在《0基础学习PyFlink——使用PyFlink的SQL进行字数统计》一文中，我们直接执行了Select查询操作，在终端中直接看到了查询结果。

select word, count(1) as `count` from source group by word;
+--------------------------------+----------------------+
|                           word |                count |
+--------------------------------+----------------------+
|                              A |                    3 |
|                              B |                    1 |
|                              C |                    2 |
|                              D |                    2 |
|                              E |                    1 |
+--------------------------------+----------------------+

在生产环境，我们往往要将计算结果保存到外部系统中，比如Mysql等。这个时候我们就要使用Sink。

Sink

Sink用于将Reduce结果输出到外部系统。它也是通过一个表（Table）来表示结构。这个和MapReduce思路中的Map很类似。

Print

为了简单起见，我们让Sink的表连接的外部系统是print。这样我们就可以在控制台上看到数据。

    # define the sinkmy_sink_ddl = """CREATE TABLE WordsCountTableSink (`word` STRING,`count` BIGINT,PRIMARY KEY (word) NOT ENFORCED) WITH ('connector' = 'print');"""t_env.execute_sql(my_sink_ddl).print()

这一步只能创建表和连接器，具体执行还要执行下一步

Execute

因为source和WordsCountTableSink是两张表，分别表示数据的输入和输出结构。如果要打通输入和输出，则需要将source表中的数据通过某些计算，插入到WordsCountTableSink表中。于是我们主要使用的是insert into指令。

    # execute insertmy_select_ddl = """insert into WordsCountTableSinkselect word, count(1) as `count`from sourcegroup by word"""t_env.execute_sql(my_select_ddl).wait()

完整代码如下

import argparse
import logging
import sysfrom pyflink.common import Configuration
from pyflink.table import (EnvironmentSettings, TableEnvironment)def word_count(input_path):config = Configuration()# write all the data to one fileconfig.set_string('parallelism.default', '1')env_settings = EnvironmentSettings \.new_instance() \.in_batch_mode() \.with_configuration(config) \.build()t_env = TableEnvironment.create(env_settings)# define the sourcemy_source_ddl = """create table source (word STRING) with ('connector' = 'filesystem','format' = 'csv','path' = '{}')""".format(input_path)t_env.execute_sql(my_source_ddl).print()tab = t_env.from_path('source')# define the sinkmy_sink_ddl = """CREATE TABLE WordsCountTableSink (`word` STRING,`count` BIGINT,PRIMARY KEY (word) NOT ENFORCED) WITH ('connector' = 'print');"""t_env.execute_sql(my_sink_ddl).print()# execute insertmy_select_ddl = """insert into WordsCountTableSinkselect word, count(1) as `count`from sourcegroup by word"""t_env.execute_sql(my_select_ddl).wait()if __name__ == '__main__':logging.basicConfig(stream=sys.stdout, level=logging.INFO, format="%(message)s")parser = argparse.ArgumentParser()parser.add_argument('--input',dest='input',required=False,help='Input file to process.')argv = sys.argv[1:]known_args, _ = parser.parse_known_args(argv)word_count(known_args.input)

执行命令如下

python sql_print.py --input input1.csv

输出结果如下

Using Any for unsupported type: typing.Sequence[~T]
No module named google.cloud.bigquery_storage_v1. As a result, the ReadFromBigQuery transform CANNOT be used with method=DIRECT_READ.
OK
OK
+I[A, 3]
+I[B, 1]
+I[C, 2]
+I[D, 2]
+I[E, 1]

因为使用的是批处理模式（in_batch_mode），我们看到Flink将所有数据计算完整成，成批的执行了新增操作（+代表新增）。这块对比我们将在后续将流处理时介绍区别。
附上input1.csv内容

"A",
"B",
"C",
"D",
"A",
"E",
"C",
"D",
"A",