from pyflink.datastream.functions import  MapFunction, RuntimeContext, KeyedProcessFunction
from abc import ABC, abstractmethod
from pyflink.datastream import StreamExecutionEnvironment
from pyflink.datastream.functions import  MapFunction, RuntimeContext, KeyedProcessFunction
from pyflink.datastream.state import MapStateDescriptor
from pyflink.datastream.connectors.kafka import FlinkKafkaConsumer
from pyflink.common.typeinfo import Types, TypeInformation
from pyflink.datastream.connectors.elasticsearch import Elasticsearch7SinkBuilder, ElasticsearchEmitter, FlushBackoffType
from pyflink.datastream.connectors import DeliveryGuarantee
from pyflink.common.serialization import SimpleStringSchema
import json
import re
from datetime import datetime
from elasticsearch import Elasticsearch
from pyflink.datastream.functions import RuntimeContext, FlatMapFunction

import re
import redis


# 创建 StreamExecutionEnvironment 对象
env = StreamExecutionEnvironment.get_execution_environment()
env.set_parallelism(1)

# 读取文件,创建 DataStream 对象
data_stream = env.read_text_file('/root/pyflink/test.txt')
class MyMapFunction(FlatMapFunction):
   def open(self, runtime_context: RuntimeContext):
       pass
   def close(self):
       pass

   def flat_map(self,line):
      r_value=str(int(line.split(' ')[1]) + 1)
      dict1={}
      dict1['r_value']=r_value
      yield dict1

env.add_jars("file:///root/lib/flink-sql-connector-elasticsearch7-3.0.1-1.16.jar")
date_str = datetime.now().strftime("%Y-%m-%d")
es7_sink = Elasticsearch7SinkBuilder() \
    .set_bulk_flush_max_actions(1) \
    .set_emitter(ElasticsearchEmitter.static_index('flink-test2023-06-07')) \
    .set_hosts(['127.0.0.1:9200']) \
    .build()
      
new_stream = data_stream.flat_map(MyMapFunction(),output_type=Types.MAP(Types.STRING(),Types.STRING())).sink_to(es7_sink)
# 输出到控制台

# 执行任务
env.execute('Add "bus_seq" to each line')

你可能感兴趣的:(Flink实时计算,flink)