PyFlink Sample Program Code Description

Updated on 2024-08-10 GMT+08:00

View PDF

Assume that you need to submit a Flink task to an MRS cluster. The main language used by the service platform is Python. The following content uses an example Python program to read and writ Kafka jobs.

This scenario applies to MRS 3.3.0 or later.

Code Example for Submitting a Flink Read/Write Kafka Job to Yarn Using a Python API

The main logic code in pyflink-kafka.py is provided. Before submitting the code, ensure that file_path is the path where the SQL statement to be executed. You are advised to use a full path.

For details about the complete code, see pyflink-kafka.py in flink-examples/pyflink-example/pyflink-kafka.

import os
import logging
import sys
from pyflink.common import JsonRowDeserializationSchema, JsonRowSerializationSchema
from pyflink.common.typeinfo import Types
from pyflink.datastream.connectors import FlinkKafkaProducer, FlinkKafkaConsumer
from pyflink.datastream import StreamExecutionEnvironment
from pyflink.table import TableEnvironment, EnvironmentSettings
def read_sql(file_path):
    if not os.path.isfile(file_path):
        raise TypeError(file_path + " does not exist")
    all_the_text = open(file_path).read()
    return all_the_text
def exec_sql():
    # Change the SQL path before job submission.
    # file_path = "/opt/client/Flink/flink/insertData2kafka.sql"
    # file_path = os.getcwd() + "/../../../../yarnship/insertData2kafka.sql"
    # file_path = "/opt/client/Flink/flink/conf/ssl/insertData2kafka.sql"
    file_path = "insertData2kafka.sql"
    sql = read_sql(file_path)
    t_env = TableEnvironment.create(EnvironmentSettings.in_streaming_mode())
    statement_set = t_env.create_statement_set()
    sqlArr = sql.split(";")
    for sqlStr in sqlArr:
        sqlStr = sqlStr.strip()
        if sqlStr.lower().startswith("create"):
            print("---------create---------------")
            print(sqlStr)
            t_env.execute_sql(sqlStr)
        if sqlStr.lower().startswith("insert"):
            print("---------insert---------------")
            print(sqlStr)
            statement_set.add_insert_sql(sqlStr)
    statement_set.execute()
def read_write_kafka():
    # find kafka connector jars
    env = StreamExecutionEnvironment.get_execution_environment()
    env.set_parallelism(1)
    specific_jars = "file:///opt/client/Flink/flink/lib/flink-connector-kafka-xxx.jar"
    # specific_jars = "file://" + os.getcwd() + "/../../../../yarnship/flink-connector-kafka-xxx.jar"
    # specific_jars = "file:///opt/client/Flink/flink/conf/ssl/flink-connector-kafka-xxx.jar"
    # the sql connector for kafka is used here as it's a fat jar and could avoid dependency issues
    env.add_jars(specific_jars)
    kafka_properties = {'bootstrap.servers': '192.168.20.162:21005', 'group.id': 'test_group'}
    deserialization_schema = JsonRowDeserializationSchema.builder() \
        .type_info(type_info=Types.ROW([Types.INT(), Types.STRING()])).build()
    kafka_consumer = FlinkKafkaConsumer(
        topics='test_source_topic',
        deserialization_schema=deserialization_schema,
        properties=kafka_properties)
    print("---------read ---------------")
    ds = env.add_source(kafka_consumer)
    serialization_schema = JsonRowSerializationSchema.builder().with_type_info(
        type_info=Types.ROW([Types.INT(), Types.STRING()])).build()
    kafka_producer = FlinkKafkaProducer(
        topic='test_sink_topic',
        serialization_schema=serialization_schema,
        producer_config=kafka_properties)
    print("--------write------------------")
    ds.add_sink(kafka_producer)
    env.execute("pyflink kafka test")
if __name__ == '__main__':
    logging.basicConfig(stream=sys.stdout, level=logging.INFO, format="%(message)s")
    print("------------------insert data to kafka----------------")
    exec_sql()
    print("------------------read_write_kafka----------------")
    read_write_kafka()

**Table 1** Parameters for submitting a regular job using Python
Parameter	Description	Example
bootstrap.servers	Service IP address and port number of the Broker instance of Kafka	192.168.12.25:21005
specific_jars	Package path: Client installation directory*/Flink/flink/lib/flink-connector-kafka-.jar**. You are advised to use a full path. NOTE: If a job needs to be submitted as yarn-application, replace the following path. Replace the JAR package version with the actual one. specific_jars="file://"+os.getcwd()+"/../../../../yarnship/flink-connector-kafka-1.15.0-h0.cbu.mrs.330.r13.jar"	specific_jars = file:///Client installation directory/Flink/flink/lib/flink-connector-kafka-1.15.0-h0.cbu.mrs.330.r13.jar
file_path	Path of the insertData2kafka.sql file. You are advised to use a full path. Obtain the package from the secondary development sample code and upload it to the specified directory on the client. NOTE: If a job needs to be submitted as yarn-application, replace the following path: file_path = os.getcwd() + "/../../../../yarnship/insertData2kafka.sql"	file_path = /Client installation directory/Flink/flink/insertData2kafka.sql

The following is an example SQL statement:

create table kafka_sink_table (
  age int,
  name varchar(10)
) with (
  'connector' = 'kafka',
'topic' = 'test_source_topic', --Name of the topic written to Kafka. Ensure that the topic name is the same as that in the Python file.
  'properties.bootstrap.servers' = 'IP address of the Kafka broker instance:Kafka port number',
  'properties.group.id' = 'test_group',
  'format' = 'json'
);
create TABLE datagen_source_table (
   age int,
  name varchar(10)
) WITH (
  'connector' = 'datagen',
  'rows-per-second' = '1'
);
INSERT INTO
  kafka_sink_table
SELECT
  *
FROM
  datagen_source_table;

Code Example for Submitting a Flink SQL Job to Yarn Using a Python API

The main logic code in pyflink-sql.py is provided. Before submitting the code, ensure that file_path is the path where the SQL statement to be executed. You are advised to use a full path.
For details about the complete code, see pyflink-sql.py in flink-examples/pyflink-example/pyflink-sql.
import logging
import sys
import os
from pyflink.table import (EnvironmentSettings, TableEnvironment)
def read_sql(file_path):
    if not os.path.isfile(file_path):
        raise TypeError(file_path + " does not exist")
    all_the_text = open(file_path).read()
    return all_the_text
def exec_sql():
    # Change the SQL path before job submission.
    file_path = "datagen2kafka.sql"
    sql = read_sql(file_path)
    t_env = TableEnvironment.create(EnvironmentSettings.in_streaming_mode())
    statement_set = t_env.create_statement_set()
    sqlArr = sql.split(";")
    for sqlStr in sqlArr:
        sqlStr = sqlStr.strip()
        if sqlStr.lower().startswith("create"):
            print("---------create---------------")
            print(sqlStr)
            t_env.execute_sql(sqlStr)
        if sqlStr.lower().startswith("insert"):
            print("---------insert---------------")
            print(sqlStr)
            statement_set.add_insert_sql(sqlStr)
    statement_set.execute()
if __name__ == '__main__':
    logging.basicConfig(stream=sys.stdout, level=logging.INFO, format="%(message)s")
    exec_sql()
Parameters for submitting a SQL job using Python
Parameter
Description
Example
file_path
Path of the datagen2kafka.sql file. You are advised to use a full path. Obtain the package from the secondary development sample code and upload it to the specified directory on the client.
If a job needs to be submitted as yarn-application, replace the following path:
file_path = os.getcwd() + "/../../../../yarnship/datagen2kafka.sql"
file_path = /Client installation directory/Flink/flink/datagen2kafka.sql
The following is an example SQL statement:
create table kafka_sink (
  uuid varchar(20),
  name varchar(10),
  age int,
  ts timestamp(3),
  p varchar(20)
) with (
  'connector' = 'kafka',
  'topic' = 'input2',
  'properties.bootstrap.servers' = 'IP address of the Kafka broker instance:Kafka port number',
  'properties.group.id' = 'testGroup2',
  'scan.startup.mode' = 'latest-offset',
  'format' = 'json'
);
create TABLE datagen_source (
  uuid varchar(20),
  name varchar(10),
  age int,
  ts timestamp(3),
  p varchar(20)
) WITH (
  'connector' = 'datagen',
  'rows-per-second' = '1'
);
INSERT INTO
  kafka_sink
SELECT
  *
FROM
  datagen_source;